Ammonia

Catalysis and Manufacture

With contributions by
K. Aika, L. 1. Christiansen, I. Dybkjaer,
1. B. Hansen, P. E. H0jlund Nielsen,
A. Nielsen, P. Stoltze, K. Tamaru

With 68 Figures and 23 Tables

Springer-Verlag
Berlin Heidelberg New York
London Paris Tokyo
Hong Kong Barcelona Budapest
Editor:

Anders Nielsen
Haldor Tops0e A/S
NYill011evej 55,2800 Lyngby/DK

Library of Congress Cataloging-in-Publication Data Ammonia: catalysis and manufacture/with

contributions by K. Aika ... [et al.; editor, Anders Nielsen]. p.cm. Includes bibliographical reference
and index.
ISBN-13: 978-3-642-79199-4 e-ISBN-13: 978-3-642-79197-0
DO I: 10.1007/ 978-3-642-79197-0
1. Ammonia. I. Aika, K. (Ken-ichi) II.
Nielsen, Anders, 1934-. TP223.A453 1995. 661'.34--dc20 94-36677 CIP

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation,
broadcasting, reproduction on microfilms or in other ways, and storage in data banks. Duplication
of this publication or parts thereof is only permitted under the provisions of the German Copyright
Law of September 9, 1965, in its current version, and a copyright fee must always be paid.

© Springer-Verlag Berlin Heidelberg 1995

Softcover reprint of the hardcover 1st edition 1995

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence
of a specific statement, that such names are exempt from the relevant protective laws and regulations
and therefore free for general use.

Typesetting: Macmillan India Ltd., Bangalore-25

SPIN: 10077499 51/3020 - 5 4 3 2 1 0 - Printed on acid-free paper
Preface

This book owes its existence to Dr. Ekkehard Fluck, Director of the Gmelin-
Institut. Dr. Fluck suggested that the individual chapters could be written by
staff members of Haldor Topsoe A/S, and that emphasis should be given to the
industrial manufacture of ammonia.
Upon careful consideration it was decided to ask two distinguished experts
in catalysis, Prof. Kenzi Tamaru of the Science University of Tokyo and Prof.
Ken-ichi Aika of the Tokyo Institute of Technology to write the chapter on
ammonia synthesis on non-iron catalysts.
When I started to work in catalysis more than 50 years ago, first as a student
in the Institute of Physical Chemistry with the late Prof. J. N. Bf0nsted and
subsequently in the laboratories of Dr. Haldor Topsoe, the use of catalytic
processes by industry and the literature on catalytic studies was still somewhat
limited and allowed a single person to reasonably acquaint himself with the field.
Today, catalysis is a step in the manufacture of most chemical products and
most refinery streams undergo catalytic reactions. The number of physical tools
applied to the study of catalysts is impressive and the literature on catalysis
overpowering. A reasonably comprehensive volume on the topic of ammonia
synthesis had to be a team effort. The first chapter entitled, "Thermodynamic
properties in Ammonia Synthesis" is written by Dr. Lars J. Christiansen. It should
be emphasized that this chapter does not contain the complete thermodynamics
on ammonia, but concentrates on the thermodynamic properties used for the
design and operation of ammonia synthesis units.
The second chapter "Structure and Surface Chemistry of Industrial Am-
monia Synthesis Catalysts" is written by Dr. Per Stoltze. This chapter deals with
the structure and surface chemistry of iron-based ammonia synthesis catalysts of
the type used by industry. Certain studies of single crystal surfaces are included
to the extent that they serve to add information to the main topic. This chapter
includes a presentation of the unreduced catalyst; the reduction process and the
bulk and surface structure of a reduced catalyst. A thorough discussion is given
of the different states of sorption of nitrogen and of chemisorption of hydrogen,
carbon oxides, ammonia and oxygen. The last part of the chapter gives a
detailed account of the mechanism of ammonia synthesis on iron.
The third chapter is written by Profs. Ken-ichi Aika and Kenzi Tamaru,
both of whom have contributed prominently to our knowledge of ammonia
synthesis. Their chapter is entitled, "Ammonia Synthesis over Non-Iron Catalysts
and Related Phenomena". It is recalled that osmium was indeed used in the early
vi Preface

work by Haber to demonstrate the feasibility of a high pressure ammonia

synthesis. Today, ruthenium is considered the most likely second candidate for
large-scale industrial production of ammonia. The first part of this chapter is a
discussion of ammonia synthesis activity of elements and promoter effects, and
the second part a discussion of the mechanism of ammonia synthesis over
metals, including a study of the adsorption of nitrogen on a number of the
surfaces that show activity for ammonia synthesis and decomposition and of the
state of the adsorbed species of nitrogen.
The fourth chapter is written by Dr. John B0gild Hansen and is entitled,
"Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts".
The scope of the chapter is limited to promoted and non-promoted iron
catalysts. The chapter includes discussions of the Temkin-Pyzhev rate equation,
it deals with rate equations derived from the Langmuir isotherm and reports on
the kinetics based on surface science techniques. In the last section ofthe chapter
a discussion of transfer phenomena - as these are found in experimental reactors
and in industrial converters - is included.
Chapter five written by Dr. Poul Erik H0jlund Nielsen, is entitled, "Poison-
ing of Ammonia Synthesis Catalysts" and this chapter is also limited to iron-
based ammonia synthesis. For many years poisoning of the industrial ammonia
catalysts, particularly with oxygen compounds played a great industrial role. In
a modern ammonia synthesis the purity of the gas converted over the iron-
ammonia synthesis catalysts is extremely high and one can say, that only when
equilibrium is approached at the lower temperatures at which iron catalysts are
active does oxygen poisoning still play a role. The blocking of the catalyst
surface by chemisorption of reactants today appears much more important -
nitrogen and, to a smaller extent, its hydrides on iron and hydrogen on
ruthenimum.
Chapter six is written by Dr. Ib Dybkja:r and is entitled, "Ammonia
Production Processes". The industrial ammonia synthesis is covered in detail
with a shorter coverage of synthesis gas production and storage of ammonia. In
a book on ammonia synthesis this chapter is, of course, a crucial one. A detailed
account is given of the complete ammonia production processes and the related
energy balances. The process schemes for steam reforming of light hydrocar-
bons, as well as processes based on partial oxydation of heavy hydrocarbons
and gasification of solid feedstocks are covered. A discussion of the integration
of the production of ammonia and other products is also included.
Chapter 7 entitled, "Ammonia Storage and Transportation-Safety" has been
written by the undersigned. It was written with the objective of bringing together
our current knowledge from the many sources dealing with this important topic.
In conclusion, I should like to thank the Gmelin-Institut, in particular Dr.
Prof. Ekkehard Fluck, the Director of the Institute, and Dr. J0rn von Jouanne,
who has been the coordinator of this volume, for making the publication of this
book possible.

Anders Nielsen
Table of Contents

Chapter 1
Thermodynamic Properties in Ammonia Synthesis
L. 1. Christiansen . . . . . . . . . . . . . . . . . . . .

Chapter 2
Structure and Surface Chemistry of Industrial Ammonia
Synthesis Catalysts
P. Stoltze. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Chapter 3
Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena
K.-i. Aika and K. Tamaru. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Chapter 4
Kinetics of Ammonia Synthesis and Decomposition
on Heterogeneous Catalysts
1. B. Hansen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

Chapter 5
Poisoning of Ammonia Synthesis Catalysts
P. E. H0jlund Nielsen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

Chapter 6
Ammonia Production Processes
I. Dybkjaer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

Chapter 7
Ammonia Storage and Transportation-Safety
A. Nielsen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
Chapter 1
Thermodynamic Properties in Ammonia Synthesis

Lars J. Christiansen
Haldor Tops0e A/S Copenhagen, Denmark

Contents

1.1 Introduction . . . . . . . . . . . . . . 2

1.2 Equations of State (PVT Properties) 2

1.3 Thermodynamic Properties . . . . . . . . . . . 3

1.3.1 Enthalpies of Pure Compounds and Mixtures 3
1.3.2 Reaction Enthalpies 4

1.4 Chemical Equilibrium 5

1.5 Phase Equilibrium .. 6

1.5.1 Vapor Concentration of Ammonia. 8
1.5.2 Phase Equilibrium of Dissolved Gases 10

1.6 Transport Properties . . . . . . . 12

1.6.1 Low Pressure Properties . 12
1.6.2 Mixture Properties . . . 13
1.6.3 Dependence on Pressure 13

1. 7 References . . . . . . . . . . . . 13

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
2 L. J. Christiansen

1.1 Introduction

The reaction of hydrogen and nitrogen, which is performed under pressure in

the presence of a catalyst, is exothermic and reversible. The conversion to NH3
is thus limited by chemical equilibrium. The ammonia is normally removed from
the gas stream by cooling condensation and the unused reactants are recycled
back to the inlet of the chemical reactor.
Design of energy efficient ammonia plants therefore requires an accurate
knowledge of the following thermodynamic properties at the actual operating
conditions:
1. PVT properties,
2. Thermodynamic Properties,
3. Phase equilibrium between condensed ammonia and gases,
4. Transport properties.
The thermodynamic properties or their derivatives must be continuous in order
to ensure that there is no discontinuity in the derived thermodynamic property
such as cooling condensation, etc.
An extensive review of the required data mentioned above has been given in
Nielsen [45].

1.2 Equations of State (PVT Properties)

The pressure range utilized in ammonia plants implies that the ideal gas law
cannot be used. This is corrected for by use of a compressibility factor Z.,
whereby
pV = ZRT (1)
where p is the pressure, V the molar volume, T the temperature, and R the gas
constant. The compressibility factor can be found from generalized charts, but it
is more common to use an equation of state, where the pressure is given as
a function of temperature, volume, and molecular parameters, such as the
critical properties,
p = f(V,T, parameters) (2)
By use of adequate mixing rules for the molecular parameters, the same type
of equation can be used to calculate mixture properties. Numerous equations of
state have been proposed and a review can be found in Reid et al. [50].
For detailed information on compressibility factors reference is given to the
experimental data by Wiebe, Gaddy [56], by Bartlett et al. [3,4]. These papers
give values of the compressibility factors of hydrogen, nitrogen, and 3: 1 mix-
 1 Thermodynamic Properties in Ammonia Synthesis 3

tures of these gases at pressures up to 1000 atm and temperatures between - 70

and + 300°C. The article by Wiebe and Gaddy also gives compressibility
factors for three other hydrogen-nitrogen mixtures. Reference is also given to an
extensive treatment by Sage et al. [52] of thermodynamic properties of two
hydrogen-nitrogen mixtures, one of them the 76:24 two-component mixture
and the other a mixture of these gases with CO 2 , CO, and CH 4 • Michels et al.
[42] have calculated thermodynamic functions of two gas mixtures containing
ammonia from compressibility isotherms, and Michels et al. [41] report com-
pressibility and thermodynamic data of three ammonia-containing synthesis gas
mixtures.
Thermodynamic properties of pure ammonia are shown in an extensive
work by Haar, Gallagher [17]. They include densities of gaseous and liquid
ammonia between - 50 and + 350°C and from 0 to 1000 bar. Their work
includes data from Circular of the National Bureau of Standard No. 142 [8],
and they find excellent agreement between this work and newer data.
Properties of liquid ammonia from - 50 to + 65°C and from saturation
pressure to 370 bar are also reported by Zander, Thomas [62].

1.3 Thermodynamic Properties

1.3.1 Enthalpies of Pure Compounds and Mixtures

The calculation of the thermodynamic properties is accomplished in two steps.

The first step is the calculation of the properties calculated for the ideal gas, the
pure component heat capacities at constant pressure, and enthalpies and free
energies of formation of the pure components. Ideal mixing rules are used. The
second step is the calculation of the deviation from the ideal gas, which can be
derived from an equation of state describing the mixture properties.
The mixture enthalpy H and the mixture heat capacity C p are calculated as:

H=IYiH~-H' (3)
i

C p = I YiC~i - C~ (4)
i

where H? and C~i are the pure component ideal gas standard enthalpy and heat
capacity, respectively, an Yi the mole fraction. The enthalpy departure from the
ideal gas, H', can be calculated using rigorous thermodynamics as:

H'= -7[P-T(oP) ]dV-RT(Z-l) (5)

v oT v,N
4 L. J. Christiansen

The corresponding heat capacity departure C~ is found by using the temper-

ature differentiation:

C~= [aH']
aT p
(6)

The pure component ideal gas standard enthalpies or heat capacities are
functions of the temperature only. They are normally represented as tables as
can be found in Rossini [51] and JANAF Tables [24] or fitted to polynomials as
shown in Reid et al. [50] and in Christiansen, Kjrer [7]. The last reference uses
a polynomial of fourth order.
(7)
The heat capacity is then simply represented by:
C~i = a2i + 2a3iT + 3a4iT2 + 4asiT3 (8)
The coefficient ali is determined so that the enthalpy at 25°C is equal to the
enthalpy of formation of the pure component. The enthalpy polynomial can
then, without further modifications, be used to calculate heats of reaction, and
consequently heat balances including chemical reactions.
The polynomial coefficients for hydrogen, nitrogen, and ammonia from
Christiansen, Kjrer [7] are determined from data given in Rossini [51] and
JANAF Tables [24] and are shown in Table 1.1 below:

Table 1.1 Enthalpy polynomial coefficients in the equation H = al + a2 T + a3 T2 + a 4T3 + as T4

[7] where H is in kcal/kmol and T in K.

NH3 - 1.320772 X 104 6.048322 4.125509 X 10- 3 - 3.692310 X 10- 8 - 1.802763 X 10- 10
H2 - 2.112450 X 10 3 7.209790 - 5.559028 X 10- 4 4.846263 X 10- 7 - 8.190294 X 10- 11
N2 - 1.976727 X 103 6.459189 5.182665 x 10- 4 2.032237xl0- 7 -7.654612xlO- 11

Concerning enthalpy and specific heat of ammonia, the extensive work by

Haar, Gallagher [17] shows values from - 50 to + 350°C and from 0 to 1000
bar. Ideal gas properties of ammonia are given by Haar [16].
Especially at high pressure it is necessary to take deviation from ideal gas
enthalpies into consideration. This can be included by use of the methods shown
by Reid et al. [50]. The value of H' increases considerably at higher pressures.

1.3.2 Reaction Enthalpies

For any reaction mixture of hydrogen and nitrogen yielding ammonia, the
standard reaction enthalpy can be evaluated from the data in Table 1.1, if stoi-
chiometric are included. It is largest close to the dew point and decreases with
increasing temperatures.
 1 Thermodynamic Properties in Ammonia Synthesis 5

Direct determination of the heat of reaction of NH3 synthesis has been

carried out by Haber, Tamaru [19J, and by Haber et al. [20]. At 1 atm and O°C
Haber found AH~ = - 11000 kcal/kmol; and at 1 atm 600 °CAH~ =
- 13000 kcal/kmol.
Heats of reaction at higher pressures have been calculated by various
authors by use of equation of state. Gillespie, Beattie [13] have used the Beattie
Bridgeman equation of state, and Kazarnovskii, Karapet'yants [27], and Kazar-
novskii [26] have used experimental PVT data to calculate the heat of reaction.

1.4 Chemical Equilibrium

The thermodynamic equilibrium constant K for the ammonia synthesis reaction

is

K = aNH3
(9)
a?/22 ai./22
where aj is the activity of component i, which is given as
(10)
where ({Jj is the fugacity coefficient, which is a function of temperature, pressure
and composition. It is conveniently derived from an equation of state, such as
eq. (1) p. 2 as

In({Jj = - 1
RT
J -p )
00

v
[(a
ani T,V,nj,Fi
- -RTJ dV -
V
RTlnZ (11)

Expressions from different equations of state can be found in Reid et al. [50].
Insertion of (11) in (10) makes it possible to separate the ideal gas from the
non-ideal gas contribution. The equation is then

(12)

The equilibrium constant K is calculated from the Gibbs free energy of

formation as
AGO = RTlnK (13)
where AGO is a function of temperature only given by the Gibbs-Helmholz
equation as

(14)
6 L. J. Christiansen

Insertion of the heat of reaction derived from equation (7) gives the following
expression for the thermodynamic equilibrium constant as a function of temper-
ature
at a4 z as 3
RInK = ao - T + azlnT + a3 T + 2 T + 3 T (15)

The constant ao is fixed so that equation (14) gives the Gibbs free energy of
formation at 25°C. - 3915 kcaljkmol [24].
The equilibrium composition is then calculated in a two step procedure. The
first step leads to the ideal gas composition as given by Kp and the second step is
the correction for the non-ideal gas behavior by use of Kq>' which includes the
influence of pressure and composition.
The most accurate experimental values appear to be those by Haber,
Rossignol (18), by Schultz, Schaefer [53J, by Haber et al. [21J for 30 atm
pressure, by Larson, Dodge [32J for 10, 30, 50, and 100 atm, and by Larson [30J
for pressures of 300, 600, and 1000 atm. Experimental and calculated data on
ammonia synthesis equilibrium at pressures from 1000 atm to 3500 atm have
been given by Winchester, Dodge [61].
The experimental data at high pressures have been analyzed by different
equations of state to calculate the fugacity coefficients in Kq>. Gillespie [IIJ and
Gillespie, Beattie [12J have used the Beattie-Bridgeman equation of state. The
method by Newton [44J is a generalized method for calculation of fugacity
coefficients.
In Figs. 1.1 to 1.3 the equilibrium % NH3 are shown as a function of
temperature at different pressures and at 0%, 10% and 20% inert (methane and
aTgon) content, respectively. The fugacity coefficient ratio is calculated by the
present author using the Martin-Hou equation of state, see [36]. Of course,
other equations of state could have been used for calculating this coefficient.
It is interesting to note that because of the composition dependence of the
fugacity coefficients the maximum yield of ammonia at high pressures may well
exist at a hydrogen-nitrogen ratio different from 3. At 200 atm and 500°C the
maximum yield at equilibrium is to be expected for a ratio of 2.9 although the
increase in only 0.01 % NH 3.

1.5 Phase Equilibrium

The content of ammonia in the vapor phase and the amount of dissolved
synthesis gases in the liquid ammonia phase are calculated from expressions for
the f3 value, f3i = Yi/Xb where Yi and Xi are the mole fractions in the gas and liquid
phases, respectively.
 1 Thermodynamic Properties in Ammonia Synthesis 7

40~-----+4.--4.~---+--r------+------~

M
I
Z

600 700 800 900 1000

Temperature in K

Fig. 1.1 Percentage of ammonia in equilibrium mixture (75% H 2 , 25% N 2 , 0% inerts).

Correlations for the {3 values are derived from the isofugacity criterion,
which can be written as

{3i = 'YJi O (16)

<PiP
where 'Yi is the liquid phase activity coefficient, which is a function of temper-
ature and composition, f? is the reference fugacity, and <Pi the gas phase fugacity
coefficient, which is calculated from an equation of state. All parameters are
given at the pressure of the mixture.
The reference fugacity for ammonia, which is the soluble component, is
evaluated from the saturation fugacity of ammonia and a Poynting correction
factor. For the dissolved gases, the reference fugacity is the Henry's constant
referred to the saturation pressure of ammonia. For further details of the theory
and the necessity of including the Poynting correction factor, see Reid et al. [50].
The activity coefficients and the fugacity coefficients consequently account
for the deviation from Raoult's law for ammonia, and for the deviations from
Henry's law for the dissolved gases. The validity of a published gas solubility for
a binary system in a multicomponent system requires therefore that Henry's law
8 L. J. Christiansen

40~----~--+-~~~----+-~---1------~

201-------t--

10~----_+------+-~~_+~~~+_--~_i

600 700 800 900 1000

Temperature in K

Fig. 1.2 Percentage of ammonia in equilibrium mixture (67.5% H 2 , 22.5% N 2 , 3% Ar, 7% CH4 ).

is valid. This is often a good approximation in ammonia synthesis, where the

amounts of dissolved gases are small.
The type of correlation as shown in equation (18) has been used by Alesan-
drini et aI. [1] to evaluate phase equilibrium properties. They used the Redlich-
Kwong equation of state and the van Laar type activity coefficients to correlate
the experimental information for the binary systems.

1.5.1 Vapor Concentration of Ammonia

The activity coefficient for ammonia is greater than one and the fugacity
coefficient decreases with increasing pressure thereby giving a higher concentra-
tion of ammonia in the gas phase than predicted by the Raoult's law alone.
Experimental measurements of the ammonia concentration in the vapor
phase have been reported by Larson, Black [31] for the equilibrium with
nitrogen and hydrogen from - 22.5 to + 18.7°C and from 50 to 1000 atm.
More recent data are reported by Michels et al. [40]. Their investigation
 1 Thermodynamic Properties in Ammonia Synthesis 9

co>
:I:
Z

600 700 800 900 1000

Temperature in K

Fig. 1.3 Percentage of ammonia in equilibrium mixture (60% H 2 , 20% N 2 , 6% Ar, 14% CH 4 )·

covered the temperature interval from 0 to 121.8°C and pressures from 25 atm
to 785 atm. Michels et al. [38] later reported data in the same pressure range but
at temperatures below O°C. Lefrancois, Vaniscotte [33] report data for the
temperature interval - 70 to + 60°C at pressures 300 and 500 kg/cm 2 • These
data agree well with the data by Michels et al. [38,39]. Reddy, Husain [49] have
given data for the ammonia vapor concentration in a mixture with gas phase
mole ratios, H 2 :N 2 :Ar:CH4 , equal to 3:1:0.18:0.44.
The available data for the vapor concentration of ammonia in equilibrium
with a 3: 1 mixture of hydrogen and nitrogen are shown in Fig. 1.4 for pressures
equal to 10,20,30, and 50 MPa. The data used are those published by Michels et
al. These data are in good agreement with the data given by Lefrancois,
Vaniscotte [33], by Heise [22] and by Zeininger [14] at 25°C, whereas the data
given by Larson, Black [31] give higher concentrations of ammonia, in particu-
lar at the high pressures.
In Fig. 1.4 the predicted values ofthe ammonia concentration are also shown
assuming that Raoult's law is valid and neglecting the dissolved gases at 10 and
50 MPa, respectively. It is seen that Raoult's law is not valid.
The data presented by Reddy, Husain [49] also agree with Fig. 1.4 in spite of
the content of inerts in their data. This is also the case for the data given by
10 L. J. Christiansen

20r-----~-------+------_+----+_~

I 15r---t----t----++-

.!:
M
::c
z
~ 10r-----~-------+--~--~~~~

5r-----~--~~~~~~50M~

I...,o'\l.'?J.. ---
. ., \n'" '?:,..--
?-~\l~_'- Fig. 1.4 Equilibrium vapor concentra-
--- tion of ammonia in 3: 1 mixture of
H2 + N2•
280 300 320
Temperature in K

Zeininger [14]. There is hence some evidence that the pressure of inerts will not
change the ammonia vapor concentration significantly although it will increase.
More experimental information is needed in order to obtain more certainty
about this point.
In the model given by Alesandrini et al. [1] ammonia concentrations are
calculated which are too small at the higher pressures. The model has therefore
been revised by Reddy, Husain [49].

1.5.2 Phase Equilibrium of Dissolved Gases

Solubilities of the gases hydrogen, nitrogen, argon, methane, and helium in the
form of f3 values (definition see p. 6) are shown in the extensive treatment in
Landolt-Bornstein [29]. The data cover the operating region normally used in
ammonia synthesis.
Data for the solubility of hydrogen in ammonia can be found in Reamer,
Sage [47], in Wiebe, Treamearne [57, 58]. In Heise [22], and in Zeininger [63].
Data for the solubility of nitrogen in ammonia can be found in Reamer, Sage
[48], in Heise [22], and in Zeininger [63]. Data for the solubility of argon in
 1 Thermodynamic Properties in Ammonia Synthesis 11

ammonia can be found in Kaminishi [25], in Michels et al. [39], and in Heise
[22]. Data for the solubility of methane in ammonia can be found in Kaminishi
[25], and in Zeininger [63]. Data for the solubility of helium in ammonia can be
found in Heise [22].
Data for the solubility of hydrogen-nitrogen in the ratio 3: 1 in ammonia are
shown in Larson, Black [31], in Michels et al. [40], in Lefrancois, Vaniscotte
[34], in Michels et al. [38], and in Atroshchenko, Gavrya [2]. Data with
methane can also be found in Zeininger [63] and in Konoki et al. [28].
f3 values for hydrogen, nitrogen, argon, and methane evaluated by using. the
method given in Alesandrini et al. [1] for a mixture with the 10% inert gases and
Hz/N z = 3 are shown in Fig. 1.5. The parameters have been evaluated by use of
the data above. f3 values are shown for 10, 20, and 30 MPa, respectively, as
a function of temperature. The same figure also shows experimental data for the
solubility of helium in the range given in Heise [22]. The f3 values decrease in the
order helium, hydrogen, nitrogen, argon, and methane, which corresponds to an
increase in solubility. The f3 values decrease with increase in temperature. The
f3 values are not very sensitive to the content of inert gases, but they increase
slightly (not significant in Fig. 1.5) with increasing inert gas level. The reason is
that argon and methane have higher solubilities than hydrogen and nitrogen.
Very few data sets exist with all components present. Zeininger [63] gives
data for a mixture with hydrogen, nitrogen and methane. The data agree well
with those at high pressure shown in Fig. 1.5 whereas there is some divergence at
low pressure. In this region, however, the solubility is very low.

Fig. 1.5 Phase equilibrium f3 val-

ues for H 2-N 2 -Ar-CH 4 mixtures
(67.5% H 2 • 22.5% N 2 • 3% Ar,
p=10MPo p=20MPa p=30MPa
7% CH 4 ).
240 260 280 300 320
Temperature in K
12 L. J. Christiansen

1.6 Transport Properties

The transport properties to be discussed in the design of ammonia plants are the
viscosity, the thermal conductivity, and the diffusion coefficients, which are used
in calculation of transport rates in order to determine equipment sizes. The
prediction of the properties for the gases are based on kinetic gas theory,
whereas the properties of liquid ammonia are predicted by empirical correla-
tions based on experimental data.
The properties of liquid ammonia with small amounts of dissolved gases are
calculated by mixing the properties of the pure liquid NH3 with the correspond-
ing properties of the dissolved gases calculated by the kinetic gas theory as
explained below. The mixing rule is the logarithmic mean.
The calculation of transport properties of gases can be divided into three
parts similar to those used in the derivation of thermodynamic properties. The
first part deals with the pure component low pressure values, which are func-
tions of molecular parameters and temperature only. The second part is calcu-
lation of low pressure mixture properties using appropriate mixing rules, and
the third part is the correction for influence of presrure. A detailed discussion of
the kinetic gas theory is given in Hirschfelder et al. [23], and in Reid et al. [50].
The recommendations given in the latter reference are used closely in the
following, since they give a good agreement with the available data for ammonia
synthesis mixtures.

1.6.1 Low Pressure Properties

The pure component low pressure viscosity is predicted from the Chapman-
Enskog equation which includes the effect of intermolecular forces. The equa-
tion for the viscosity I'J is:

1'J=26.69~
v (J
(17)

where Mw is the molecular weight, T the temperature in K, (J the hard-sphere

diameter in A and Qv the collision integral which is calculated from an inter-
molecular potential. Since ammonia is a polar gas it is necessary to use the
so-called Stockmayer potential, which is identical with the Lennard-Jones
potential but includes a term correcting for dipole-dipole interactions. The
Stockmayer potential is then evaluated as a function of temperature. It contains
two molecular parameters, the characteristic energy of interaction, and the
dipole moment. The thermal conductivity of pure gases at low pressure can also
be estimated from kinetic gas theory. The theory has been used by Bromley [6]
to develop equations for the thermal conductivity of gases depending on the
molecular structure. The equation gives the thermal conductivity A, when the
 1 Thermodynamic Properties in Ammonia Synthesis 13

viscosity YJ is known in the form of the ratio AMw/YJ which is equal to a function
of the heat capacity and the various modes of vibration. The method of Bromley
[6] gives acceptable results for the components present in ammonia synthesis
gas.

1.6.2 Mixture Properties

The binary diffusion coefficients are also predicted using kinetic gas theory. The
potential function used in evaluation of the collision integral is that given by
Neufeld et al. [43] with the modification for polar gases given by Brokaw [5].
The low pressure mixture properties are calculated using an appropriate mixing
rule, which also has been derived from kinetic gas theory. The formula includes
a binary interaction term. For viscosity the interaction parameter given by
Wilke [60] can be used. For thermal conductivity the form given by Lindsay,
Bromley [35] for the binary interaction parameter can be used, and for the
diffusion coefficients the formula given by Wilke [59] can be used.

1.6.3 Dependence on Pressure

The correction for the pressure dependence of transport properties of gases can
be tnade by use of various correlations. These include the gas density which in
turn is calculated from the compressibility factor. For viscosity the method
proposed by Dean, Stiel [9] can be used, and for the thermal conductivity the
method proposed by Stiel, Thodos [54] can be used. For the bulk diffusion
coefficient it is normally assumed that it is inversely proportional to the density.
Viscosities of hydrogen-nitrogen and of hydrogen-ammonia mixtures have
been reported in Pal, Barua [46]. Dembovskii [10] has reported viscosity data
for mixtures of ammonia, hydrogen, and nitrogen. Thermal conductivities of
nitrogen-hydrogen-ammonia mixtures are given by Golubev, Kiyashova [15].
Reference is also given to the work by Tsederberg [55]. Binary diffusion
coefficients for hydrogen-ammonia and nitrogen-ammonia are given in Mason,
Monchick [37]. The agreement between these experimental data and those
calculated by the above-mentioned methods is acceptable for design purposes.

1.7 References

1. Alesandrini CG, Lynn S, Prausnitz JM (1972) Ind Eng Chern Process Design Develop 11: 253
2. Atroshchenko VI, Gavrya NA (1959) Zh Prikl Khim 32: 100; J Appl Chern [USSR] 32: 100
3. Bartlett EP, Cupples HL, Tremearne TH (1928) J Am Chern Soc 50: 1275
14 L. 1. Christiansen

4. Bartlett EP, Hetherington HC, Kvalnes HM, Tremearne TH (1930) 1 Am Chern Soc 52: 1363
5. Brokaw RS (1969) Ind Eng Chern. Process Design Develop 8: 240
6. Bromley LA (1952) UCRL-1852: 1
7. Christiansen LJ, Kjrer 1 (1982) Enthalpy Tables of Ideal Gases, Haldor Tops0e A/S, Copen-
hagen pp. 1
8. U. S. Department of Commerce (1923) Bur Std [U.S.] Circ No. 142: 1
9. Dean DE, Stiel LI (1965) Am Inst Chern Eng 1 11: 526
10. Dembovskii VV (1968) Zavodsk Lab 34: 42; Ind Lab (USSR) 34: 52
11. Gillespie LJ (1925) 1 Math Phys 4: 84
12. Gillespie LJ, Beattie lA (1930) Phys Rev 36: 743
13. GillesIJie LJ, Beattie lA (1930) Phys Rev 36: 1008
15. Golubev IF, Kiyashova VP (1979) Tr GIAP No. 52: 57
16. Haar L (1968) 1 Res Natl Bur Std A 72: 207
17. Haar L, Gallagher lS (1978) 1 Phys Chern Ref Data 7: 635
18. Haber F, Le Rossignol R (1907) Ber Bunsenges Physik Chern 40: 2144
19. Haber F, Tamaru S (1915) Z Electrochem 21: 191
20. Haber F, Tamaru S, Oeholm LW (1915) Z Electrochem 21: 206
21. Haber F, Tamaru S, Ponnaz C (1915) Z Electrochem 21: 89
22. Heise F (1972) Ber Bunsenges Physik Chern 76: 938
23. Hirschfelder 10, Curtis CF, Bird RB (1954) Molecular Theory of Gases and Liquids, Wiley, New
York, pp. 1
24. Dow Chemical Corp (1971) lANAF Thermochemical Tables. 2nd Ed. NSRDS NBS-37: 1
25. Kaminishi G-I (1965) Intern Chern Eng 5: 749
26. Kazarnovskii YaS, (1945) Zh Fiz Khim 19: 392; CA. (1946) 1727
27. Kazarnovskii YaS, Karapet'yants MK (1941) Zh Fiz Khim 15: 966; CA. (1942) 6884
28. Konoki K, Takeuchi K, Kaminishi G-I, Toriumi T (1972) 1 Chern Eng 1apan 5: 103
29. Landolt-Bornstein (1980) 6th Ed. Pt. 4C 2: 189
30. Larson AT (1924) 1 Am Chern Soc 46: 367
31. Larson AT, Black CA (1925) 1 Am Chern Soc 47: 1015
32. Larson AT, Dodge RL (1923) 1 Am Chern Soc 45: 2918
33. Lefrancois B, Vaniscotte C (1960) Chaleur Ind No. 419: 183
34. Lefrancois B, Vaniscotte C (1960) Genie Chim 83: 139
35. Lindsay AL, Bromley LA (1950) Ind Eng Chern 42: 1508
36. Martin 11, Hou Y-C (1955) Am Inst Chern Eng 1. 1: 142
37. Mason EA, Monchick L (1962) 1 Chern Phys 36: 2746
38. Michels A, Dumoulin E, Th. Van Dijk 11 (1959) Physica 25: 840
39. Michels A, Dumoulin E, Th. Van Dijk 11 (1961) Physica 27: 886
40. Michels A, Skelton GF, Dumoulin E (1950) Physica 16: 831
41. Michels A, Wassenaar T, Wolkers Gl, de Graaf W, Louwerse P (1953) Appl Sci Res A 3: 1
42. Michels A, Wassenaar T, Wolkers G1, van Seventer W, Venteville Al (1954) Appl Sci Res A
4: 180
43. Neufeld PD, lanzen AR, Aziz RA (1972) 1 Chern Phys 57: 1100
44. Newton RH (1935) Ind Eng Chern 27: 302
45. Nielsen A (1968) An Investigation on Promoted Iron Catalysts for the Synthesis of Ammonia.
3rd Ed., Gjellerup, Copenhagen, pp. 1
46. Pal AK, Barua AK (1967) 1 Chern Phys 47: 216
47. Reamer HH, Sage BH (1959) 1 Chern Eng Data 4: 152
48. Reamer HH, Sage BH (1959) 1 Chern Eng Data 4: 303
49. Reddy KV, Husain A (1980) Ind Eng Chern Process Design Develop 19: 580
50. Reid RC, Prausnitz, 1M, Sherwood TK (1977) The Properties of Gases and Liquids 3rd Ed.
McGraw-Hili, New York pp. 1
51. Rossini FD (1953) Selected Values of Physical and Thermodynamic Properties of Hydrocarbons
and Related Compounds. Carnegie Press, Washington pp. 1
52. Sage BH, Olds RH, Lacey WN (1948) Ind Eng Chern 40: 1453
53. Schulz G, Schaefer H (1966) Ber Bunsenges Physik Chern 70: 21
54. Stiel LI, Thodos G (1964) Am Inst Chern Eng 1 10: 26
55. Tsederberg NV (1965) Thermal Conductivity of Gases and Liquids, MIT Press, Cambridge,
Mass., pp.l
56. Wiebe R, Gaddy VL (1938) 1 Am Chern Soc 60: 2300
 1 Thermodynamic Properties in Ammonia Synthesis 15

57. Wiebe R, Treamearne JH (1933) J Am Chern Soc 55: 975

58. Wiebe R, Treamearne JH (1934) J Am Chern Soc 56: 2357
59. Wilke CR (1950) Chern Eng Progr 46: 95
60. Wilke CR (1950) J Chern Phys 18: 517
61. Winchester LJ, Dodge BF (1956) Am Inst Chern Eng J 2: 431
62. Zander, M. Thomas W (1979) J Chern Eng Data 24: 1
63. Zeininger H (1973) Chern Ing Tech 45: 1067
Chapter 2
Structure and Surface Chemistry of Industrial
Ammonia Synthesis Catalysts

Per Stoltze
Haldor Tops0e Research Laboratories Lyngby, Denmark

Contents

2.1 Introduction................................ 21

2.2 The Unreduced Catalyst . . . . . . . . . . . . . . . . . . . . . . . .. 21

2.2.1 Structure.............................. 22
2.2.1.1 Magnetite . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.1.2 Grain Boundaries . . . . . . . . . . . . . . . . . . . 22
2.2.1.3 Wustite.......................... 23
2.2.1.4 Ferrites.......................... 23
2.2.1.5 Glass phases . . . . . . . . . . . . . . . . . . . . . . 23
2.2.2 Location of Promoters in the Unreduced Catalyst . . . . . .23
2.2.2.1 Aluminum........................ 23
2.2.2.2 Calcium . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.2.3 Potassium........................ 24
2.2.2.4 Other Additives. . . . . . . . . . . . . . . . . . . . . 24
2.2.3 Texture............................... 25
2.2.4 Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . 26

2.3 The Reduction Process . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.3.1 Reduction Temperature . . . . . . . . . . . . . . . . . . . .. 26
2.3.2 Development of Properties During Reduction . . . . . . .. 27
2.3.2.1 Bulk Structure . . . . . . . . . . . . . . . . . . . . . 27
2.3.2.2 Surface Properties . . . . . . . . . . . . . . . . . .. 28
2.3.3 Mechanism and Kinetic Models . . . . . . . . . . . . . . .. 28
3.3.1 Single Crystal Studies . . . . . . . . . . . . . . . . .. 29
2.3.4. Influence of Promoters on the Reduction Process . . . . . 29
2.3.4.1 Aluminium . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.4.2 Alkali........................... 29
2.3.4.3 Other Elements . . . . . . . . . . . . . . . . . . . .. 30

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
18 P. Stoltze

2.4 Bulk Structure of the Reduced Catalyst . 30

2.4.1 Iron . . . . . . . . 30
2.4.2 Aluminium. 31
2.4.3 Calcium . . . . . 31

2.5 The Surface Structure of the Reduced Catalyst 31

2.5.1 The Pore System 32
2.5.2 Particle Size . . . . . 32
2.5.3 Surface Area . . . . . 32
2.5.4 Effect of Promoters . 33
2.5.4.1 Aluminium 33
2.5.4.2 Calcium . . 34
2.5.4.3 Potassium. 35
2.5.4.4 Other Promoters 36
2.5.5 Models of Structural Promotion 36
2.5.6 Alloying with Transition Metals 37

2.6 Chemisorptive Properties of the Catalyst 38

2.6.1 Chemisorption of H . . . . . . . . 38
2.6.1.1 Structure of Chemisorbed H 38
2.6.1.2 Thermodynamics .. 39
2.6.1.3 Adsorption Kinetics . . . . . 40
2.6.1.4 Desorption Kinetics . . . . . 40
2.6.1.5 Properties of Chemisorbed H ...... . 41
2.6.1.6 Effect of Promoters . . . . . . . . . . . . . 41
2.6.1.7 Effect of Preadsorbed Species ...... . 42
2.6.1.8 The Hydrogen artha-para Conversion . . . . . . . 43
2.6.1.9 H2 + D2 Exchange . . . . . . . . . . . . . . . . . . . 43
2.6.2 Chemisorption of CO . . . . . . . . . . 43
2.6.2.1 Structure of Chemisorbed CO . 43
2.6.2.2 Thermodynamics .. 44
2.6.2.3 Adsorption Kinetics . 45
2.6.2.4 Desorption Kinetics . 45
2.6.2.5 Dissociation Kinetics 45
2.6.2.6 Properties of Chemisorbed CO . . . . . . . . . . . 45
2.6.2.7 Effect of Promoters . . . . . . . . . . . . . . . . . . 46
2.6.2.8 Effect of Preadsorbed Species . . . . . . . . . . . . 47
2.6.2.9 Correlation with Activity . . . . . . . . . . . . . . . 47
2.6.3 Chemisorption of CO 2 . . . . . . . . . . . . . . . . . . . . .. 47
2.6.3.1 Structure of Chemisorbed CO 2 . . . . . . . . . . . 48
2.6.3.2 Thermodynamics ... 48
2.6.3.3 Dissociation Kinetics . . . . . . . . . . . . . . . . 48
2.6.3.4 Effect of Promoters . . . . . . 48
2.6.3.5 Effect of Preadsorbed Species 48
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 19

2.6.4 Physisorption of N2 . . . . . . . . . . 48
2.6.4.1 Structure of Physisorbed N2 49
2.6.4.2 Thermodynamics.. 49
2.6.4.3 Adsorption Kinetics . . . . . 49
2.6.4.4 Desorption Kinetics . . . . . 49
2.6.4.5 Kinetics of Conversion into aN 2 • 49
2.6.4.6 Properties of Physisorbed N z .......... . 49
2.6.5 Molecular Chemisorption of N2 . . .. . . . . . . . . . . . 50
2.6.5.1 Structure of Chemisorbed N 2 50
2.6.5.2 Thermodynamics.. 50
2.6.5.3 Adsorption Kinetics . . . . . . 50
2.6.5.4 Desorption Kinetics . . . . . . 50
2.6.5.5 Properties of Chemisorbed N z 50
2.6.5.6 Effect of Promoters . . . . . . 51
2.6.5.7 Effect of Preadsorbed Species 51
2.6.6 Dissociative Chemisorption of N .... 51
2.6.6.1 Structure of Chemisorbed N 52
2.6.6.2 Thermodynamics .. 52
2.6.6.3 Adsorption Kinetics . . . . . 53
2.6.6.4 Desorption Kinetics . . . . . 54
2.6.6.5 Hydrogenation of Chemisorbed N . 54
2.6.6.6 Properties of Chemisorbed N 54
2.6.6.7 Effect of Promoters . . . . . . 55
2.6.6.8 Effect of Preadsorbed Species 56
2.6.6.9 Isotopic Exchange . . . . . . 56
2.6.7 Kinetic Models of N2 Chemisorption . 57
2.6.8 Chemisorption of NH3 . . . . . . . . . 59
2.6.8.1 Structure of Chemisorbed NH3 59
2.6.8.2 Thermodynamics.. 59
2.6.8.3 Adsorption Kinetics 60
2.6.8.4 Desorption Kinetics 60
2.6.8.5 Dissociation..... 60
2.6.8.6 Properties of Chemisorbed NH3 60
2.6.8.7 Effect of Preadsorbed Species . . . . . . . . . . . . 61
2.6.9 Adsorption of N2H4 . . . . . . . . . . . . . . . . . . . . . . . 61
2.6.10 Chemisorption of O 2 . . . . . . . . . . . . . . . . . . . . . . . 61
2.6.10.1 Structure of Chemisorbed 0 62
2.6.10.2 Thermodynamics . . 62
2.6.10.3 Adsorption Kinetics . . . . . 63
2.6.10.4 Desorption Kinetics .... . 63
2.6.10.5 Properties of Chemisorbed 0 63
2.6.10.6 Oxygen Isotopic Exchange . 64
2.6.10.7 Effect of Promoters . . . . . . 64
2.6.10.8 Effect of Preadsorbed Species. 64
20 P. Stoltze

2.6.11 Adsorption of H 2 0. 65
2.6.12 Adsorption of H 2S 65

2.7 The Mechanism of NH3 Synthesis 65

2.7.1. Nature of the Active Structure. 66
2.7.1.1 Structural Sensitivity. . 67
2.7.1.2 The Effect of K . . . . . . . . 68
2.7.2 Results from Chemisorption Studies. 69
2.7.3 Kinetic Models of NH3 Synthesis . . 70
2.7.3.1 Formulation of the Models. 70
2.7.3.2 Input Parameters. . . . . . . . . 72
2.7.3.3 Test................ 72
2.7.4 The Nature of Reaction Intermediates . . . . . . . . . . . . 73
2.7.4.1 Nitrogen Dissociation . . . 74
2.7.4.2 Stability of Intermediates 74
2.7.4.3 Coverage by Intermediates 76
2.7.4.4 Lifetime of Intermediates 78
2.7.5 The Nature of Rate Limiting Step . 78
2.7.5.1 N2 Adsorption as Rate Limiting Step. 79
2.7.5.2 N2 dissociation as Rate Limiting Step . . . . . . . 79
2.7.5.3 N2 Hydrogenation as Rate Limiting Step . . . . . 79
2.7.5.4 Changes in Rate Limiting Step. . . . . . . . . . 80
2.7.6 Kinetics of NH3 Synthesis 80
2.7.6.1 Reaction Orders . . . . . . . . . 81
2.7.6.2 Activation Energy . . . . . . . . 82
2.7.7 Deuterium Isotope Effect for NH3 Synthesis. 84
2.7.8 The Stoichiometric Number for Ammonia Synthesis 84
2.7.9 Poisoning 85

2.8 References . . . . 88
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 21

2.1 Introduction

This chapter deals with the structure and surface chemistry of industrial ammo-
nia synthesis catalysts. Results on catalyst models and single crystal surfaces are
included to the extent that they illuminate the behavior of industrial catalysts.
The industrial catalyst is prepared by fusion. The catalyst may be supplied in
the unreduced state after crushing and screening to the desired particle size or
the catalyst may be reduced and subsequently stabilized by controlled oxidation
in the catalyst factory. Although the reduced catalyst is pyrophoric, the
prereduced catalyst can be safely handled. In the ammonia synthesis plant the
catalyst is activated by reduction with a mixture of hydrogen and nitrogen as the
final step in the start-up procedure for the plant. The reduction of the
prereduced catalyst is faster and simpler than the start-up of the unreduced
catalyst.
A number of useful reviews on the structure and properties of ammonia
synthesis catalysts [1-18J and on ultra-high vacuum investigations related to
ammonia-synthesis [10, 19-25J have been published.
Catalyst constituents, which have little or no catalytic activity by themselves,
but which increase the catalytic activity for the catalyst are referred to as
promoters. Promoters which increase the catalytic activity primarily by increa-
sing the active area of the sample are referred to as structural or textural
promoters. Promoters which increase the activity of the catalyst primarily by
increasing the reaction rate per area are referred to as chemical or electronic
promoters. Constituents, which decrease the activity of the catalyst when pre-
sent in small amounts are referred to as poisons.
Catalysts containing Fe, one structural promoter and no electronic pro-
moter are commonly referred to as singly promoted. Catalysts containing Fe, one
structural promoter and one electronic promoter are referred to as double
promoted, while catalysts containing Fe, more than one structural promoter and
one or more electronic promoters are referred to as multiply promoted.
In the following, the notation e.g. (Fe, AI, K) will indicate a sample contain-
ing the elements Fe, AI, K and possibly non-metallic elements. The sequence
indicates the relative concentrations of the metals, the first metal being the most
abundant. The asterisk (*) represents a surface site; X * represents a species
X adsorbed .on a surface site.

2.2 The Unreduced Catalyst

The unreduced catalyst consists of oxides of iron with up to a few percent of AI,
Ca and K. Other elements may be present in small amounts.
22 P. Stoltze

2.2.1 Structure

The unreduced catalyst is produced by melting a mixture containing the differ-

ent elements. The main phase of the resulting product is magnetite [14, 15,
26-29]. The details of the preparation procedure [30, 31] and the homogeneity
in the unreduced state [32-35] have been found to influence the properties.
From an examination of the shapes of the magnetite grains [14,28], it is
apparent that this phase is the first to solidify during the cooling of the melt. The
grain boundaries [36,37] between the magnetite grains may contain several
different phases which will be discussed below.

2.2.1.1 Magnetite
For an industrial catalyst, energy dispersive X-ray analysis, X-ray powder
diffraction, optical microscopy and Mossbauer spectroscopy show that part of
the Al and Ca atoms are dissolved in the magnetite lattice [15,27,28,38-43].
The lattice constant of the magnetite phase of an industrial catalyst is 8.377 kX
[15]. The lines in the X-ray powder diffractions diagram are broadened [15]; the
broadening is independent of particle size [15].
From the Mossbauer spectrum [27,44,45] and the X-ray power diffraction
diagram [39] of the unreduced catalyst, it has been estimated that 85% of the Al
in the unreduced catalyst is dissolved in the magnetite. Evidence for the dissolu-
tion of K[41,46], Mg[41,46], V[41], Si[41], W[46], and Mo[46] in the
magnetite has been reported. However, due to the large size of K +, only a small
amount of K is found in the magnetite phase of the industrial catalyst.
Additional information on the structure of the magnetite phase comes from
the study of catalyst models, in particular of (Fe, AI) solid solutions. Mossbauer
spectroscopic studies of Fe304 [47] and of unstochiometric Fe-spinels [44]
have been reported.
For unreduced precipitated (Fe,AI) samples, solid solutions of Al in Fe-
oxides and of Fe in AI-oxides may be observed [48, 49] depending on composi-
tion [48] and the preparation method [49]. In experimental (Fe,AI)-oxide
samples, dissolution of Al in the magnetite has been shown by X-ray powder
diffraction [50]. The lattice constant decreases from 8.413 A to 8.365 A for
Fe304-Ah03 up to 13 atom % Al 2 0 3 and is then constant [51] indicating the
formation of a saturated solid solution with segregation of excess Ah03 as
a separate phase.

2.2.1.2 Grain Boundaries

The grain boundary regions may constitute about 7% of the volume of the
unreduced catalyst [46], and may contain much higher concentrations of
promoter [14, 52] than the magnetite. The phases detected in the grain bound-
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 23

aries are small amounts of wustite [14, 15,28,40], calcium ferrites with dis-
solved promoters and a glass phase rich in silicon [26,40,41]).

2.2.1.3 Wustite
Based on X-ray powder diffraction studies, some authors have concluded that
the structure of the wustite is that of natural wustite [15, 45]. From energy
dispersive X-ray analysis and X-ray powder diffraction studies, others have
found evidence for the dissolution of AI[28,40], Ca[28], Mg[41], K[40],
V[41], or Si[41] in the wustite. In fused (Fe,AI)-oxide samples the grain
boundaries contain small but significant amounts of wustite [40].

2.2.1.4 Ferrites
Two calcium ferrites differing in their content of dissolved promoters may
coexist in the grain boundaries [28,41]. The dissolution of AI[28], K[28],
Mg[41], V[41], or Si[41] in the calcium ferrites has been reported.

2.2.1.5 Glass Phases

The grain boundaries in the unreduced catalyst contain a glass phase. The main
element of the glass phase is Si[26] or Fe [41]. The composition 17% SiOl, 18%
FeO, 9% CaO, 30%AI l 0 3 , 24%K l O[40] has been found from energy disper-
sive X-ray analysis.

2.2.2 Location of Promoters in the Unreduced Catalyst

While the composition of the solid solutions which constitute the unreduced
catalyst is not known in detail, an extensive knowledge exists on the location of
various additives both for catalyst models and for an industrial catalyst.

2.2.2.1 Aluminum
Evidence for the presence of Al dissolved in the magnetite phase has been found
by X-ray powder diffraction [27,28, 38-42, 53] and by chemical analysis of
powders of varying particle size [54]. The solubility of Al in the (Fe,Alh
04-phase has been determined to be 30 atom% AI[55] from measurement ofthe
Curie temperature and 50 atom % AI[55], or 67 atom % AI[39] from measure-
ment of the X-ray powder diffraction lattice constant. Other studies have
indicated homogeneous solution of Al in magnetite, at least for small amounts of
Al [56] and not too high temperatures [57].
In (Fe, AI) oxide catalyst models, some or all of the Al is dissolved in the
Fe-oxide phases, as solid solutions between Fe-oxide and AI-oxide are readily
24 P. Stoltze

formed [58,59]. The solubility of Al in (Fe,Alh03 is < 15 atom % Al [58, 59]

determined by X-ray powder diffraction, and < 9.2 atom % Al [58] determined
by M6ssbauer spectroscopy. In calcined (Fe,Co,AI) samples, M6ssbauer spec-
troscopy shows ocFe203 with dissolved Ah03 for 0.10% Co and Fe-Co-spinel
with a little ocFe203 for 10-100% Co [60].
In the industrial catalyst, smaller amounts of Al are found dissolved in the
calcium ferrites [28], and possibly in the wustite [28, 40] and in the glass phase
[40].

2.2.2.2 Calcium
For an industrial catalyst, energy dispersive X-ray analysis, X-ray powder
diffraction and optical microscopy indicate that Ca is found dissolved in the
magnetite [28], in calcium ferrites [28,41], possibly in the glass phase [40], and
in the wustite [28]. Ca has been found in grain boundaries in a sintered (Fe,Ca)
oxide catalyst [61].

2.2.2.3 Potassium
In the unreduced catalyst [11, 14,40,61,62] and in (Fe,Mg) catalyst models
[63], K is found in the grain boundaries. Additional amounts of K may be
present as K-ferrites [64]. K has been reported to be associated with Si [54].
X-ray powder diffraction indicates that K is insoluble in magnetite [56].
The addition of Al makes the distribution ofK more homogeneous [62], and
. the addition of Al or Zr decreases the volatility of K during preparation [65].
Both observations indicate the potential for K to react with acid oxides in the
catalyst. CaO and Si0 2 decrease the water solubility of K [66]. Annealing
increases the water solubility of K [66].
The amount of K which may be extracted by H 20 has been reported to
increase [67,68] or to decrease [66] by the addition of Si; the amount of
K which may be extracted by H 20 is decreased [66] by the addition ofCa, and
decreases [67,68] or increases [66] by heating the unreduced catalyst. By
scanning electron microscopy and energy dispersive X-ray analysis it was found
that K segregates to the outer part of the catalyst particles with storage and
prolonged use [69].

2.2.2.4 Other Additives

After sintering of a (Fe,Co,AI) based catalyst containing 15 atom % Co, Fe304
and CoFe204 were found [70]. In calcined (Fe,Co,AI) samples, M6ssbauer
spectroscopy shows ocFe203 with dissolved Ah03 for 0.10% Co, and Fe-Co-
spinel with a little ocFe203 for 10-100% Co [60]. The surface area of the
unreduced sample decreases with an increasing Co content [71, 72]. For precipi-
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 25

tated (Fe,Co,AI) the Fe (II) : Fe (III) ratio increases [73J with increasing Co
content.
In the unreduced catalyst, Mg is present dissolved in magnetite [63J and as
an unidentified phase in grain boundaries [63].
The effect of Mn on the precursor has been studied [74J,
In Mo containing samples, Mo has been detected as grains ofK 2 Mo0 4 [75J,
CaMo0 4 [75J, FeMo0 4 [75J Fe3(Mo04h [76J, or Mo0 3 [76J depending on
the composition and preparation procedure. Mo is soluble in magnetite [77].
For precipitated (Fe,Ni,AI), Ni increases the Fe (II) : Fe(III) ratio [73].
In the unreduced catalyst, Si is found in the calcium ferrites [41J and in the
glass phase [41, 26]. Smaller amounts may be present dissolved in the magnetite
[41J or in the wustite [41]. Evidence for the association of Si with K has been
reported [54]. Si is insoluble in magnetite in the absence of other promoters
[56]. The addition of Si decreases the solubility of basic oxides in magnetite
[56].
For W containing catalysts, W is found dissolved in the magnetite [46].

2.2.3 Texture

The size of the magnetite grains in the unreduced catalyst is rather variable [78].
The cross sectional area of magnetite in KM1 is (1.28 ± O.l4)·1O- 2 mm 2 deter-
mined from planimetry [79]. The density is 4.8 gjcm 3 [14]. The porosity is
negligible [14].
For the prereduced catalyst the density is 3.73 gjcm 3 [14]. The porosity is
0.11 cm 3 jg, i.e., 41% [14J, and electron microscopy [11, 14,80J shows the
presence of a well developed pore system.
Precipitated (Fe,AI) catalyst models in the unreduced state contain pores
with a 19-20 A radius and approximately 70 A radius [48J, while in the
prereduced state these models have maxima in pore volume distribution at 20 A
and at 140 A [48]. The 20 A peak is thus unaffected by the reduction process.
The BET area is smaller for a catalyst prepared by sintering than for
a similar catalyst prepared by precipitation [81]. The BET area decreases with
increasing calcination temperature [58J, and increases with increasing Al con-
centration [61J for precipitated catalysts. For samples calcined at 600 DC, the
BET-area increases from 13 m 2 jg at 0% Al 2 0 3 to 215 m 2 jg at 88% Al 2 0 3 [48].
From a X-ray photoelectron spectroscopy study, the surface composition
3.2% Fe, 33.2% K, 8.4% AI, 3.9% Ca and 51.3% 0 (atomic %) was found [82J
for the unreduced catalyst. The surface is enriched in K and Al compared to the
bulk [52]. The prereduced catalyst shows more Fe in the surface by X-ray
photoelectron spectroscopy than the unreduced catalyst [52]. Energy dispersive
X-ray analysis of prereduced catalyst shows that the surface consists mainly of
iron oxide [83J; AI, Ca and Si are inhomogeneously distributed in the surface
[83].
26 P. Stoltze

2;2.4 Physical Properties

For unpromoted Fe-oxide the magnetization is maximum for

Fe (II) : Fe (III) = 0.50 (magnetite) [84]. The Curie temperature is constant for
Fe (II) : Fe (III) = 0.352 to l.278 [84]. The Curie temperature decreases from
575°C to 535 °C for Fe304 by the addition of up to 13 atom % Al 2 0 3 [51]. The
Curie temperature is 535°C for higher AI-concentrations and is then constant
[51]. For singly promoted samples containing Na, K, Cs, Ba, B, Al or Si, the
Curie temperature is unchanged compared to unpromoted magnetite [84,85].
The magnetization is reduced relatively to pure magnetite [84,85]. For doubly
promoted samples containing Al + Na, Al + K, B + K, Al + Ba, or Al + Si,
both the Curie temperature and the magnetization is lower than for pure
magnetite [84]. Addition of Li, Na, K, Mg, Ca, Ba or Si to (Fe, AI)-oxide samples
either does not affect the Curie temperature, or decreases it slightly compared to
(Fe,AI)-oxide [85].
The mechanical strength increases with increased Fe(II) content [86].
The heat capacity of an unreduced (Fe, Ca)-oxide sample displays a second
order phase transition at 718 K [87].
The electrical conductivity of the unreduced catalyst has been studied [88].

2.3 The Reduction Process

The reduction of the industrial catalyst has been extensively studied, [89-92].
The reduced catalyst mainly consists of metallic Fe while the promoters
remain in their oxidic state. The reduction process serves two purposes, firstly
the surface of metallic iron is the active structure, and secondly the removal of
the oxygen makes the material porous and increases the surface area by a large
amount.

2.3.1 Reduction Temperature

During a temperature programmed reaction in hydrogen, an industrial catalyst

is reduced in the range 500-900°C with a maximum rate at 800°C [73]. The
prereduced catalyst reduces under the same experimental conditions at 523 °C
[73]. The highest activity is reached by slow reduction in a specific temperature
interval, this interval is 380-400 to 500-525°C [93].
For (Fe, AI)z03 with 5% AI, a peak at 950°C is interpreted as the reduction
of FeAl 2 0 3 [73].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 27

The reduction of 10% Fe supported on Alz0 3 has been studied by Mos-

sbauer spectroscopy [94]. At 460 DC Fez03 is reduced to Fe304, at 470 to 600 DC
Fe304 is reduced to aluminates without consuming Hz, at 790 DC Fe (III) is
reduced to Fe(O) and at 850°C Fe(II) is reduced to Fe (0).
Cyanide based Fe catalysts reduce at 300-350 DC [95].
The reduction proceeds at a lower temperature [96] and higher catalytic
activity is obtained for catalyst reduced as small grains.
For (Fe,Al) samples the reduction proceeds at higher temperatures and the
activity is improved [12] or unchanged [19] after reduction in hydro-
gen + nitrogen mixtures compared to reduction in pure hydrogen. For un-
promoted Fe the activity is improved [19] after reduction in mixtures of
hydrogen and nitrogen compared to reduction in pure hydrogen
The kinetics of reduction for unpromoted magnetite in hydrogen is unaffec-
ted by the presence of water [97] while the reduction of an industrial catalyst is
somewhat inhibited by water [97,98]. The presence of water during the reduc-
iton increases the average pore diameter [99] and decreases the activity of the
catalyst. The activity of the catalyst increases with increasing space velocity
during reduction [100]. The space velocity during reduction is more critical for
samples reduced as small grains [100].
The rate of reduction increases with increasing Fe(II) content for (Fe,Al,K)-
samples [101]. For multiply promoted samples [102, 103] the rate has also been
reported to increase with increasing Fe(II) content. However, the rate does
depend on the detailed structure of the catalyst.

2.3.2 Development of Properties During Reduction

2.3.2.1 Bulk Structure

The fully reduced catalyst consists of metallic Fe with little or no oxides of Fe
present.
By the use of Mossbauer spectroscopy [27,80, 104, 105] it has been found
that wustite reduces before magnetite. The Fe formed from wustite has a more
rounded shape than Fe formed from magnetite [28]. The reduction of the
wustite leaves behind a system of pores with 600-2000 A diameter [80]. For
samples with an initially high concentration (27%) of wustite, the reaction zone
is more diffuse than for samples of low initial wustite content.
The formation of wustite under conditions where wustite is metastable has
been reported from magnetic measurements during reduction [106-108] and
from X-ray powder diffraction [109] during reduction. Evidence against the
formation of wustite has been reported from chemical analysis [110] and
Mossbauer spectroscopic studies [27, 111].
For a (Fe,Al) based catalyst the presence of a paramagnetic phase under
reduction has been detected by Mossbauer spectroscopy [112]; this phase was
28 P. Stoltze

interpreted as FeAl 2 0 4 inclusions in Fe [112]. Ca-ferrites are only partially

reduced [40]. M6ssbauer spectroscopy shows the formation Fe304 as an
intermediate phase [111] during reduction of (Fe,Al)z03 in H 2 . A minimum in
the reduction rate was observed [111] at a degree of reduction corresponding to
the quantitative formation of Fe304'

2.3.2.2 Surface Properties

The outer shape of the grains is conserved during reduction and the porosity of
the reduced sample thus has a simple relation to the amount of iron oxide in the
unreduced sample.
For an industrial catalyst the pore volume [113, 114] increases during the
entire reduction. The pore volume distribution of the reduced catalyst has
a maximum at a pore radius of 100-120 A and at a 260-430 A [99, 115]. The
smaller pores are formed by the reduction of magnetite [99]; the larger pores are
formed by the reduction of wustite [99].
For an industrial catalyst the pore radius is constant at 50-100% reduction
[114]. For (Fe,Mg), (Fe,Si), (Fe,Cr), (Fe,K) samples the pore radius is constant
at 20-90% reduction and increases at 95-100% reduction [113, 116].
For an industrial catalyst the BET area increases during the entire reduction
[114], increases at first and passes through a maximum at 90% reduction [113]
or 95% reduction [117, 119].
After reduction, chemisorbed hydrogen has been detected by Laser Raman
spectroscopy [120].
The CO chemisorption area increases slowly at 30-90% reduction and then
increases rapidly at 96-100% reduction [113, 117-119, 121].
The oxygen chemisorption for an industrial catalyst is proportional to the
degree of reduction [122] in the later stages of reduction.
The catalyst activity increases slowly at 30-90% reduction and then in-
creases more rapidly than the CO-area at 96-100% reduction [117, 119]. This
observation has been interpreted [117, 119] as the manifestation of the poison-
ous effect of the oxides on the surface.
Reduction at 1073 K increases both the NH3 synthesis and the N 2 -
chemisorption rate; this has been interpreted as the reduction of Fe(II)-spinel
[123].

2.3.3 Mechanism and Kinetic Models

The reduction of samples of an industrial catalyst with a small (4%) wustite

content has been found by gravimetry [115, 124-126], M6ssbauer spectroscopy
[27, 127, 128], and electron microscopy [27,80,98, 115, 129, 130] to follow the
core-and-shell model.
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 29

A kinetic model involving the rate of interface reaction, pore-diffusion and

gas-film diffusion has been formulated for the core-and-shell reduction [115].
The enthalpy of adsorption of water on the reaction interface is 163.2 ± 45.2
kllmole in this model [115]. The core-and-shell model is invalid in the final
stages of reduction for samples containing high (27%) concentrations of wustite
[26,27,80] in the presence of small amounts of water [97, 131].
The reduction in a wet atmosphere has been found by scanning electron
microscopy to follow the cracking-core model [98, 132].
As the main phase of the reduced catalyst is iron, the influence of a magnetic
field on the reduction process has been studied [133].

2.3.3.1 Single Crystal Studies

The reduction of chemisorbed 0* on Fe(100) by Hz has been studied by Augeer
electron spectroscopy and LEED at 473 and 673 K [134]. The reaction pro-
ceeds via the formation of H* [134]; no reaction is detected if the surface is
completely covered by 0* initially [134]. The apparent activation energy of the
reaction is 59 ± 4 kllmole [134]. The transport of 0 to the surface is fast
compared to the surface reaction [134].

2.3.4 Influence of Promoters on the Reduction Process

While the reduction of unpromoted Fe proceeds at a relatively low temperature

and the reduced sample has a low porosity, the presence of the promoters leads
to a much higher porosity for the reduced samples. However, the presence of the
promoters also has a profound influence on the kinetics ofthe reduction process.

2.3.4.1 Aluminum
Al decreases the rate of reduction for (Fe,AI)-oxide samples [130], in particular
in wet atmospheres [135, 136]. For Al 20 3 supported catalyst models, the partial
dissolution of Al during impregnation leads to a more difficult reduction [137].

2.3.4.2 Alkali
M6ssbauer spectroscopic studies showed that alkali promotes the reduction of
Fe203 to Fe at 300°C [111, 138, 139]. Yet more direct measurements by
temperature programmed reactions demonstrate that K decreases the rate of
reduction for (Fe,K)-oxide samples [121] and for (Fe,AI,K)-oxide samples
[121]. The rate of reduction increases through the sequence (Fe,AI,M), M = Li,
Na, K, Rb, Cs for Fe203 based catalysts [139].
30 P. Stoltze

2.3.4.3 Other Elements

A number of elements, which are not present in an industrial catalyst, have been
reported to affect the reduction. The elements which have been reported to
increase the reduction temperature are Er [140], La [140], Mo [141~144], Pr
[140], rare earths [145], Sc [140], and Sm [146~148], while the elements which
have been reported to decrease the reduction temperature are Co [73], Cu
[149], Ni [73,150], Pd [151], and Re [152]. Ag [149] has been found to have no
effect on the reduction.

2.4 Bulk Structure of the Reduced Catalyst

Some information on the structure and texture of the active catalyst may be
inferred from studies of the spent catalyst [153], although the details of com-
position and structure may differ due to the violence of the reaction of the
reduced catalyst with air and due to structural changes during the oxidation.

2.4.1 Iron

In the catalyst [14, 15,27,41, 112, 154] Fe is present in the reduced state mainly
as the metal. The lattice constant of the iron is 2.8601 kX [15]. On the lean side
of the gas phase equilibrium of the synthesis gas mixture Fe films do not form
bulk nitrides [155]. On the rich side of the equilibrium Fe4N may be formed
[155].
In an industrial catalyst traces of unreduced Fe are detected by Mossbauer
spectroscopy [27,42]. These traces of Fe may be present in Ca ferrites with
dissolved promoters [28,41,156] or in the glass phase [41,156]. It has been
suggested that the glass phase is inactive in the formation of the active catalyst
[157].
For unpromoted Fe oxides, both Fe304 and FeO are completely reduced
[40] at 550°C in H 2 •
For a number of (Fe,AI,K) and (Fe,AI,Cs) catalysts, in situ EXAFS and
XANES [158, 159] indicate complete reduction of Fe in the reduced state of the
catalyst. For an (Fe,AI)-oxide catalyst model containing 3% ASI 2 0 3, Mos-
sbauer spectroscopy indicates complete reduction of Fe [112] while for an
(Fe,AI)-oxide catalyst model containing 10.2% A1 2 0 3, traces of Fe (II) have been
detected in the reduced state [154].
The degree of reduction for 0.05~ 15% Fe on Al 2 0 3 is 77~97% after reduc-
tion in 3H 2 + N2 at 1 atm, 673 K [160]. This indicates that Fe supported on
Al 2 0 3 is more difficult to reduce that Fe promoted with Al 2 0 3 and that (Fe,AI)
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 31

oxide samples of high Ah03 concentration may be poor models for the indus-
trial catalyst.
The X-ray powder diffraction diagrams show line broadening for Fe due to
particle size effects [14,15,27, 154]. The particl~ size for Fe is 300 A [15].
Others have interpreted the in situ X-ray powder diffraction diagram as evi-
dence for Fe being present as a metallic glass [161].

2.4.2 Aluminum

In the reduced catalyst Al 2 0 3 have been shown by Auger electron spectroscopy

[162] and X-ray photoelectron spectroscopy [52] to remain in its oxidic state.
The AI-containing phases have been interpreted as Ah03 segregated to the
space between the Fe-crystallites based on evidence from chemisorption
measurements [163] and X-ray powder diffraction studies [164]; as Al 2 0 3 and
FeAh04 based on evidence from EM and X-ray powder diffraction studies [48];
or as FeAl 2 0 4 present as paracrystalline defects in the Fe crystallites based on
evidence from secondary ion mass spectroscopy and Mossbauer studies
[50,154,165,166]. The presence of FeAl 2 0 4 as paracrystalline defects has been
challenged on the basis of Mossbauer spectroscopic studies [112, 165].
In reduced samples of (Fe,Alh03 with more than 2.4% A1 2 0 3, Ah03 is
observed in the X-ray powder diffraction diagram [167].

2.4.3 Calcium

In the reduced catalyst CaO has been shown by Auger electron spectroscopy
[162] and X-ray photoelectron spectroscopy [52] to remain in its oxidic state.
Chemisorption measurements [163] and X-ray powder diffraction studies [164]
show that during the reduction CaO segregate to the space between the Fe
crystallites.

2.5 The Surface Structure of the Reduced Catalyst

The outer shape of the grains is conserved during reduction [28, 110,
113,156,168,169], expands by up to 0.6% [168] or contracts by up to 0.5%
[168] during reduction depending on the composition and reduction temper-
ature.
32 P. Stoltze

2.5.1 The Pore System

The density of the reduced catalyst is 2.7-3.7 gjcm 3 [14]. The pore volume is
0.15 cm 3 jg [67, 6SJ, independent of the K concentration [[67,6S].
For the reduced catalyst, electron microscopy shows the presence of a well
developed pore system [92, 170, 171]. An industrial catalyst with pores of 100
and 300 A radius [14, 115J in the reduced state has been found to have 177 A
pores [l72J after passivation. The reduction of promoted magnetite creates
a system of pores with < 400 A diameter [SO]. The average pore radius
increases with increasing K concentration from 50-200 A (0% K) to 200-S00 A
(0.7S% K) after reducing at 600°C [67, 6S].
Precipitated (Fe,AI) catalyst models have pores with 19-20 A radius and
larger pores of 50-250 A radius [4S, 167].

2.5.2 Particle Size

The particle size has been studied by X-ray powder diffraction [173]. The cause
of the line broadening in the X-ray powder diffraction diagram has been
assigned to particle size effects [l64J, to the combined effect of particle size,
defects and strain [174J, or to the presence of paracrystalline defects
[166, 173, 175].
If the line broadening is assigned to particle size effects alone, the calculated
particle radius is 100-1000 A [53J, 175-250 A [174J, or ISO A [164]. For
catalyst models the particle radius is 305 ± 15 A for (Fe, AI) [154J 125-155 A for
(Fe,Mg) and 305 A for (Fe,Mg,K) [176].
If the line broadening in the X-ray powder diffraction diagram of an
industrial catalyst in the reduced state is assigned to the presence of paracrystal-
line defects, the calculated particle radius is 200 A [173, 175]. The calculated size
of the paracrystalline defects is the same in both the unreduced and in the
reduced states [173].

2.5.3 Surface Area

Physisorption measurements using N z , CO, Ar, O 2 or COz give the same area
[177J; the areas determined by this method are 0.44-10.4 m 2 jg [17S]. The BET
area for the reduced catalyst depends on the composition and structure prior to
reduction and on the conditions during the reduction. Consequently very
different values have been reported: 8 m 2 jg [l64J, 11.6 m 2 jg [179J, 15.S m 2 jg
[179J, 15 mZjg [93J, or 20.9 m 2jg [ISO]. After passivation an area of 13.1 m 2jg
found [172].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 33

For an industrial catalyst the calculated area of the Fe crystallites based on

the particle size obtained from X-ray powder diffraction is 2.6 times larger than
the measured BET area of the sample [164].
The surface concentration of Fe in the reduced catalyst as determined by
electron spectroscopy is 4% [162],6.4% [82], or 15% [52].
The BET area of the catalyst increases with increasing Al content [48]. The
increase is most pronounced at small Al concentrations. This has been at-
tributed to the limited solubility of FeAlz0 4 in Fe [175].
An increase of the wustite content in the unreduced sample has been
reported to increase the BET area [181], to have no effect on the BET area [182]
or to decrease the BET area [183, 184] of the reduced sample. Analogously an
increase of the wustite content in the unreduced sample has been reported to
increase [102, 180, 182, 184-187] the activity, to have no effect on the activity
[180] or to decrease the activity [101,102,181,188] of the reduced catalyst. The
existence of a maximum in catalyst activity with respect to wustite content has
been reported [188-190].
For (Fe,AI) catalyst models the BET area after reduction is 5.57 [191] or
5.86 m2jg(Fe,AI) [191]. For unpromoted Fe the BET area after reduction is
approx 1 m 2 jg [192].
The stability of an industrial catalyst at temperatures higher than normal
has been studied. High temperatures causes a decrease in the BET area
[113, 193] and an increase in the average pore diameter [113]. The BET area
and the CO chemisorption area remain proportional [193]. A phas'e rich in K,
Al and Ca [28] seems to be formed.
The stability toward impure synthesis gas has been studied. Under exposure
to'water at synthesis temperatures the BET area [194] and the CO area [194]
decreases.

2.5.4 Effect of Promoters

The cause of the structural promotion has been assigned to the segregation of
refractory oxides to the surface [53, 154, 195, 196] or to the formation of
paracrystalline defects [154, 166]. Theoretical considerations show that the
structural promoters must be located near or on the surface to have any effect
[197].

2.5.4.1 Aluminium
Al is a structural promoter [67,68, 198-204]. The distribution of Al in the
sample changes during reduction [40] and, in particular, the segregation of Al to
the surface has been the subject of a large number of studies.
From studies by chemisorption measurements [163, 198, 199,205-207], by
X-ray powder diffraction studies [164], by electron microscopy [92, 170, 171],
34 P. Stoltze

by X-ray photoelectron spectroscopy [205, 206, 208], and by scanning Auger

electron spectroscopy [52,209-211] it was concluded that Al z0 3 is present as
a thin surface layer in the reduced catalyst. Others have concluded that Al is
segregated as both FeOAl z0 3 and Al z0 3 [53]. Al has been shown by Auger
electron spectroscopy [162] and X-ray photoelectron spectroscopy [52] to
remain in the oxidic AI(III)-state.
The content of Al in the catalyst is sufficient for a monolayer coverage [53]
on the BET area. From chemisorption measurements for a (Fe,AI) sample it was
concluded that 1% Al 2 0 3 in the precursor leads to 35% total surface coverage
[198,199], while 10% Al 2 0 3 in the precursor leads to 55% total surface cover-
age [198, 199].
For concentrations of Al in excess of the optimum, Al may be observed in
bulk phases such as Al z0 3 [167] by X-ray powder diffraction. Evidence for the
existence of paracrystalline defects in reduced (Fe,AI) samples [212] has been
found.
While an increase in Al content causes a monotonic increase in total area,
the active area has been found to increase [198, 199] or decrease [67,68]. The
ratio of the catalytic activity to the BET area decreases [202,203,213] with
increasing Al concentration. For this reason no simple relation exists between Al
content and activity [65].
For (Fe,AI) catalysts the catalytic activity is maximum at 3.5-4 % Al z0 3
[201],3-4 % Al z0 3 [214],2.5-5 % Al z0 3 [48], or 3 % Al 2 0 3 [67,68]. It has
been suggested that the increase in activity with the Fe (II) : Fe(III) ratio is
caused by Al being more soluble in magnetite than in hematite [57,215].
A maximum in BET area has been found by the addition of 2.5 % Al z0 3
[67,68] or 5-6 % KAIO z [204]. The addition of Al to multiply promoted
samples decreases the pore radius [216]. Al increases the work function [217]
for (Fe,AI,K) samples. These changes may be caused by changes in the
K-coverage following the changes in Al content [217].

2.5.4.2 Calcium
Ca is a structural promoter [198,199,201]. The effect ofCa on the properties of
the catalyst has been much less studied than the effect of AI.
Ca has been shown by Auger electron spectroscopy [162] and X-ray photo-
electron spectroscopy [52] to remain in the oxidized state. Chemisorption
measurements [163] and X-ray powder diffraction studies [164] show that CaO
is segregated to the space between the Fe crystallites during the reduction. The
segregation of Ca to the surface has been demonstrated by scanning Auger
electron spectroscopy[52, 209-211].
Ca has been reported to increase the activity [218], to increase [198, 199] the
surface area, and to increase the resistance toward impurities in the gas [67,68].
The optimum Ca concentration in a (Fe,AI,K,Ca)-oxide catalyst is 2-2.5 %
[201].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 35

2.5.4.3 Potassium
K is an electronic promoter. K acts as an promoter both by impregnation and
by addition to the melt [219].
The segregation ofK to the surface has been demonstrated by chemisorption
measurements [207, 220J, by scanning Auger electron spectroscopy
[52, 209-211J, by X-ray photoelectron spectroscopy [208J, and by electron
microscopy [92]. Single crystal studies of K overlayers on Fe(110) demonstrate
that K is not a structural promoter [221J and that K may even reduce the ability
of Al to disperse Fe [221].
The migration of K to the surface of the reduced catalyst [40J has been
demonstrated by energy dispersive X-ray analysis [28J, by field iron mass
spectroscopy [222, 223J, by chemisorption of CO, CO 2, N 2and H2 [207, 220J
by scanning Auger electron spectroscopy [209-211J, and by high-voltage elec-
tron microscopy [224].
For the catalyst it has been concluded that KH, KNH2 and K 20 are less
stable than KOH under NH3 synthesis conditions [225,226]. These consider-
ations were based on the bulk phase thermodynamics.
For K adsorbed on an Al 20 3 overlayer on Fe(100), the desorption tempera-
ture is increased to 600°C [227J, well above the typical reaction temperature
[227].
On clean single crystal surfaces X-ray photoelectron spectroscopy and
ultraviolet photoelectron spectroscopy show that K is present as the metal
[228]. On Fe(100) [229, 230J chemisorbed K is disordered. On Fe(llO) a hexag-
onal close packed structure is formed [228]. On Fe(1l1) K is disordered [230J
or forms a (3 x 3) structure [231]. At room temperature it is unlikely that
multilayer adsorption will occur [228J due to the low heat of vaporization for K.
K chemisorbed on Fe(100) desorbs under NH3 synthesis at 20 atm. [232].
This is consistent with typical temperatures for NH3 synthesis being above the
desorption temperature for K. K*jFe(lOO) is stabilized by the presence of 0*
[232].
The presence of a K or a K + Al overlayer does not cause a recrystallization
or an increase in catalytic activity after steaming and reduction in the NH3
synthesis gas [233].
The migration of K on FejAl 20 3 has been studied by Auger electron
spectroscopy [234]. The migration is faster in H2 than in O 2, and faster in moist
gas than in dry. The kinetics of surface migration was found to be consistent
with a surface diffusion mechanism [234].
Addition of small amount of K results in an increase in average pore
diameter [67, 68, 235J, no changes in the pore volume [67, 68J, an decrease in
the BET area [67,68,176, 236-241J a decrease of the active area [67, 68J, an
increase in average particle diameter [176J, a decrease [176,217,237, 242-244J
or an increase [243J in the work function. For single crystal surfaces at small
K-coverages, the work function decreases with increasing K coverage
[228,229,245].
36 P. Stoltze

The importance of K for the kinetics of NH3 synthesis is discussed further in

7.1.2.

2.5.4.4 Other Promoters

Besides Al and Ca a number of elements have the ability to act as structural
promoters, either alone or as a part of a mixture.
The following elements have been reported as structural promoters:
Ba[246], Be [198,199], Ce [200, 247, 248-250], Dy [200, 248-252], Er [252],
Eu [250], mixtures of rare earths [132, 145,253-256], Ho [252],
Mg[56, 176, 198, 199,202,257-259], Mo [31, 76, 77,141, 144,260-270],
Sc [262,271,272], Si [67, 68], Sm [147, 148,273], U [274, 275], V [264],
W [77,141,261,263,265,270,276], Y [277-280], Yb[277, 278,280], and
Zr[202,239,260,278,281].
Cr has been found to be structural promoter [198,199,260,282,283], while
others have found that Cr has no effect on the catalytic activity [284] or acts to
reduce the catalytic activity [285]. The large reactivity of Cr toward N2 [286]
and the considerable stability of nitrides of Cr [287] are complications in the
deduction of the effect of small amounts of Cr added to the catalyst. Undoubted-
ly, this complication is the cause of much confusion in the study of the effects of
trace amounts of a number of transition metals on the activity of ammonia
synthesis catalysts.
The BET area increases through the sequence (Fe,AI), (Fe, Ti), (Fe, Cr),
(Fe,Mg), (Fe,Mn), (Fe,Ca), (Fe,Si), (Fe,Be) for a small promoter content [198,
199]. The BET area increases with the promoter content [198, 199].
The alkali metals are electronic promoters, increasing the activity
[139,240, 288, 289], and decreasing the area [240]. The rate of synthesis in-
creases through the sequence (Fe,AI,M), where M = Li, Na, K, Rb, Cs
[139,288,290]. A decrease in the work function generally results in an increase
in catalytic activity [291,217,242].

2.5.5 Models of Structural Promotion

Single crystals of Fe with evaporated overlayers of K and/or Al have been

studied as models for the structure of the industrial catalyst. After heating of an
Fe(11O) crystal with an Al overlayer, the surface looks rough under the electron
microscope [227]. The reaction leading to a dispersion of Al and Fe is primarily
a reaction between oxidic phases [221]. The activity increases after reaction
between the crystal and the Al-overlayer [227], possibly due to the formation of
facets [227]. After heating of an Fe(110) crystal in 5 atm of N 2 to 450 C, no
increase is found in the activity of the crystal [227]. The presence of Ai stabilizes
the Fe(111) and Fe(211) surfaces during NH3 synthesis at 20 atm [221]. The
heating of Fe (1 10) and Fe(lOO) with an Al overlayer causes a recrystallization of
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 37

the Fe during heating in steam followed by reduction in NH3 synthesis gas

[233]. After reduction the surface is as active as F(111). The increase in activity
is assigned to the formation of high index planes. For Fe(1l1) with an Al
overlayer, the resulting increase in activity is very small [233].
Fe evaporated on a nonporous Al 20 3 film has been studied as a model for
structural promotion [292-296]. After heating in O 2, transmission electron
microscopy shows a multilayer film of Fe-oxide coexisting with 3-dimensional
crystallites of Fe [294]. After reduction of oxidic Fe/Al 20 3 in H2 at 500°C with
traces (1 ppm) of H 20 or O 2, torus-shaped crystallites are formed [295]. The
composition of the tori approach the composition FeAl 20 4 and/or A12Fe206'
The core consists of less oxidic material [295]. After heating to higher tempera-
tures the shapes are less distinct [295]. After prolonged heating the crystallites
fragment, possibly due to mechanical fatigue [295]. Fe in oxidic form is present
even after prolonged reduction at high temperatures [296]. The stabilization of
oxidic Fe-compounds implies that the (Fe, AI) interfacila energy is small [296].
After heating in steam to 700°C, no Fe particles are found by transmission
electron microscopy [293], presumably because the Fe spreads as a thin film
[293]. After reduction small Fe crystallites are found [293, 295]. The crystallites
are smaller than the crystallites present before oxidation and reduction
[293, 295]. The crystallites formed after oxidation and reduction are only
smaller than the original crystallites if the reduction is of short duration
[293, 295]. Sintering of the particles is caused by coalescence and by Ostwald
ripening [296].
A large number of other catalyst models have been studied. The mostly
frequently studied models are Fe/C [297-316]. Fe/AI 20 3 [94,108,
137,317-321], Fe/MgO [108,167,319,320,322-330], Fe/SiO
[319-321,331,339] Fe/Ti0 2 [340,342], FeTi intermetallics [343,349] and
FeZr intermetallics [350, 354].

2.5.6 Alloying with Transition Metals

A number of dilute alloys of transition metals in Fe are active for ammonia

synthesis. The elements reported to form Fe alloys active for ammonia synthesis
are Co [60, 123, 194,355-363], Cu [338,364], Ir [365], Ni
[216,268,269,356,357,363,366-371], Os [372], Re [152,264,373], and Ru
[264,336,363,372,374].
The information on the effect of alloy formation on the catalytic activity is
generally conflicting. The effect of Sc during superheating of the catalyst is
interpreted as an activity increase due to reduction of SC203 to Sc followed by
an activity decrease due to Sc + Fe alloy formation [271]. Small amounts of Co
increases the rate ofNH 3 synthesis and Nrchemisorption [123]. Small amounts
ofNi increases only the rate ofNH 3 synthesis, not the rate ofN 2 chemisorption
[123].
38 P. Stoltze

A promising technique for the study of the catalytic activity of alloys is the
study of chemisorption on iron overlayers on single crystals of other metals such
as FejRu [375], FejRe [376], and FejW [377].

2.6 Chemisorptive Properties of the Catalyst

An important step in heterogeneous catalysis is the adsorption and desorption

of reactants and products of the reaction. Important information on the mecha-
nism of ammonia synthesis has come from the study of the adsorption ofH 2, N2
and NH 3. The adsorption of H 20 and O 2 is interesting because of the role of
H 20 as a poison for NH3 synthesis.
As mentioned above, not all of the surface of the catalyst is active. The
adsorption of other gases, such as CO and CO 2, is interesting because of these
gases adsorb selectively on the catalyst surface.
The presence of adsorbed atoms on a metal surface may have consequences
for subsequent adsorption of another gas. There are several possible outcomes
of this procedure. The preadsorbed atoms may weaken or prevent the sub-
sequent adsorption; the strength of adsorption for both species may increase;
a compound may be formed or the preadsorbed atoms may be displaced to the
bulk or to the gas phase.
While the reactions of the catalyst may be fairly complicated, studies of
chemisorption on single crystals has resulted in a detailed understanding of the
more important adsorption reactions.

2.6.1 Chemisorption of H

H2 is a reactant in several reactions of the catalyst surface. The role of H2 in the

reduction of the surface has been treated above. The role of H2 as a reactant in
the synthesis of NH3 will be treated below. The present section will treat the
adsorption and desorption of H 2, the ortho-para conversion, and the H2 + D2
isotopic exchange.

2.6.1.1 Structure of Chemisorbed H

On single crystals a c(2 x 2) low-energy electron diffraction pattern is observed
for HjFe(11 0) at 140 K, 7 L [378] and a p(1 x 1) low-energy electron diffraction
pattern is observed for HjFe(1 10) at 140 K, 3500 L [378]. Disorder is observed
for HjFe(1 00), HjFe(ll 0) and HjFe(111) [378] at higher temperatures. The
complete H2 + D2 isotopic scrambling upon adsorption-desorption [378] indi-
cates that H2 is adsorbed as atoms. For coverages higher than the c(2 x 2)
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 39

struCture, adsorbate-adsorbate interactions are reflected in the low-energy elec-

tron diffraction structures observed at coverages between Fe: H = 1: 0.25 and
1: 1 [378].
High-resolution electron-energy-Ioss spectroscopy spectra of c(2 x 2)
H/Fe(110) [191] are interpreted from selection rules, isotope effect, and wave
number as H* in the short bridge position [191].
By secondary ion mass spectroscopy from H2 adsorbed on Fe the signal for
H2 is stronger than for H [379]. This is taken as evidence that H2 does not
dissociate [379]. However, this argument cannot be correct in view of the large
difference in stability between H-atoms and Hrmolecules in the gas phase.
For the chemisorption of hydrogen on the catalyst, Emmett et al found
a complex behavior [163, 380, 381]. Adsorption of H2 on the catalyst was
detected at - 90 DC and above + 100 DC [381]. Transients in the adsorption
when the temperature is suddenly changed in the range 0-210 °C has also been
observed by others [382]. Presumably this behavior is caused by two reactions
where the low temperature reaction is a weakly exothermic equilibrium adsorp-
tion and the high temperature reaction is a reaction limited by a high activation
energy.

2.6.1.2 Thermodynamics
The initial enthalpy of chemisorption has been determined for single crystal
surfaces of Fe. For H/Fe(1 00) - 86 kJ/mole was found by temperature pro-
grammed desorption [383]; for H/Fe(11 0) - 109 kJ/mole [378] by tempera-
ture programmed desorption or - 24.2 kcaljmole [384] by He-scattering; for
HjFe(111) - 88 kJ/mole by temperature programmed desorption [378]. For
D/Fe(110) - 24.2 kcaljmole has been determined from He-scattering [384]
and for D/Fe(111) - 104 kJ/mole by calorimetry [385].
Adsorbate-adsorbate repulsion at high coverage is reflected in the low-
energy electron diffractions patterns, in the occurrence of temperature pro-
grammed desorption peaks, and in the enthalpy of chemisorption.
For hydrogen adsorption on polycrystalline Fe, the enthalpy of chemisorp-
tion has been determined as - 98 kJ/mole by calorimetry [385], - 96 kJ/mole
by calorimetry [386], - 81.0 kJ/mole by electrical conductivity [387],
- 85.0 kJ/mole by volumetric chemisorption [388], - 36 kcaljmole at zero
coverage by volumetric chemisorption [389], - 15 kcaljmole at 90% saturation
[389], - 20 kcaljmole at 140 K, zero coverage [390], - 17.5 kcaljmole at 1%
of saturation calculated from the equilibrium pressure [391], and
- 5 kcaljmole at 10% of saturation calculated from the equilibrium pressure
[391].
For Fe/MgO the saturation coverage depends on particle diameter [328],
i.e., the chemisorption appears to be structure sensitive. Structural sensitivity of
NH3 synthesis will be discussed further in Sect. 7.1.1.
The studies of the chemisorption of H2 on the catalyst has been comp-
lemented by studies at low [391, 392] and at high pressures [393].
40 P. Stoltze

Calorimetry on a (Fe, AI, Ca, K) sample at 35°C yields a Freundlich

isotherm [394]. The decrease in heat of chemisorption is the same for promoted
and unpromoted samples [391]. By volumetric chemisorption the adsorbed
phase was found to behave as a 2 dimensional gas with a constant enthalpy of
chemisorption, - 85.0 kl/mole, up to 3.3 I1mole/m2 [388].
For an industrial catalyst the enthalpy of chemisorption is - 25 kcal/mole
by calorimetry [395], - 28 kcal/mole at zero coverage from calorimetry [394],
and - 12.3 kcal/mole calculated from calorimetry of NH3 cracking [395].
From a thermodynamic model [396] of the chemisorption of H2 we have
found that under an equilibrium pressure of 1 atmosphere, the coverage by H*
decreases from ~ 1 at 500 K to ~ 0.15 at 1000 K. By calculating the outcome of
a volumetric chemisorption experiment from a model of the kinetics of ammonia
synthesis [396], it is concluded that H* is by far too weakly adsorbed to be used
in a titration of the number of active sites on the catalyst surface.

2.6.1.3 Adsorption Kinetics

For single crystal surfaces the sticking coefficient for H2 into 2H* is 0.03 on
Fe(1 00) at 250 K [383],0.16 on Fe(11 0) at 140 K [378], and unity on Fe (pc) at
78-298 K [397]. The activation energy for Fe(110) is zero [384] or
0.70 kcal/mole [398].
The presence of H-H interactions [399] on the surface results in the
formation of ordered domains at low temperatures. The kinetics of domain
growth for H/Fe(11 0) has been studied by Monte Carlo simulations [400]. At
e = 0.500 the domains grow according to an Allen-Cahn power law; the size of
the domains is proportional to to. 50 [400]. At e = 0.667 the rate is lower due to
diffusion limitations [400]. The diffusion and reactions of H* at the surface has
a number of similarities to the diffusion of H in the bulk [401].
For a film of Fe the rate of adsorption for H2 at low temperature is
proportional to the square root of the hydrogen pressure [402] and decreases
exponentially with coverage [402]. The activation energy is 3-6 kcal/mole and
increases with coverage [402]. The H2 chemisorption is activated for small Fe
particles on MgO [328], and for (Fe, Ir)/AI 20 3 [365].

2.6.1.4 Desorption Kinetics

The temperature programmed desorption peak maxima at low coverage are
H/Fe(100) 400 K [378], H/Fe(11 0) 480 K [378], H/Fe(lll) 370 K [378]
H/Fe(pc) 430 K [386]. For both H2 and D2 the desorption shows second-order
kinetics [398, 384].
Additional peaks caused by adsorbate-adsorbate repulsion are seen for H2
desorbing from single crystal surfaces at high initial coverages [378, 386].
The desorption temperature is 150°C [403] or 120-170, 280-380 and
480-540°C [404] for (Fe, AI) samples; 120-170,280-380 and 480-540 °C for (Fe,
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 41

AI, Cal samples [404J; and 120-170,280-380 and 480-540°C for (Fe, AI, Ca, K)
samples [404].
Temperature programmed desorption of H2 from the reduced catalyst has
been reported to proceed at 90-240°C [405J or to show 4 peaks [406]. The
kinetics of thermal desorption of H2 from the catalyst is second order [407].
Evidence that chemisorbed hydrogen remains on the surface after evacuation at
550°C for 24 h has been reported [408].

2.6.1.5 Properties of Chemisorbed H

For H* on a Fe surface the ultra violet photoelectron spectroscopy (Is) peak is
found at - 5.6 eV [378J.
High-resolution electron-energy-Ioss spectroscopy spectra of c(2 x 2)
H/Fe(110) [191J are interpreted from selection rules, isotope effect, and wave
number as H* in the short bridge position [191J. The 2 bands are interpreted as
the symmetric Fe-H-Fe stretch at 1060cm- 1 and the asymmetric Fe-H-Fe
stretch at 880 cm -1 [191].
Chemisorbed H decreases the work function for industrial catalysts [243,
279]. The effect of hydrogen on the work function has been reported to depend
on the pressure [409]. There is evidence that the measured values of the work
function change is distributed by the lack of mobility at 140 K [378]. Whereas if
the temperature is increased, significant desorption takes place before equi-
li-brium is established [378].
Magnetic measurements of hydrogen chemisorption on Fe/Si0 2 are re-
versible at 210 torr, 310°C [410]. For 15 A Fe/MgO, H* affects the magnetic
moment below the superparamagnetic transition. Above this temperature no
effect is found [326J for 80 A Fe/MgO, H* does not affect the magnetic moment
[326].

2.6.1.6 Effect of Promoters

For Fe single crystals the saturation coverages by H* is not effected by the
presence of K [411J, whereas for (Fe, AI) samples the amount of hydrogen
chemisorption is increased by K-promotion [391, 412J. For Fe/AhO, tem-
pera-ture programmed desorption shows that K increases the strength of
adsorption for H [412].
At low coverages by K, preadsorbed K on Fe(1 00) increases the sticking
coefficient for hydrogen [383]. The increase indicates that the sticking co-
effi-cient is unity for K-promoted sites [383].
The heat of adsorption is increased by 6-10 kJ/mole for K promoted sites on
Fe(1 00) [383, 411J and 8 kJ/mole for K-promoted sites on Fe(lll) [411J. Even
at low coverages by K, the temperature programmed desorption spectra of
hydrogen desorbing from H/K/Fe(1 00) or from H/K/Fe(111) are not split into
peaks assignable to promoted or unpromoted sites respectively [411].
42 P. Stoltze

The hydrogen chemisorption becomes activated when Al is present [193,

321]. A plausible explanation for this observation is that the structural pro-
moters decorates the surface [321].

2.6.1.7 Effect of Preadsorbed Species

Preadsorbed Oz inhibits Hz chemisorption [207,413] and decreases the rate of
adsorption [383]. On p(1 x 1) 0/Fe(100) the initial sticking coefficient for
hydrogen is 1.0· 10 -4 at 200 K [383].
The reaction between H2 and preadsorbed 0* is an important model for the
reduction of the catalyst. For Fe (1 00) this reaction has been studied at 473 and
673 K at 2· 10 - z torr of Hz [134]. The apparent activation energy is
59 ± 4 kJ/mole [134]. The reaction proceeds via the dissociative chemisorption
of Hz [134]. For this reason it is not unexpected that the reaction becomes quite
inhibited by the presence of a monolayer of 0* [134].
Preadsorbed S decreases the rate of adsorption. No H* is adsorbed on
c(2 x 2) S/Fe(1 00) after 2000 L exposure at 200 K [383].
Preadsorbed C decreases the rate of adsorption. On c(2 x 2) C/Fe(1 00) the
initial sticking coefficient for hydrogen is 10 - 3 at 200 K [383].
Preadsorbed CO inhibits Hz chemisorption [207, 406], but has no effect on
the enthalpy of chemisorption for Hz [389].
Preadsorbed COz inhibits Hz adsorption at - 78°C [207].
F or catalysts the results obtained when N z is adsorbed on preadsorbed H *
and vice versa are not readily interpreted [380]. For single crystals the results
are easier to interpret. When preadsorbed H * is exposed to N z (g) at 77 K, N z (g)
is displaced [386,414] and physisorbed N z may be formed on top of the H*
layer [386].
By coadsorption of N2 and Hz, FeN z is seen in secondary ion mass
spectroscopy [379]. An Eley-Rideal mechanism has been suggested,
(1 )
[379]. This conclusion is in disagreement with more direct studies of the
chemisorption of N zH z (Sect. 2.6.9) and of the mechanism of ammonia synthesis
(Sect. 2.7).
Preadsorbed N * does not inhibit dissociative adsorption of HzO [415].
Preadsorbed N * has been reported to decrease [207, 416, 406, 414] or to
increase [417] the amount of chemisorbed hydrogen. Preadsorbed N * has been
found to decrease [389], to increase [394] or to have no effect [414] on the heat
of adsorption for Hz [389]. Presumably, the results of the coadsorption of N *
and H * depends sensitively on the experimental conditions as both displace-
ment to the gas phase and the formation of NH3 are realistic possibilities.
The presence of 0.1 % N z in the reduction gas used for chemisorption
measurements decreases the Hz chemisorption by 25-30% [193, 380].
Preadsorbed N * is removed as NH3 upon exposure to hydrogen [418,419].
The synthesis of ammonia by hydrogenation of chemisorbed N * is faster
than by the reaction of Hz and N z over the same catalyst [420, 421]. This
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 43

suggests that N z and N * are not in equilibrium under ammonia synthesis

conditions. This conclusion is in agreement with more direct studies of the
mechanism of ammonia synthesis (Sect. 7).
15N isotopic labeling shows that all chemisorbed N-atoms undergo hydro-
genation with equal probability [422]. This suggest that all adsorbed N * atoms
have identical properties, i.e., the hypothesis that the surface should be strongly
heterogeneous (Sect. 2.7.6) may be incorrect.
The rate ofN * removal from bulk Fe by reaction with Hz is first order in N *
[423].

2.6.1.8 The Hydrogen ortho-para Conversion

The nuclear spin of H results in the existence of H2 in two states, a singlet state
called ortho and a triplet state called para. The interconversion of these two
forms is a peculiar reaction as it requires the interaction with a magnetic
substance but does not require the fission of the H-H bond.
The hydrogen ortho-para conversion is fast on (Fe, AI, K) and (Fe, AI, Si, Zr)
samples at - 195°C [424]. The conversion over (Fe, AI, K) and (Fe, AI) is
:2:: first order in H2 [425].
The ortho-para conversion is not poisoned by preadsorbed H* on an (Fe, AI,
Si, Zr) sample [424] but has been found to be poisoned by pre adsorbed H on an
(Fe, AI, K) sample [424, 405].

2.6.1.9 H2 + D2 Exchange
The H2 + D2 isotopic exchange is interesting as one of the simplest chemical
reactions of H2 on the catalyst surface.
The activity of H2 + D z exchange increases as Fe(pc) > Fe(ll 0) >
Fe(l 00) > Fe(lll) [426].
Fast Hz + D2 isotopic scrambling has been found for an (Fe, AI, Si, Zr)
sample at - 195°C [424], while the scrambling was slow on an (Fe, AI, K)
sample at the same conditions [424].

2.6.2 Chemisorption of CO

The chemisorption of CO is interesting for the study of ammonia synthesis

catalysts since this reaction provides a way of titrating the number of active sites
on the surface. The reaction is complicated by the dissociation of CO.

2.6.2.1 Structure of Chemisorbed co

The CO chemisorption is molecular at low temperature [207]. At higher
temperatures the CO is dissociated. Evidence for the existence of more than one
44 P. Stoltze

molecular species at low temperature has been found from X-ray photoelectron
spectroscopy and ultra violet photoelectron spectroscopy [245, 427].
On single crystals a c(2 x 2) structure of CO/Fe (1 00) is formed at 373-400 K
[427,383]. For CO/Fe(11 0) at 300 K, c(2 x 4) is formed at low coverage [428,
429], p(l x 2) is formed at high coverages [428]. By NEXAFS it has been found
that for CO/Fe(100) the CO molecule is tilted 45 ± 10° [430].
The disorder observed below room temperature [383, 427] [429] is caused
by the lack of mobility [427]. Just below the dissociation temperature ordering
is observed.
While molecular and dissociated CO have similar X-ray photoelectron
spectroscopy spectra, O(ls) at 531 eV and C(ls) at 285 eV [245,427,431], the
ultra violet photoelectron spectroscopy [245, 427] and laser Raman [432]
spectra are different.
During sequential adsorption of 12CO and 13CO on Fe (1 00), isotopic
exchange is observed only among the two strongest bound molecular states
[433]. For the catalyst, isotopic scrambling between 13C160 and 12C180 is
observed at - 33°C indicating the existence of dissociated CO at these tempera-
tures. There is partial isotropic scrambling between the first and second exposure
at -195°C [434] and at -78°C [435].

2.6.2.2 Thermodynamics
For single crystal surfaces the initial enthalpy of chemisorption for CO
is -105 kJ/mole on Fe(100) by temperature programmed desorption
[383], - 96 kJ/mole on Fe(ll 0) by temperature programmed desorption [245,
436], - 91 kJ/mole by temperature programmed desorption on Fe(lll) [437],
and -155 kJ/mole by calorimetry on Fe(pc) [438].
The enthalpy of chemisorption for CO on Fe surfaces is - 9.3 to
- 4.2 kcaljmole at O°C [439] on (Fe, AI), - 23.1 to - 16.9 kcaljmole at O°C
[439] on (Fe, AI, K), - 8.0 to - 4.2 kJ/mole at - 183°C [439] on (Fe, AI, K), and
- 32 kcaljmole at 22°C [389] on (Fe, AI, K). For Fe supported on Alz0 3
a Freundlich isotherm has been found [160]. The enthalpy of adsorption is
12-30 kJ/mole at 37% of saturation [160].
For Fe/MgO the amount of CO chemisorption depends on particle diameter
[328]. The ratio of the area determined from CO chemisorption to the area
determined by the BET method decreases through the sequence Mg > Be > AI,
Si, Cr > Mn > Ca > Ti [198, 199]. Infra-red spectroscopy and microcalorimetry
show that the amount of weakly bound CO increases with the dispersion of
Fe/MgO [328].
For multiply promoted samples the CO area is 5.6 m 2/g [180].
For the catalyst, pulse chemisorption and volumetric chemisorption give
identical results at - 78 °C and - 196°C [440]. For an industrial catalyst the CO
chemisorption was initially found to be 0.7-1.4 m 2/g [14,164], i.e., 9-18 flmole/g;
in later studies the value 28-39 flmole/g was found [179]. Increa-sing the wustite
content in the unreduced sample was reported to increase the CO area [181] or to
decrease the CO area [183] of the reduced catalyst.
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 45

2.6.2.3 Adsorption Kinetics

The sticking coefficient is close to unity and remains constant up to 60% of
a monolayer on Fe(l 00) [383] and up to 40% of a monolayer on Fe(111) [437]
indicating the existence of a mobile precursor for the adsorption. The sticking
coefficient decreases with temperature. For the catalyst the adsorption kinetics is
first order [402].

2.6.2.4 Desorption Kinetics

At low initial coverage, temperature programmed desorption of CO from CO/Fe
show peaks from adsorbed molecules at 430 K for Fe(100) [383], 400 K for
Fe(110) [245, 436], 300 K [437] or 360 K [441] for Fe(lll), and 350 K for
Fe(pc) [438]. At low initial coverages, temperature programmed desorption of
CO from dissociated CO on Fe show peaks at 800 K [383] for Fe(l 00), 710 K
[441] for Fe(lll) and 700 K [438] for Fe (pc). By sputtering of CO/Fe(111) the
major desorption product is CO [442].
Isotopic labeling shows that physisorbed CO is desorbed in vacuum at
-195 C while chemisorbed CO is not [434]). Temperature programmed desorp-
tion of CO from the reduced catalyst yields 2 peaks, one peak at - 50 to
+ 180°C, another peak at 650 °c[406]. Presumably, the peak at 650°C is caused
by the recombination of C * and 0 *.

2.6.2.5 Dissociation Kinetics

Upon heating, the dissociation of adsorbed CO is complete at 390 K on Fe(11 0)
[245]. The activation energy for dissociation of adsorbed molecules is
105 kJ/mole [383].

2.6.2.6 Properties of Chemisorbed co

For single crystal surfaces the ultra violet photoelectron spectroscopy CO(40")
peak is - 10.6 eV and the ultra violet photoelectron spectroscopy CO(n + 50)
peak is - 7.6 eV for CO/Fe(11 0) [245], while the ultra violet photoelectron
spectroscopy C(2p) and the ultra violet photoelectron spectroscopy 0(2p) peak
are both - 5.6 eV for C* + 0* on Fe(ll 0) [245].
For CO/Fe (1 00) at exposures below 1 L at 350 K the CO stretch frequency is
observed by high-resolution electron-energy loss-spectroscopy at
1180-1245 cm -1 [443]. At higher exposures at 110 K the CO stretch frequency is
observed at 1900-2055 cm -1 [443]. This change in frequency is attributed to the
transition from a side-on bonding at low coverages at a end-on bonding at higher
coverages [443].
High-resolution electron-energy-loss spectroscopy spectra of CO/Fe(11 0)
indicate that CO is positioned on-top with the C-O stretching frequency at
1890-1950cm- 1 [428]. The vibration frequency for the C-O stretch for
46 P. Stoltze

CO/Fe(110) is increased upon increasing exposure to CO [428] indicating

adsorbate-adsorbate interactions.
On Fe(111) 3 different coordinations have been observed [437]. High-resolu-
tion electron-energy-Ioss spectroscopy spectra of C + 0/Fe(111) show the Fe-C
stretch at 420 cm - 1 and the Fe-O stretch at 540 cm - 1 [437].
At 100 K the work function increases for CO/Fe (1 00) until c(2 x 4) is
complete at 3.8 L [436] and then remains constant due to depolarisation until
p(l x 2) is complete at 4.5 L [436]. Above room temperature the work function
depends on pressure [436], presumably because adsorption is followed by
dissociation. At small exposures the work function increases with CO coverage
[428, 436, 438].

2.6.2.7 Effect of Promoters

A high-resolution electron-energy-Ioss spectroscopy study of CO/K/Fe(111)
shows that up to one half monolayer of K, on-top, shallow-hollow and deep-
hollow CO* are seen [437]. At higher K-coverages, on-top CO* is observed in
high-resolution electron-energy-Ioss spectroscopy together with bands inter-
preted as CO* near K (1360-1435 cm -1) and CO-bending modes (820 cm -1)
[437].
Temperature programmed desorption shows that the presence of K in-
creases the heat of adsorption for CO [230, 444]. An increase of the heat of
adsorption by 40-80 kJ/mole [230] has been deduced from the change in peak
temperature.
The presence of K was found to increase the saturation coverage for CO
. [230, 245, 444]. For Fe(11 0) the saturation coverage is increased from
7.3 flmole/m2 for the unpromoted surface to 11.7 flmole/m2 for a surface
precovered by 6.3 flmole of K per m 2 [245].
Other studies have found that K has no effect on that saturation coverage for
CO on Fe (1 1 1) [437]. A possible explanation for the observed effects is that the
adsorbate-adsorbate interactions limiting the CO adsorption on unpromoted
Fe are screened by K. This will tend to increase the saturation coverage in the
presence of K, while the blocking of sites by K will tend to decrease the
saturation coverage by CO.
The initial sticking coefficient has been reported to increase [245], to remain
constant [230] or to decrease [230] in the presence of preadsorbed K. Presum-
ably conflicting results are obtained because the K blocks sites while increasing
the strength of CO adsorption.
For single crystal surfaces the temperature programmed desorption peak
maximum is 390 K for CO/K/Fe(1 00) [383],420 K for CO/K/Fe(ll 0) [245],
and 500 K for CO/K/Fe(111) [437]. Temperature programmed desorption
shows an extra peak at 700 K at small K coverages. This peak is due to the
simultaneous desorption of K and dissociated CO [230]. Desorption of disso-
ciated CO from K/Fe showed peaks at 800 K for Fe(l 00) [383], 700 and 820 K
for Fe(111) [437].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 47

The presence of adsorbed K leads to an apparently lower barrier to dissocia-

tion [444]. Presumably this is a reflection of the changed stability of the
adsorbed molecule. If the adsorbed molecule is stabilized, the rate of desorption
at a given temperature will be lower and this may appear as an increased
tendency to dissociation.
For COjKjFe(11 0) with 8.2 f1molejm2 of K, the ultra violet photoelectron
spectroscopy (40-) peak is - 11.4 eV, and the ultra violet photoelectron spectro-
scopy CO(ln + So-) peak is -7.8 eV [24S].
The X-ray photoelectron spectroscopy O(ls) peak for COjKjFe(11 0) and
respectively for C + OjKjFe(11 0) is unaffected by the presence of K [24S].
Preadsorbed K reduces the frequency of the CO stretch vibration [230].

2.6.2.8 Effect of Preadsorbed Species

Preadsorbed 0 * inhibits the chemisorption of CO [383, 207,413], decreases the
tendency for dissociation of CO * [383], and decreases the bond strength for
CO* [44S].
Preadsorbed S * partially blocks the adsorption of CO [446], decreases the
initial sticking coefficient for CO [383], and decreases the tendency for dissocia-
tion of CO * [383].
Preadsorbed C* decreases the initial sticking coefficient for CO [383] and
decreases the tendency for dissociation of CO * [383].
Preadsorbed N * partially inhibits CO chemisorption [207, 447, 448] and
decreases the heat of adsorption for CO at 32°C [389].
Preadsorbed CO 2 inhibits CO adsorption at - 78 DC [207].
Preadsorbed CI * blocks the adsorption of CO [449].

2.6.2.9 Correlation with Activity

Extensive chemisorption measurements on reduced (Fe, AI) and (Fe, AI, K)
samples have been reported by Brunauer and Emmett [163]. From these studies
it has been concluded that the Fe-area of the samples can be measured by low
temperature chemisorption of CO [163, 193].
The active fraction of total area generally decreases with increasing pro-
moter content [198, 199]. The activity is maximum when the ratio of the area
determined by CO chemisorption to the area determined by the BET method is
0.3 [67, 68].

2.6.3 Chemisorption of CO 2

The chemisorption of CO 2 on the surface of the catalyst is primarily interesting

as CO 2 appears to bind selectively to some parts of the surface of the catalyst.
48 P. Stoltze

2.6.3.1 Structure of Chemisorbed CO 2

For CO 2 adsorption on a stepped Fe(11 0) and on Fe(111) at 77 K, a linear and
an unidentified species is formed [450]. At 140 K the unidentified species is
dominating [450]. Above 140 K adsorbed CO 2 dissociates to CO*, 0* and C*
[450].
X-ray photoelectron spectroscopy and UPS for CO 2 on a Fe film shows that
CO 2 is present at 80 K as a linear molecule and as a bent CO 2 species [451].
For the catalyst, CO 2 is assumed mainly to bind to alkali at - 78 C without
dissociation [207]. Pure MgO, Cr203 and Al 2 0 3 chemisorbs some CO 2 [198,
199].

2.6.3.2 Thermodynamics
For Fe(11 0) no adsorption of CO 2 is detectable in the range 77-340 K [450].
For a (Fe, AI) catalyst model the enthalpy of chemisorption is ~ 8.7 to
- 6.5 kcaljmole at - 78°C [439] and - 17.7 to - 8.9 kcaljmole at - 78°C
[439] for (Fe, AI, K).
Pulse chemisorption and volumetric chemisorption give identical results for
the CO 2 chemisorption [440]. The enthalpy is difficult to determine because of
the tendency to dissociate at higher temperatures.

2.6.3.3 Dissociation Kinetics

Above 140 K adsorbed CO 2 dissociates to CO*, 0* and C* [450].
During passivation with CO 2 at 663-773 K, some carbide is formed in the
industrial catalyst [452--457].

2.6.3.4 Effect of Promoters

The addition of basic promoters increases the CO 2 area [198, 199]. The ratio of
the area determined by CO 2 chemisorption to the area determined by the BET
method decreases through the sequence Ca > Mn, Mg, Si > Cr, Be> Al > Ti
[198, 199]. For the industrial catalyst the CO 2 area is 59% of the BET area
[164].

2.6.3.5 Effect of Preadsorbed Species

Preadsorbed O 2 does not inhibit CO 2 chemisorption [207,413].

2.6.4 Physisorption of N2

On Fe films at 78-273 K, three forms of N exists: physical adsorption, weak

chemisorption and strong chemisorption [458, 459].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 49

2.6.4.1 Structure of Physisorbed N2

Based on detailed interpretation of the X-ray photoelectron spectroscopy spec-
trum the geometry of physisorbed N 2* is assigned to end-on [460, 461].
For the catalyst a phase transition in the physisorbed N2 is observed at 5%
of a monolayer at 79.8-90.3 K [164].

2.6.4.2 Thermodynamics
The saturation coverage is 5.8 ± 0.7 {lmole/m 2 at 91 K for Fe(111) [462,463].
The heat of physisorption is 5.2-3.1 kcal/mole [390] for Fe powder at
78-90 K.
For single crystal surfaces the heat of adsorption is 25-37 kllmole depending
on the coverages by molecular and physisorbed N2 * [22, 462].

2.6.4.3 Adsorption Kinetics

The initial sticking coefficient for N 2(g) into the physisorbed N 2 state is 0.7 for
Fe(111) at 85 K [462].

2.6.4.4 Desorption Kinetics

For Fe(l 11) a temperature programmed desorption peak is found at 84 K [22]
or at 96 K [231]. Sequential adsorption of 14N and 15N shows that this peak is
most likely a new N 2* species; possibly this peak is physisorbed N 2 upon N 2*
[22].

2.6.4.5 Kinetics of Conversion into aN 2

The activation enthalpy for the conversion of physisorbed N 2* to chemisorbed
N 2 * is 18 kllmole for Fe(ll1) [462]. Estimation of the reaction rates indicates
that the chemisorbed N 2* may be formed directly, rather than via physisorbed
N 2 *, at typical reaction conditions for NH3 synthesis [464].
Below 90 K the equilibrium between the molecular chemisorbed state and
the physisorbed state is not established [465].

2.6.4.6 Properties of Physisorbed N2

Ultra violet photoelectron spectroscopy N (1s) peaks are found at - 12 and
- 8 eV [462]. At 85 K X-ray photoelectron spectroscopy N(1s) peaks are found
at 405.9 and 401.2 eV [462, 463]. High resolution electron energy loss spectro-
scopy shows a N-N stretch in yN2 at 2100 cm -1 [231]. For N 2* the work
function decreases linearly with the coverage up to saturation [22].
50 P. Stoltze

2.6.5 Molecular Chemisorption of N z

Even if CO is isoelectronic with N z, the bonding to the surface is different for the
two molecules. CO adsorbs with a high sticking probability and high binding
energy in end-on geometry, while N z adsorbs side-on in a more weakly adsor-
bed species. The dissociation of N z * is much easier than the dissociation of
CO*.

2.6.5.1 Structure of Chemisorbed N z

A molecular, chemisorbed species, N z *, is formed during the adsorption of N z
on Fe [229,466, 468].
The absence of 14N_15N exchange during adsorption-desorption [386, 469]
and isotope labeling of N z in high-resolution electron-energy-loss spectroscopy
demonstrates that this species is molecular [231].
The geometry of N z * is assigned to side-on based on detailed interpretation
of the X-ray photoelectron spectroscopy spectrum [460,461] and of the high-
resolution electron-energy-loss spectroscopy spectrum [470].

2.6.5.2 Thermodynamics
The saturation coverage for N z * on Fe(lll) is 1.16 pmole/m z [462,463]. This
is significantly less than for the physisorbed state.
The enthalpy of chemisorption for N z * is - 31 kJ/mole for N z /Fe(1 00)
[466],[229], -31 kJ/moleforN z/Fe(110)[467],and -21 kJ/moleforFe(pc)
[469] from temperature programmed desorption studies.

2.6.5.3 Adsorption Kinetics

The sticking coefficient for N z into N z * is 10 - z for Fe (11 1) [466]. This is lower
than the sticking coefficient into the physisorbed state.

2.6.5.4 Desorption Kinetics

Temperature programmed desorption of N z from Nz*/Fe is 160 K for
Nz/Fe(l 00) [466], 160 K for N z/Fe(ll 0) [467], 260 K for N z/Fe(lll) [469,
471], and 290 K for Nz*/Fe(pc) [471]. Others have reported desorption ofN z
from N 2 */Fe(pc) at 77,100,200-250 and 350 K [386].

2.6.5.5 Properties of Chemisorbed N2

High-resolution electron-energy-Ioss spectroscopy spectra of N 2 */Fe(111)
show a N-N stretch at 1490 cm -1 [463, 470, 472] indicating a considerable
weakening of the N-N bond compared to the stretch found at 435 cm -1 [463,
472].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 51

Laser Raman during NH3 synthesis shows peaks at 2040, 1940, 423,
443 cm- 1 [473]. These peaks have been assigned to N z * [473].
The adsorption of N z * increases the work function on Fe [22, 474]. The
dipole moment of N z * is 0.4 D [467].

2.6.5.6 Effect of Promoters

K increases the stability of N z * [466]. The maximum stability is 10.5 kcaljmole
[466]. This effect is strongest on Fe(l 00) [466]; this erases the difference in
sticking coefficient between low index planes [466].
The decrease in N z * adsorption at high coverages by K is consistent with
N 2* adsorbing on Fe adjacent to K* rather than adsorption on top of K*
[466].
The enthalpy of chemisorption has been determined from temperature
programmed desorption and from equilibrium coverages. K increases the stabil-
ity of N 2* [466]. The maximum stability is 10.5 kcal/mole [466]. The enthalpy
of chemisorption is - 46 kllmole for Nz/K/Fe(l 00) [466] and - 43.9 kllmole
for Nz/K/Fe(lll) [466]; the heat of adsorption is increased by 3 kcal/mole by
the presence of K [229].
From theoretical considerations it has been argued that the effect of K + is
a local effect extending at most to the next neighbor position.
On exposing K/Fe(111) to N 2, promoted and unpromoted sites are simul-
taneously filled [466]. This indicates that the sticking coefficient into the N 2 * is
approximately identical for both promoted and unpromoted sites.
The temperature programmed desorption peak maximum for
N 2 */K/Fe(111) is 210 K [466]. At small coverages by K, both promoted and
unpromoted sites are seen in temperature programmed desorption [466]. At
large K-coverages a short range Nz*-K* interaction is detected [231]. The
temperature programmed desorption peak maximum is increased to 210 K. The
stabilization is 16.2 kllmole [231].
High-resolution e1ectron-energy-Ioss spectroscopy of N2 */K/Fe (111)
shows a N-N stretch at 1390 cm -1 [470] or 1370 cm -1 [470]. The reduction of
the N-N stretching frequency in the presence ofK [231,472] does not represent
a significant weakening of the N-N bond [475].

2.6.5.7 Effect of Preadsorbed Species

Preadsorption of 0* inhibits the formation of N2 * [470,476,477].

2.6.6 Dissociative Chemisorption of N

Above room temperature the exposure of Fe to N z leads to the formation of N *

through dissociative chemisorption, N z (g) + 2 * = 2 N *. A number of reviews of
52 P. Stoltze

ultra-high vacuum single-crystal studies of the kinetics and mechanism of N2

adsorption on Fe are available [10, 20-23,478].

2.6.6.1 Structure of Chemisorbed N

The exposure of Fe to N2 at low and moderate pressures does not lead to the
formation of bulk nitrides.
The adsorbate is atomic as 28N + 30N are completely equilibrated after
adsorption-desorption on Fe(ll 0) at 680 K [468J, on Fe(1ll) at 140-1000 K
[468J, on Fe(pc) [479, 480J and on the catalyst [481]. Complete isotopic
scrambling indicates that the adsorption is dissociative. Others have concluded
that a significant part of the adsorbed nitrogen at 380-400°C is not dissociated
[379,482, 483]. These latter conclusions are based on less direct evidence.
On single crystals, N2 adsorbing on Fe(l 00) forms c(2 x 2) [229,466,474,
484]. Detailed interpretation of the low-energy electron diffraction patterns for
NjFe(l 00) shows that N * is located in sites with C4 symmetry 0.27 Aabove the
outermost plane of Fe-atoms [485]. The distance between the first and the
second layer of Fe atoms is expanded by 7.7% compared to the bulk lattice of Fe
[485].
Fe(110) [468J and Fe(lll) [468, 474J reconstruct upon N2 adsorption.
Fe(ll 0) forms c(2 x 3) [468J, while Fe(lll) forms (3 x 3), (J19 xJ19) R23.4 0,
(j21x j21)RlO.90, (foxfo)R30°, (2x2) [474]. The c(2x2) pattern on
Fe(lll) is only observed when N segregates from the bulk [474]. Nt implanta-
tion in Fe(1ll) yields p(l x 1) [486].
The observed structures have been rationalized [468, 474]. The c(2 x 2)
NjFe(1 00) pattern is structurally very similar to the (002) plane of Fe4N while
recrystallisations of NjFe(1 10) and NjFe(lll) can be interpreted as stepwise
transitions toward the Fe4N(1Il) structure. The sequence of patterns can be
seen as an attempt to keep as many Fe-atoms fixed while accommodating an
increasing number of N-atoms.
On Fe(ll 0) the original surface structure is recreated during desorption in
N * [468]. For F e( 1 00) partial desorption of N * results in disorder [474 J;
accommodating at 620 K recreates the order [474]. N2 chemisorption on
Fe(12, 1,0) at 750 K leads to an increase in the density of steps and to facetting
[487].
The chemisorption of N 2 has been studied on catalysts and catalysts models
[488, 489]. The studies include high pressures [393J and high temperatures
[490].

2.6.6.2 Thermodynamics
From a study of N2 adsorption by volumetric chemisorption it was concluded
that the chemisorption of N2 on a (Fe, AI, K, Si) catalyst follows the Freundlich
isotherm [491, 492]. The enthalpy of chemisorption is - 38 kcaljmole 10g(fJ)
[492].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 53

For polycrystalline samples the enthalpy of chemisorption for N 2, i.e. for

N 2(g) + 2* = 2N*, has been determined in a number of ways [493]. For a (Fe,
AI) sample
H = - 208.2 + 446.58 kJ/mole (2)
for converges 0.00 < 8 < 0.22 [494], and
H = - 136 + 122.28 kJ/mole (3)
for 0.22 < 8 < 0.90 [495] has been determined from gravimetric studies.
F or an industrial catalyst, - 209 ± 16 kJImole at 400°C [493],
30 ± 5 kcaljmole at 197°C [395], and 32 kcaljmole [496] were obtained from
direct calorimetric measurement; - 70 kcaljmole at 8 = 0, and - 16 kcaljmole
at 8 = 0.18 [389] from volumetric chemisorption; - 56 to - 19 kcaljmole
[497] from temperature programmed desorption; -175 kJ/mole [498] from
the solubility of N in Fe; and - 126 ± 21 kJ/mole at 197°C [499] from
a calorimetric study of NH3 adsorption.

2.6.6.3 Adsorption Kinetics

Emmett and Brunauer pointed out that the rates of N2 chemisorption and of
NH3 synthesis are comparable and slow [163].
The sticking coefficient, defined as the probability that one N 2(g) molecule
hitting the surface will chemisorb as 2N *, is small. The sticking coefficient for N 2
determined by Auger electron spectroscopy is on Fe(l 00) 10- 7 at 383 K [474,
484],10- 7 at 400 K [229],1.4-10- 7 at 430 K [466],4.0'10- 7 at 508 K [484];
onFe(11 0) 1 -7 at 583-733 K [468]; and on Fe(111)10-7 to 10- 6 [468] or
6'10- 6 at 583-733 K [467,474]. On polycrystalline Fe, the sticking coefficient
is less than 10- 6 at 290 K [471], 5·10 -6 at 430 K [466],4'10 -6 at 423 K [500]
determined by Auger electron spectroscopy, and 1.6· 10 -7 at 273 K, [479] and
2.5' 10 - 7 at 348 K [479] determined by volumetric chemisorption.
For Fe(l 00) the activation enthalpy for the sticking coefficient is indepen-
dent of coverage [229], while for Fe(ll 0) [229,466,468,474,484] and Fe(lll)
[474] the activation enthalpy becomes more positive with increasing coverage.
The activation energy is approximately 5 kcaljmole [484] for Fe(pc). From the
low value of the activaiton energy for the sticking coefficient, it is apparent that
the small sticking coefficient is due to the unusually low value for the preex-
ponential factor.
For a molecular beam of N z directed on Fe(ll 0) crystal at 550 K, N2
translational energy 0.176 and 0.647 eV, no diffuse scattering was observed
[501]. This is not unexpected in view of the low sticking probability.
The dynamics ofN z chemisorption on Fe(111) has been studied by molecu-
lar beam techniques using kinetic energies in the range 0.3 to 0.6 eV [502, 503,
504] and observation of the sticking by electron spectroscopy. The sticking
probability increases from 10 -6 at 0.09 eV to 10 -1 at 4.3 eV [503]. The sticking
coefficient increases at low surface temperatures [503]. This demonstrates that
54 P. Stoltze

the barrier to dissociation is found in the entrance channel [502-504]. Conse-

quently this barrier is more easily passed if energy is supplied to the molecule in
the form of kinetic rather than vibrational energy. Vibrational energy is about
half as effective as kinetic energy [502-504].
The high activity on the more open planes has been assigned to the existence
of C 7 sites on such planes [505].
For (Fe, AI) samples the rate of N2 chemisorption has been studied by
gravimetry [494, 495]. The activation energy of chemisorption was found to
increase, 22.0 + 3240 kJ/mole up to 0 = 0.22, and then to remain constant [191,
495]. The rate of adsorption was still significant after 8 h at 201 torr and 250°C
[494]. The activation energy for desorption decreases with converge [191,495].
A change in the preexponential factor in the rate of adsorption [494, 495] and
for desorption [495] is interpreted as a compensation effect or as a change in the
properties of the molecular precusor with coverage [494].
The entropy of the adsorbate is calculated from transition state theory.
For 0 < 0.10 the adsorbate is immobile; for 0.22 < 0 < 0.30 the adsorbate is
immobile and dissociated [191].

2.6.6.4 Desorption Kinetics

On single crystal surfaces the desorption temperatures are 980 K [474,484] for
N/Fe(l 00),920 K [468] for N/Fe(11 0), and 860 K [474] for N/Fe(III).
N2 desorption from N * chemisorbed on a Fe-overlayer on Ru(OO 1) shows
a temperature programmed desorption peak at 850-950 K [506].
The desorption of N2 from the catalyst has been studied [406,497].

2.6.6.5 Hydrogenation of Chemisorbed N

The hydrogenation of adsorbed N 2* and N * has been studied [418, 507].
Preadsorbed N * is removed exclusively as NH3 at 218-313 °C [418]. The rate of
reaction of N * with H2 is first order in H2 [508], the rate increases with
increasing ON. [508] and the rate is higher for D2 than for H2 [508]. 15N
isotopic labeling shows that all chemisorbed N-atoms undergo hydrogenation
with equal probability [422]. By exposing N * adsorbed on single crystal Fe
surfaces to an increasing pressure of H 2(g) at 580 K, the disappearance of N * is
observed at approximately 100 torr of H2 [414]. For catalysts the rate of
hydrogenation of preadsorbed N * depends on the adsorption temperature
[509,419].
By field mass spectroscopy of N2 + H2 interacting on an Fe tip at room
temperature H, H2 [510], N2 [510], N [510] N2H [510] and NH3 [511] have
been detected. By coadsorption of N2 and H 2, FeN2 is seen [379].

2.6.6.6 Properties of Chemisorbed N

The N (Is) ultra violet photoelectron spectroscopy peaks are - 5.0 and
-1.8 eV [474] for Fe(1 00) and - 5.4 and -1.8 eV [474] for Fe(III). Ultra
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 55

violet photoelectron spectroscopy of N z adsorbed on Fe(lll) at 140-1000 K

shows peaks at 5 eV, interpreted at N* [468].
For N z adsorbed on single crystal surfaces, the X-ray photoelectron spectro-
scopy N(1s) peaks are 380eV for Fe(1 00) [474] and 397 eV for Fe(111) [462,
463]. X-ray photoelectron spectroscopy of N (Is) in Nz/Fe(pc) adsorbed at
- 80°C shows two peaks at 405.3 eV and 400.2 eV [471], while X-ray photo-
electron spectroscopy of N (Is) in N/Fe (pc) adsorbed at 290°C shows one peak
at 397.2 eV [471]. The absence of a shift in core level energy for Fe upon N *
adsorption shows that the Fe-N bond in N * is covalent [52].
High-resolution electron-energy-loss spectroscopy of N */Fe(lll) shows
N-Fe at 450 cm -1 [463, 470].
For catalyst models infrared spectroscopy after N z adsorption shows a band
at 1820cm- 1 [419] for Fe/SiO z . This band disappears upon exposure to Hz
[419]. On Fe/MgO bands are observed at 2200 and 2050cm- 1 [512].
On single crystal surfaces the work function increases smoothly with cover-
age [474] on Fe(1 00), while for the Fe(lll) surface the work function behavior
is more complicated [474]. The change in work function at saturation is
+0.33 eV for N*/Fe(1 00) [474] and +0.25 eV for N*/Fe(lll) [468,474].
The work function during N z adsorption at - 80°C on Fe-films goes
through a minimum [513]. The shape but not the absolute value is reproducible
[513].
On catalysts N z chemisorption has been reported to decrease the work
function [243, 514] or to have no effect on the work function [243, 279].
For N z chemisorption on Fe/SiO z at 400°C little change in the mag-
neti-zation is observed [410].

2.6.6.7 Effect of Promoters

The reported structures of N */K/Fe (100) are c (2 x 2) [229,466] and (3 x 3) for
N*/K/Fe(lll) [500]. With up to 30% of the saturation coverage by K, the
presence of K does not affect the saturation coverage of N * [229].
K increases the strength of N z chemisorption [412]. At small coverages by
K, the number of promoted sites is proportional to the coverage by K, each
K forming 1-2 promoted sites [229,466]. At higher K-coverage, the number of
promoted sites decreases with increasing coverage by K due to surface blocking
by K [466]
K increases the rate of chemisorption for N z [231,466, 515]. The sticking
coefficient is 1.4·10 -7 for Fe (1 00) at 430 K, 0 /-lmole K/mz [466]; 3.9' 10- 5 for
Fe (1 00) at 430K, 2.5/-lmoleK/m z [229,466]; 5'10- 6 for Fe(111) at 430K
O{tmoleK/mz [466],4'10- 5 for Fe(lll) at 430K, 3.3/-lmoleK/m z [466];
4.10- 6 for Fe(pc) at 423 K, 0 /-lmoleK/m z [500], and 2.5 .10- 5 for Fe(pc) at
423 K, 2.2 /-lmole K/m z [500]. A maximum sticking coefficient is seen for
3.3 /-lmole/m z [229] of K atoms.
The activation energy for N z adsorption on K/Fe(1 00) is approximately
0[229].
S6 P. StoItze

Thermal desorption of K from N */K/F e shows peaks at 570--810 K [466,

500]. Temperature programmed desorption shows that N * increases the stabi-
lity of K * [466].
From electron spectroscopy studies it was found that N * and 0 * interacts
with K [515]. On the surface K is associated with 2 or 3 chemisorbed O-atoms
[515].
Al decreases the N2 chemisorption on an area basis [210,211, 516].

2.6.6.8 Effect of Preadsorbed Species

Preadsorption of 0* inhibits the formation of N* [476,477,490]. One 0*
blocks the formation of one N * [477].
In a study [517] ofN 2 adsorption on Fe(111) contaminated with 1.2,umole
O 2 per m 2, the initial sticking coefficient for N2 into 2N * was 10- 7 at 420 K and
2.10- 7 at 470K.
For (Fe, AI) samples the rate of adsorption for N2 is increased by the
presence of H2 [518, 519]. Chemisorption on an industrial catalyst give evidence
that Nand H chemisorb on identical sites [496].
CI blocks the adsorption of N2 [449].

2.6.6.9 Isotopic Exchange

The rate of 14N_15N exchange becomes measurable at typical reaction tem-
pera-tures or NH3 synthesis [520].
The N 2 isotopic exchange does not occur on Fe films [488], but does occur
on fused Fe [490]. These findings may be related to the low surface area of these
materials.
The kinetics of N-isotopic exchange was interpreted as N-N bond breaking
being rate limiting [482, 520--523].
The activation energy for N 2 isotopic exchange is 36 kcal/mole [482],
33 kcal/mole [522] or 58 kcal/mole [524, 525].
N-isotopic marking show that under NH3 synthesis, NH3 (g) and N * are in
equilibrium [526, 528], while N * and N2 (g) are not in equilibrium [527, 528].
The N2 + N2 isotopic exchange, NH3 + N2 isotopic exchange is much
faster [529, 530].
The rate of isotopic exchange measured for the gas phase in the presence of
a catalyst is the same as the rate of exchange between the gas and chemisorbed
N * [482,521], indicating that the isotopic exchange predominantly proceeds on
the surface of the catalyst.
The rate of isotopic exchange has been found to correlate with the catalytic
activity for a series of catalysts [531] or even to be the same as the rate of
ammonia synthesis [521]. Model calculations show that the coverage by N * is
high during NH3 synthesis [396]. This is not necessarily the case during
a N2 isotopic scrambling experiment. This may explain why the rate of NH3
synthesis does not always correlate perfectly with the rate of N2 isotopic
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 57

scrambling. These results give evidence for the dissociative mechanism of am-
monia synthesis [520].
Addition of Hz to N z has been reported to increase the rate of 14N + 15N
isotopic scrambling [288,481,488, 520, 522, 532, 533], to have no effect of the
rate [534] or to decrease the rate [288]. The observation of an increase in the
rate of isotopic exchange in the presence of H has been interpreted as H remov-
ing an oxide from the surface [481, 533].
The presence of K has been found to be important for the effect of H [288].
N 2 isotopic exchange over fused Fe is inhibited by Oz [490]. The activation
energy for nitrogen isotope exchange is higher in the presence of chemisorbed
oxygen [534].

2.6.7 Kinetic Models of N z Chemisorption

The numerical models of the kinetics of NH3 synthesis will be discussed in Sect.
2.7.3 All contain a model of the kinetics of N z adsorption
(4)
(5)
as a special case. From the kinetic models the coverage of N * may be calculated
as a function of exposure. These calculations, shown in Figure 2.1, are in good
agreement with experiments at low and moderate exposures [396]. The cal-
culated peak temperatures, see Figure 2.2, as well as the peak shapes are in
reasonable agreement with experiments [396].

0.8

Q.o 0.6
2
~
u 0.4

Fig.2.1. Calculated coverages by N * vs expo-

sure for N2 adsorption on Fe. Pressure
10- 6 torr. Temperature 300 K (upper curve),
350 K, 400 K and 500 K (lower curve). Repro-
duced from [396]

Exposure (Po' s)
58 P. Stoltze

e
OJ

c
o
~
I-
o
Vl
OJ
o

Fig. 2.2. Calculated TPD curves for N2

desorbing from Fe. Heating rate 10 K/s.
Initial coverage 0.20 (lower curve), 0.40,
0.60, 0.80 and 1.00 (upper curve). Repro-
duced from [396J
500
Temperature (K)

N z * is weakly adsorbed on the surface of Fe. One model [396] calculates

that the equilibrium coverages of N z * under 101 kPa of N z is 0.27 for Fe and
0.30 for K/Fe at 500 K. The coverage is even smaller at higher temperatures
[396]. These calculations are made based on the assumption that N z * does not
dissociate to form N *. N * is much more strongly adsorbed than N z *. Under
101kPa ofN z , the equilibrium coverage by N* is close to unity at 500-1000 K
[396].
The activation energy for the sticking coefficient is small and the low value of
the sticking coefficient must be ascribed to an unfavorable activation entropy.
The negative activation energy found for adsorption on Fe(111) and K/Fe may
be attributed to the top of the activation barrier for N z * being below the level
for N z (g) [396]. An equivalent description may be developed from the statement
that the decrease in sticking coefficient at higher temperatures is caused by the
shorter lifetime of N z * at higher temperatures [396].
Estimates of the sticking probability based on transition state theory are not
particularly successful. This may be caused by incomplete equilibration of the
transition state [396], by an underestimate of non-adiabatic quantum effects, or
because the one-dimensional potential energy surface is an oversimplification of
the problem.
If a Fe surface is given a brief but large exposure to N z at low temperature,
the surface will be covered by N z * [396]. The dissociation to N * is thermodyn-
amically favorable, but does not proceed as virtually no free sites are left [396].
At a higher temperature the coverage by N z * will be less complete and there will
be a sudden transition to complete coverage by N * [396]. At even higher
temperatures the coverage by N * will be less than unity [396].
The thermal desorption of N z has been treated by Bowker, Parker and
Waugh and by Stoltze and N0rskov. In their first paper [535], Bowker, Parker
and Waugh used a conventional prefactor for the rate of N * recombination.
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 59

This resulted in a somewhat too high binding energy for N * and led to the
conclusions that the enthaply measured at high coverages by N * might be more
appropriate [536]. Later the pre factors were adjusted [537] resulting is a better
value for the stability of N *. Using the data of Stoltze and N0rskov, Bowker,
Parker and Waugh [537] found a broad peak for the thermal desorption ofN 2 .
This discrepancy was resolved by Trivino and Dumesic, who concluded that
both the data-set of Stoltze and N0rskov and the revised data set of Bowker,
Parker and Waugh reproduce the experimental data [474].

2.6.8 Chemisorption of NH3

2.6.8.1 Structure of Chemisorbed NH3

NH3 may be adsorbed on Fe without decomposition at low temperatures; at
higher temperatures a number of species are formed [538]. For NH3 chemisorp-
tion on a stepped Fe(1 00) surface at low temperatures, ESDIAD shows that
NH3 is bonded to the surface through the N-atom [539] and that the molecule
has free rotation around the nitrogen-surface bond [539]. Under increasing
exposure at 110 K, chemisorbed NH3 in on-top geometry and in multico-
ordinated geometry; adsorption on top of the chemisorbed layer and finally
multilayer chemisorption is observed [540]. Above 155 K only chemisorbed
NH3 is formed [540]. Based on an interpretation of the ultra violet photo-
electron spectroscopy spectra, the species formed upon adsorption of NH3 on
Fe is assigned to NH3 * at 120-300 K [541], to NH * at 350 K [541] and to N *
at 500 K [541].
NH3 * + D* isotopic scrambling is observable at low temperature for
NH3/Fe(lll) [397] but not for NH3/Fe(11 0) [541].
Low-energy electron diffraction for NH3/Fe(1 00) [542] and NH3/Fe(lll)
[542], [429] shows disorder. Low-energy electron diffraction for NH3/Fe(11 0)
shows disordered c(2 x 3) at 120 K [541], (2 x 2) at 280 K [429] and at 340 K
[541]. The (2 x 2) has been interpreted as NH* [541]. By NH3 adsorption on
Fe(lll) followed by flashing to 310 K, a (3 x 3) structure is observed [543].
More complex patterns are observed on stepped surfaces [543].
Facetting of Fe(1 00) after high temperature exposure to NH3 has been
observed [542].
By exposure of a Fe film to NH 3, several layers of nitride may be formed
[544]. By reaction of this nitride layer with D 2 , the product is mostly NH3
indicating that the film contains H [544].

2.6.8.2 Thermodynamics
The enthalpy of chemisorption is - 71 kJ/mole [541] for NH3/Fe(11 0) and
- 84 kJ/mole [397] for NH3/Fe(lll) from temperature programmed desorp-
tion.
60 P. Stoltze

2.6.8.3 Adsorption Kinetics

At 120 K the initial sticking coefficient is 0.16 on Fe(11 0) [541].

2.6.8.4 Desorption Kinetics

At low coverage NH3 desorbs from Fe(11 0) at 255 K [541], from Fe(111) at
330 K [397] or 380 K and 435 K [543]. Desorption of N2 is observed at 900 K
for NH3/Fe(111) [397] and at 900 K from NH3/Fe(pC) [229]. Desorption of
H2 is observed at 390 K from NH3/Fe(111) [397].

2.6.8.5 Dissociation
For NH3/Fe(100) [542] and NH3/Fe(111) [397, 542] the dissociation be-
comes observable at 160 K and is complete at 320 K. For NH3/Fe(11 0) the
dissociation temperature is 260-290 K [540].
The dissociation of NH3 * on Fe(11 0) was reported to give N * and H * as
the only reaction products [540]. Others have found evidence for the intermedi-
ate formation of NH * [545].
By field ion mass spectroscopy of NH3 on Fe at room temperature,
4.10- 4 torr, NxHy species and FeNxHy species are detected [510]. Secondary
ion mass-spectroscopy of NH3/Fe(11 0) at 130 K shows FenNHW+ with
n, m = 1,2, NH n+ with n = 0,1,2,3,4, H+, Hi and Fe+ [545]. The spectrum is
interpreted as fragments of molecularily adsorbed NH 3* [545]. When the
temperature is increased, FeNHt and FeNH~ + decrease smoothly, while Fe +
increases smoothly, illustrating the decrease in surface coverage and the reaction
of NH3 (g) with the surface at higher temperatures [545]. The intensity of NHt
with NHi decreases steeply above 300 K consistent with the temperature
programmed desorption studies [545]. Both ions are probably formed from
adsorbed species [545]. The intensity ofNH+ decreases more slowly than NHi
and NHj; the reason is probably that NH+ is formed from both NH3(g) and
NH* [545].

2.6.8.6 Properties of Chemisorbed NH3

Ultra violet photoelectron spectroscopy peaks for NH3 * are - 7.4 and
- 11.8 eV on Fe(1 00) [542], - 6.4 and - 11.2 eV on Fe(11 0) [541], -7.2
and - 11.6 eV on Fe(111) [397, 542]. NH*/Fe(11 0) formed by thermal
decomposition of NH3 */Fe(11 0) shows ultra-violet photoelectron spectro-
scopy peaks at - 5.2 and - 8.4 eV [541].
The X-ray photoelectron spectroscopy N(1s) spectrum of NH3/Fe (pc) shows
one peak at 400 eV at 85 K [471] and two peaks at 397.2 and 400 eV at 290 K
[471]. X-ray photoelectron spectroscopy and ultra violet photoelectron spectro-
scopy of NH3/Fe(PC) are interpreted as Fe, after 2.10 7 L at 670 K, 5·10 -4 torr
[229] and FexN, with x approximately equal to 2, after 5.4· 10 10 L at 670 K,
1 torr [229].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 61

Auger electron spectroscopy shows that the core-level energy for Fe is

independent ofNH 3 coverage [229] indicating that the N-Fe bond is covalent.
High-resolution electron-energy-Ioss spectroscopy spectra ofNH3/Fe( 11 0)
at 120-315 K were reported and interpreted [540, 546]. The geometry of the
adsorbed molecule is C 3v , i.e., the symmetry of the surface has a negligible effect
on the adsorbed molecule. The high-resolution electron-energy-Ioss spectro-
scopy spectra are interpreted [540] as Fe-N stretch at 420 cm -1 for NH3 *,
400cm- 1 for ND 3*/Fe, symmetrical NH3 stretch at 1170cm- 1 for NH3*'
905cm- 1 for ND3*, symmetrical NH3 stretch at 3310cm- 1 for NH3*'
2410 cm -1 for ND3+ *.
The change in work function at saturation is - 1.98 eV for NH3/Fe(1 00)
[542], - 2.4 eV for NH3/Fe(11 0) [541] and - 2.05 eV for NH3/Fe(lll)
[542]. For NH3/Fe(11 0) the work function decreases with exposure and passes
through a minimum, - 2.4 eV, at 15 L, 120 K [541].

2.6.8.7 Effect of Preadsorbed Species

The adsorption of NH3 on Fe overlayers on Ru(OO 1) has been studied [506].
Preadsorbed N * decreases the initial sticking coefficient and the enthalpy of
chemisorption for NH3 [541]. The saturation coverage for NH3 is unchanged
[541] or decreased [547].
The tendency to dissociate is reduced by preadsorbed c(2 x 2) N/Fe(1 00)
[542] and (3 x 3) NfFe(lll) [397].
For a Fe(lll) surface contaminated with 1.21lmole 0/m 2, the initial stick-
ing coefficient is 0.03 at 300 K [517].

2.6.9 Adsorption of N2H4

Fe is an active catalyst for N2H4 decomposition at 26 and 365°C [548].

Labeling with 15N shows that the N2 bond is not split [548]. Exposing Fe to
N2H4 at 243 K leads to the formation of NH 3, H2 and N2 [549].
During adsorption ofN 2H 4 on Fe(III), the tendency of dissociate is great.
At 126 K ultra violet photoelectron spectroscopy shows the presence of N 2H4 *
after 0.5 L exposure [550]; condensation is detected at 80 L, 126 K [550]. At
220 K ultra violet photoelectron spectroscopy and X-ray photoelectron spectro-
scopy detect N 2H 4* and NH x * after 31 of exposure [550]. A 550 K N * is
formed during exposure [550].

2.6.10 Chemisorption of O 2

The chemisorption of O 2 on Fe has been the subject of a number of studies. This

reaction is important for the catalyst during NH3 synthesis and for the passiva-
62 P. Stoltze

tion of the catalyst to prevent an uncontrolled oxidation upon exposure to air

[89, 551].

2.6.10.1 Structure of Chemisorbed 0

The chemisorption of O 2 is complicated, as chemisorption is not clearly distinct
from bulk oxidation.
On single crystal surfaces a c(2 x 2) low-energy electron diffraction pattern is
formed on Fe(1 00) [552J and one Fe(11 0) [553, 555]. A c(3 xl) [553, 555J and
a split c (3 x 1) [554J patterns are seen on Fe(11 0) at intermediate coverage.
A p(1 x 1) pattern is seen for Fe:O = 1: 1 on Fe(1 00) [552, 556, 557J and on
Fe(110) [553].
Only at low temperatures will the p(1 x 1) surface structure be reasonably
completed before the bulk oxidation starts [558, 559]. The chemisorbed layer
may be amorphous if the exposure is made at too Iowa temperature [557]. At
room temperature, exposure of the c(2 x 2) structure to O 2 results in the growth
of a 3 dimensional oxide [552, 554].
Auger electron spectroscopy and ultra violet photoelectron spectroscopy of
O 2 chemisorption at 77 K on Fe have shown that the formation of oxide may be
followed by the formation of molecularily adsorbed O 2 [476]. Evidence for the
adsorption molecular O 2 , even in the absence of oxide, has come from X-ray
photoelectron spectroscopy spectra for O 2 adsorbed on single crystal surfaces,
where a peak at - 533.6 eV is interpreted as the molecular precursor [560].
In suitable temperature ranges FeO can be observed to grow epitaxially on
Fe(100) [561J and on Fe (1 10) [553, 555, 562]. With further exposure the
sticking coefficient increases and the low-energy electron diffraction picture
disappears [561]. This is interpreted as the nucleation and growth of FeO,
which finally results in the formation ofp(1 x 1) oxide on Fe(l 00) [552J and on
Fe(110) [553].
For (Fe, AI) samples an unstochiometric magnetite [563-566J or Fe(III)
oxide [563, 564, 567J may be formed. Passivated (Fe, Co, AI) samples consists of
a Fe-Co alloy core; the oxide skin is enriched in Co [60]. The surface is covered
with small particles of Fe(III) [566].
Mossbauer spectroscopy of the industrial catalyst shows that some Fe304 is
formed [452]. Oxygen may be present in more than one oxide phase [568]. The
chemisorption of oxygen on the surface of the catalyst is not always sufficient to
prevent the reaction with air [569J, possibly because metallic Fe may migrate
through the oxide layer during storage [569, 570].

2.6.10.2 Thermodynamics
The amount of O 2 taken up under passivation in 0.75-1.0% O 2 in N2 is 3-4%
by weight [571]. The oxide film formed is approximately 30 A [122J or 6-12
atomic layers of oxide [572].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 63

The enthalpy of chemisorption is 0-120 kcaljmole for (Fe, AI) samples [439J
at -183°e and 0-100 kcaljmole for (Fe, AI, K) samples [439J at the same
conditions.
The enthalpy of chemisorption for an industrial catalyst is 420 kJjmole up to
0.25 monolayer of O 2 , then the enthalpy drops to 34 kJjmole [573]. The
strength of adsorption and the adsorption capacity for O 2 on the catalyst
increases with temperature [574]. The optimum temperature range for passiva-
tion is 673-773 K [575].
By pulse chemisorption of O 2 on the catalyst, the O 2 chemisorption area is
1.7 m 2 jg at -78°e and 1.9 m 2 jg at 20 e for a sample with a BET area of
0

1Om 2 jg [576].
In the later stages of reduction the amount of oxygen adsorption is propor-
tional to the degree of reduction [122J.

2.6.10.3 Adsorption Kinetics

The initial sticking coefficient at room temperature for O 2 on Fe is 0.20 ± 0.01
on Fe(ll 0) measured by molecular beam techniques [577J, 0.13 [554J or 0.20
[577J measured by Auger electron spectroscopy and near unity measured by
high-resolution electron-energy-Ioss spectroscopy and LEED [555]. Below
room temperature the sticking coefficient remains constant almost to saturation
[578J, indicating the existence of a weakly adsorbed precursor. At room tempera-
ture the sticking coefficient decreases as expected for site blocking until the
c(2 x 2) structure is complete [552-554].
-- Following an interruption in exposure for Fe films an increase in sticking
coefficient is observed [577]. This is interpreted as the diffusion of Fe out
through the oxide layer [573, 577].
For the catalyst the rate of chemisorption of O 2 decreases to 1% ofthe initial
rate at 1 monolayer of 0* and to 10- 6 times the initial rate at 50 monolayers
[570J.

2.6.10.4 Desorption Kinetics

During temperature programmed desorption of the passivated catalyst, H2 and
N 2 , but no O 2 is observed [579].

2.6.10.5 Properties of Chemisorbed 0

For the c(2 x 2) OjFe(100) structure the ultra violet photoelectron spectro-
scopy O(ls) peak is - 5.5 eV [552, 580]. This feature is interpreted as chemisor-
bed 0 * [552]. At higher coverages peaks are observed at - 1.5, - 2.5 and
- 5.0eV [552, 580]. X-ray photoelectron spectroscopy shows peaks at 531.7 eV
for chemisorbed oxygen [552, 578J and for oxide at 530.3 eV [560, 578]. X-ray
64 P. Stoltze

photoelectron spectroscopy of Fe (2P3/2) shows that for Fe: 0 < 1: 1.5 the oxi-
dation state for Fe is + 3 [578] and not + 2 which has been suggested earlier.
At small coverages the work function increases with the coverage until
a maximum is reached at the c(2 x 2) structure [552]. At higher coverage the
work function decreases and goes through a minimum [552, 581].
The complicated behavior of the work function is not reflected in the bulk
electronic structure. The electrical resistance of a Fe film increases smoothly
with O 2 exposure [582].

2.6.10.6 Oxygen Isotopic Exchange

For an industrial catalyst passivated in industry no 18 0 + 17 0 exchange is
detected, while catalyst passivated in the laboratory catalyses this reaction
[583]. After evacuation, the surface of the prereduced catalyst is active in the
isotopic exchange for molecular oxygen [584]. The occurrence of O-isotopic
exchange was interpreted as some 0 being loosely associated with metallic sites
[569]. This reaction is poisoned by CO and water [584].

2.6.10.7 Effect of Promoters

Scanning electron microscopy shows that Al diffuses into the bulk during
oxidation [585].
Low-energy electron diffraction shows the formation of a c(2 x 4) structure
on Fe(11 0) [54]. There is no evidence for the formation of K-oxide at
K-coverages less than one monolayer [554]. The bulk oxidation starts approx-
imately at Fe: 0 = 1: 0.5 independent of K coverage [554]. N * and 0 * interacts
with K [515].
For O 2 chemisorption on KjFe(11 0) the sticking coefficient is unity [554].
The desorption temperature for K from KjFe is increased from 670 K to 810 K
in the presence of co adsorbed 0* [500].
The 0(1s) X-ray photoelectron spectroscopy peak at 530.0 eV is indepen-
dent ofK coverage and ofO-coverage [554]. At small coverages by 0* the work
function is decreased for OjKjFe due to depolarization of K-K interactions
[500].

2.6.10.8 Effect of Preadsorbed Species

Preadsorbed N * does not inhibit O 2 chemisorption [415] since the preadsorbed
N * is displaced to the bulk Fe upon O 2 exposure at 300 K [477, 486]. The
presence of N in the bulk decreases the tendency for oxidation of the Fe [477].
The N dissolved in the bulk can only be removed by desorption after the 0 *
layer has been removed by reduction [477].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 65

2.6.11 Adsorption of HzO

At 77 K HzO is chemisorbed without dissociation [586, 587]. The chemisorp-

tion energy ofHzO is 54.6-66 kJ/mole [586, 587]. Sputtering of the surface does
not destroy the passivating effect of the adsorbed HzO [588], while the electron
beam during a Auger electron spectroscopy experiment leads to oxidation and
destruction of the passivation [588].

2.6.12 Adsorption of HzS

The chemisorption of S on Fe(100) by exposure to sulfur vapors leads to

a number of structures [589]. At a Fe: S-ratio of 0.5 the structure is c(2 x 2)
[589]. There is some evidence of a p(2 x 2) structure at lower coverages [589].
The rate limiting step for the formation of chemisorbed S * from HzS on Fe is
the dissociation of HzS [423].
HzS poisons the adsorption ofN z on Fe-films [513]. When a (Fe, AI) sample
is poisoned by HzS, FeS is observed in the X-ray power diffraction diagram if
the H1S: Hz ratio is sufficiently high [590]. At smaller HzS concentrations
a monolayer of S is formed [590].
After poisoning with 1.6 ppm HzS in 3H z + N z at 303°C, a monolayer (i.e.
0.4-0.5 mg/mZ) is adsorbed [590]. The catalyst activity is lost a 0.2 mg S/ml
[590].
Chemisorbed HzS may not be removed with 3H z + N z below 620°C [590].
The addition of K and Al to Fe increases the resistance to poisoning by HzS
at 400-500°C, 1 atm [591].

2.7 The Mechanism of NH3 Synthesis

The key step in the synthesis of NH3 form N z + 3 Hz is to dissociate the N-N
triple bond in the N z molecule. The direct gas phase reaction would involve
extremely endothermic and exothermic reactions. The resulting activation ener-
gies would be prohibitively high according to the principle of Sabatier.
All imaginable reaction mechanisms can be separated into two main cases.
In the associative mechanism H is added to the N z molecule before dissociation
of the N-N bond, e.g.
(6)
N z*+ 2H *::::::;;:NzH z * + 2* (7)
NzH z * + *::::::;;:2NH* (8)
66 P. Stoltze

Very endothermic or exothermic reaction steps are avoided if the N-N bond is
broken synchronously with the addition of H to the N-atoms.
In the dissociative mechanism the N-N bond is dissociated before any N-H
bond is formed, e.g.

(9)

N 2 * + *~2N* (10)

N* + H*~NH* (11)

Very endothermic or exothermic reaction steps are avoided if the N-N bond is
broken synchronously with the formation of the N * surface bond.
The number of distinct mechanisms is further varied by the number of
intermediate steps and the number of sites involved in the bonding of each
intermediate. A large number of mechanisms result in kinetics expressions which
are indistinguishable from an experimental point of view.
Catalysis may be understood at several levels. In recent years the under-
standing of ammonia synthesis has been taken to the level where the high
pressure reaction has been treated in terms of numerical models based on
a description of the reactants at the atomic level. In the present section we will
first address the questions on the nature of the catalytically active structure in
the catalysts and the information on the mechanism of ammonia synthesis
available from chemisorption studies. We will then describe some models of
ammonia synthesis in some detail and then proceed to discuss the remaining
aspects of the mechanism of ammonia synthesis based on these models.

2.7.1 Nature of the Active Structure

Experimentally it has been verified that the synthesis of ammonia takes place on
the surface of the catalyst [592] rather than in the bulk. However, isotope
labeling experiments seems to indicate that small amounts of ammonia may be
formed in the bulk at very high temperatures [593]. X-ray photoelectron
spectroscopy demonstrates that Fe is present in the surface of the reduced
catalyst as Feo and not as Fe oxide [52, 208].
Under ammonia synthesis conditions, ultra-high vacuum data demonstrate
that a bulk nitride is not formed [594] whereas during NH3 decomposition the
kinetics indicate that the reaction may proceed on a completely nitrided surface
[595]. The increase in reaction rate observed by the addition ofH 2 is interpreted
as the reaction being faster on Fe metal than on Fe-nitride [595].
The coverage of the total surface by catalytically inactive structural pro-
moters has been determined by a number of techniques. The coverage by
catalytically inactive material is 60%, calculated from kinetic data for NH3
synthesis [596], 55% [191,207],60% [207], or 45% [195] from chemisorption
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 67

measurements, 45% from oxygen exchange with l80-labeled water [195], and
26% from DzjH z isotopic exchange [191]. The active fraction of the total area
generally decreases with increased promoter content [198, 199]. The active area
decreases through the sequence Mg > Be > AI, Si, Cr > Mn > Ca > Ti [198,
199] for a fixed promoter concentration. Reduction at a very high temperature
(1073 K) increases the NH3 synthesis and Nz-chemisorption rate; this has been
interpreted as reduction of a Fe(II)-spinel [123].
Calculated from the activation entropy, the density of active sites is 10 12 per
cm z for a porous Fe-catalyst and 2.5' 10 16 per cm z for a Fe-film [597].
For (Fe, AI) samples large N z chemisorption correlates with high catalyst
activity [524, 525]. In some cases no correlation between N z chemisorption and
catalytic activity has been found. This lack of correlation has been ascribed to
heterogeneity of the surface [598] or to structural sensitivity for FejMgO [328].
While the catalytic activity for the catalyst does not correlate with BET-area
[599], the catalytic activity has been found to be proportional with the Fe area
as determined by CO-chemisorption [53, 600]. Others have found no correla-
tions or a complex relationship [164, 601], probably because promotion is
important.

2.7.1.1 Structural Sensitivity

If we conclude that the active structure is the metallic iron surface, an important
question is whether all basal planes have the same catalytic activity [602]. From
the results for the chemisorption of N z on single crystals, the answer is that the
rate of chemisorption does differ among the basal planes. Also the presence of
chemisorbed K changes the rate of chemisorption significantly. For the catalyst
differences in catalytic activity between the basal planes would be reflected in
a dependence of the catalytic activity on the particle size, since smaller particles
expose a relatively large number of atoms as high index planes or edge atoms.
The rate of NH3 synthesis has been measured at 20 atm on single crystals of
Fe [603, 604] in a high pressure micro reactor. The ratio of activities of NH3
synthesis on a Fe-single crystal is (111):(100):(110) = 418:251 [604].
The catalytic activity for FejMgO depends on the particle size of Fe; the
cause for this has been assigned to the structure sensitivity [322-324, 328]. The
increase in synthesis rate measured for FejMgO after reduction in pure ammo-
nia is interpreted as surface reconstruction [324].
The high activity of the more open planes of Fe for both N z chemisorption
and NH3 synthesis [605] has been assigned to the existence of C 7 sites [505].
The high activity of C 7 sites has been explained in geometric terms [505]. Based
on X-ray photoelectron spectroscopy studies of active clusters in (Ni, Fe)-alloys
with 0--20% Fe it was concluded that an active site for NH3 synthesis consists of
approx 7 Fe atoms [606].
The investigation whether the catalytic activity differs among the basal
planes of the iron is complicated by the recrystallization of the Fe surface during
NH3 synthesis. This has been observed both for unpromoted Fe [607], for (Fe,
68 P. Stoltze

AI) catalyst models [608] and for Fe single crystals with evaporated overlayers
of Al and K [232].

2.7.1.2 The Effect of K

The segregation of K to the surface has been demonstrated by chemisorption
measurements [207, 220], by scanning Auger electron spectroscopy [52,
209-211], and by X-ray photoelectron spectroscopy [208]. Chemisorption
measurements [207] and spectroscopic studies using a Fe film in a microreactor
[609] have shown that K is in close contact with Fe on the active surface.
Addition of a small amount of K results in an increase in the catalytic
activity for ammonia synthesis [139, 217, 240, 242, 244, 248, 249, 288,
289, 609-611] and an increase in activity for nitrogen isotopic exchange
[611]. For ammonia synthesis the acitivity effect depends on the pressure [563,
564].
Addition of too much K increases the activation energy for ammonia
synthesis [176, 242]. The optimum concentration of K [67,68] depends on the
nature and concentration of the structural promoters [291], and on the prepara-
tion method [612].
Kinetic and spectroscopic studies of an Fe film after NH3 synthesis suggest
that the absorbed K atoms are associated not only with N but also with 0 on
the surface [609]
While it is fairly obvious that Al and Ca are structural promoters and
increase the activity of the catalyst by increasing the specific area, it is also
obvious that K does increase the catalytic activity, but does not. do so by
increasing the specific area. While the effects of K on the kinetics of NH3
synthesis may fairly easily be observed in a high pressure reaction experiment,
the cause of the effects are very hard to deduce from the observed kinetics.
Evidence for more complex changes in the kinetics [613] has been presented.
A complication in the interpretation of the effect of K is that the observed
changes in the kinetics depend on the operating conditions for NH3 synthesis
[563, 564, 614, 615] and decomposition [236].
The synthesis of ammonia has been studied in micro reactors over single
crystal surfaces. At 20 atm the equilibrium coverage by K is 0.15 [232]. K in-
creases the activity of Fe(1 00) and Fe(111). At 0.3% conversion the rate is
increase by a factor of 2 in the presence of K [232]. The effect of K is larger at
higher conversions [232]. Fe(11 0) and KjFe(11 0) do not show any activity for
ammonia synthesis [232]. The results for NH3 synthesis over single crystals thus
show that a significant difference in activity between the basal planes persists in
the presence ofK [616]. However, the results obtained for NH3 synthesis over
single crystals are limited to small conversions and thus the results may be more
directly related to the case of a catalyst operating under unusual conditions than
to the case of typical reaction conditions.
A detailed study of the rate of the catalytic reaction over Fe films precovered
with K has been performed in a micro reactor [609]. This study clearly demon-
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 69

strated the promoting effect of K and the existence of an optimum coverage

ofK.
Based on studies of the catalyst, the origin of the promoting affect of K has
been assigned to a decrease of the work function [217J, to changes in the
popUlation of reaction intermediates [614, 617-619J, to a shift in rate limiting
step [620J, to a decrease in the dipole moment of the transition state [621, 622J,
or to a destabilization of NH3 * [221, 232]. Others have speculated that the
binding energy for intermediates may be too high on the surface of small Fe
particles and that the primary effect of K is to induce particle growth [623,624].
The numerical models show that the coverage by NH 3* during ammonia
synthesis is small [396, 625]. This makes it difficult to understand why a de-
stabilization of NH3 * by K should be responsible for the promoting effect of
K [232, 227].
The most direct information on the origin ofthe effects comes from quantum
mechanical calculations and single crystal chemisorption studies. From single
crystal studies it is concluded that the effect of K is to increase the stability of
N z * [466]. High-resolution electron-energy-Ioss spectroscopy spectra of N z *
Fe (11 1) and N z *jKjF e (1 1 1) show that K does not promote the dissociation of
N z * by weakening the N-N bond in N z * [475].
Based on quantum-mechanical calculations, the stabilization is expected to
be mainly electrostatic and of short range [626-629]. From studies of the Xe(5p)
photo emission from XejKjRu (00 1) it has been concluded that K decreases the
work function for the neighbor Ru-sites [630J and that this effect extends
approximately 6 Afrom the K nucleus [630].
The idea that the promoting effect of K is due to a stabilization of N z * has
been examined within the model of NH3 synthesis by Stoltze and N0rskov. The
stabilization of N z * in the presence of K is 12 kJ jmole for Fe (111) [396, 625J as
deduced from the changes in the temperature programmed desorption peak
temperatures [231, 466]. Although it is a complication that the different basal
planes exposed in the catalyst have very different catalytic activity in the absence
of K, the results of the calculations strongly support the electrostatic picture of
the promoting effect of K [396, 625].

2.7.2 Results from Chemisorption Studies

The ultra-high vacuum studies show that during NH3 synthesis N * is formed
[414]. As the desorption temperature for N* is well above normal synthesis
temperature, the desorption of N * as N z (g) will be slow at synthesis conditions
[414]. If N * were not consumed by the reaction, it would soon inhibit the
synthesis due to blockage of active sites [414]. These observations are consistent
with Nz(g) + 2* = 2N* being the rate limiting step for NH3 synthesis [414].
Temperature programmed desorption of N z from the industrial catalyst
resembles temperature programmed desorption ofN z in the range 85 to 220 K
70 P. Stoltze

[631]. This means that physisorbed N z * and chemisorbed N z * on the catalyst

have the same properties of KjFe(lll) as on the surface on the industrial
catalyst [631]. To the extent that the desorption ofN z from N * near 850 K may
be followed by temperature programmed desorption, it appears that N * also
has the same properties on KjFe(111) as on the surface ofthe industrial catalyst
[631].
Chemisorption ofN z [286, 632-640J and Hz [634J on Fe in relation to NH3
synthesis has been the subject of quantum-mechanical calculations. The dyna-
mics of N 2 chemisorption has been simulated using a semiclassical wave packet
technique [641]. The simulations agree with the molecular beam experiments in
the conclusion that vibrational excitation is of some importance.

2.7.3 Kinetic Models of NH3 Synthesis

Data for the kinetics measured on single crystal surfaces are important for the
study of catalytic reactions as the surface of single crystals are approximations
to the more complicated surface of catalysts [642]. This is supported by the
agreement between rate measurements for Fe single crystals [603, 604J and the
rate measured for an industrial catalyst at 1 atm [643].
The purpose of developing models of the kinetics and mechanism of catalytic
reactions starting with a description of the reactants at the atomic level is to
understand the kinetic phenomena, rather than to give a very accurate descrip-
tion of a few phenomena. While a model of a single phenomenon may give an
accurate description of this phenomenon, this description may not be unique
and extrapolation may lead to ambiguities for situations where no experimental
data exist. However, if all aspects of interest are described using one model and
this model is in reasonable agreement with available data, this model may also
be used with some confidence in situations where experimental data are sparse
or ambiguous.

2.7.3.1 Formulation of the Models

A number of models have been developed. The model by Bowker, Parker, and
Waugh is based on the reaction sequence
Nz(g) + *~Nz* (12)
N z* + *~2N* (13)
Hz (g) + *~Hz * (14)
H z* + *~2H* (15)
N* + H*~NH* + * (16)
NH* + H*~NH2* + * (17)
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 71

NH z * + H*::::;;;NH 3* + * (18)
NH3 *::::;;;NH 3 (g) +* (19)
Bowker, Parker, and Waugh proceed by using Arrhenius expressions with
known or estimated values for all prefactors and activations energies.
The model by Stoltze and N0rskov [396, 625, 644, 645J is based on the
reaction sequence

Nz(g)+ * ::::;;;Nz* (20)

N z* + *::::;;;2N* (21)
N* + H*::::;;;NH* + * (22)
NH* + H*::::;;;NHz* + * (23)
NH z* + H*::::;;;NH3* + * (24)
NH 3*::::;;;NH 3(g) + * (25)
Hz(g) + 2*::::;;;2H* (26)

with the explicit assumption that the rate limiting step is the dissociation of N z *
(27)
Stoltze and N0rskov proceed by applying statistical mechanical methods to this
sequence, essentially expressing the thermodynamic properties of reactants and
intermediates in terms of spectroscopic properties. Further they treat within the
same model, the kinetics of adsorption of N z as well as the thermodynamics of
adsorption for Hz and NH 3.
Trivino and Dumesic have considered both of these reaction sequences and
compared the results and the differences in approach.
While the models of Bowker, Parker, and Waugh and by Trivino and
Dumesic do not make a priori assumptions on the nature of the rate limiting
step, Stoltze and N0rskov make the explicit assumption that the dissociation of
N z * is rate limiting. The assumption leads to a considerable simplification in the
further treatment of their model. While Bowker, Parker, and Waugh, and
Trivino and Dumesic must calculate reaction rates iteratively, the model by
Stoltze and N0rskov allows the derivation of explicit solutions for the coverages
and reaction rates. Further, a number of aspects of the kinetics of ammonia
synthesis, such as the activation enthalpy and the reaction orders may be
investigated analytically in the model by Stoltze and N0rskov.
The solution of the models involves a number of approximations.
The reaction sites are treated as identical [396, 625, 645J; the reactants,
intermediates and products chemisorb competitively on these sites. The com-
petition for the sites important kinetic consequences. The identity of the sites is
justified by the results from temperature programmed desorption from single
crystals and by the result from quantum mechanical calculations.
72 P. Stoltze

The adsorption is assumed to be random. This is justified by the experi-

mental measurements of the temperatures of ordering for the reaction inter-
mediates. The binding energy on a given site is treated as independent of the
occupation of the neighbor sites [396J.
The gas phase is treated as ideal [396, 625, 645].
For calculations for comparison with laboratory data measured in plug-flow
reactors [14, 617J, the reactors are treated as isothermal, plugflow reactors
[396]. The diffusion limitations are negligible [617, 646].

2.7.3.2 Input Parameters

The input parameters in the model by Stoltze and N0rskov are taken from
experimental results. Vibration frequencies are taken from spectroscopic data
on single crystals [396, 625, 645]. The rate data [396, 625, 645J for the
dissociation of N2 * are taken from the sticking coefficient for N2 and its
activation energy. The binding energy for the intermediates is taken from
temperature programmed desorption spectra. For the desorption ofN * as N2 it
is important that the low sticking coefficient for N2 be taken into account [396].
Treatment of the desorption using a "normal" prefactor amounts to the implicit
assumption of a different mechanism for desorption than for adsorption and the
deduced binding energy for N * is too high.
The number of sites found by CO chemisorption is used for the number of
active sites for the catalyst [396]. Quantum mechanical calculations show that
the effect of K is to stabilize N 2*. Experimentally the differences in sticking
coefficient among the basal planes of Fe is small in the presence of K. The
activity of all sites are assumed to be that of KjFe(111) [396, 625, 645]. This
may be further justified by the close similarity of the experimental temperature
programmed desorption spectra of N 2* and N * for the industrial catalyst and
for KjFe(111) [631].

2.7.3.3 Test
By construction the model reproduces the thermodynamics of adsorption for
the intermediates and the kinetics of adsorption ad desorption for N *. As no
measured data for the catalytic activity have been used in the determination of
the input parameters, the model may be tested by a comparison between the
calculated and experimental rates of ammonia synthesis. The test is made by
calculating the exit ammonia concentration from the input composition and
operating conditions for the reactor.
Bowker, Parker, and Waugh in their first paper [535J did not appreciate that
the uniquely low sticking coefficient for N2 must also be reflected in a uniquely
low prefactor for the recombination ofN *. The somewhat too high value for the
binding energy of N * resulted in a reaction rate for NH3 synthesis too low by
a factor of 10 5 [535]. Bowker, Parker, and Waugh suggested the use of binding
energies appropriate for high coverages by N * available from gravimetric
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 73

Fig. 2.3. Comparison of the calculated and

experimental ammonia production over
a Topsoe KMl catalyst. The data set spans
a broad range of pressures (1-300 atm), tem-
perature (375-500 0C) and gas velocities. The
synthesis gas is a 1 : 3 mixture of N 2 and H 2 .
Reproduced from [724]

Experimental output

isotherms [191]. However, it has been pointed out that these data are not
consistent with measurements for the sticking coefficient over single crystals.
Bowker, Parker, and Waugh later adjusted their data and obtained agreement
with experimental rates within an order of magnitude.
Using a number of data sets at 1, 150 and 300 atm, Stoltze and N0rskov
found that the calculated exit concentrations are in good agreement with the
experimental results [396, 625, 645J, see Fig. 2.3. The differences between
calculation and experiment corresponds to an error in the rate of less than
a factor of 1.5 [396, 645]. For (Fe, AI) the calculated rates are too high by
a factor of about 3. This is clearly a consequence of assuming that all sites have
the activity of Fe(lll).
Trivino and Dumesic concluded [330J that the remaining discrepancy be-
tween the results by Bowker, Parker, and Waugh and by Stoltze and N0rskov
area caused by an unusual small number of active sites assumed in the model by
Bowker, Parker, and Waugh.
Not all the input parameters are equally important for the success of the
model. Stoltze and N0rskov found that the critical parameters are the prefactor
and the activation energy for the sticking coefficient, the ground state energy for
N 2 * and the ground state energy for N * [396, 645]. These parameters are all
rather accurately known from experimental data. The reason why these are the
critical parameters is that the first three parameters determine the rate constant,
while the groundstate energy for N * determines the number of free sites [396,
645].

2.7.4 The Nature of Reaction Intermediates

There is no gas reaction since no difference is seen between quenching the gas to
o°C or flow through a hot quartz tube [592J.
74 P. Stoltze

2.7.4.1 Nitrogen Dissociation

An important question for the reaction mechanism for NH3 synthesis is whether
any H is added to N z before the fission of the N-N bond. Mechanisms involving
the additions of H before the fission of the N-N bond are referred to as
associative while mechanisms involving N* are called dissociative. Even if the
kinetic expression by Temkin and Pyzhev was derived from considerations for
an associative mechanism, the evidence for the associative mechanism is rather
meager.
Attempts have been made to observe the intermediates under conditions
where the formation of NH3 could be hoped for. By field ion mass spectroscopy
of a Fe tip at room temperature the observation of the ion Nt [510,511] and
the observation of N z* by laser Raman spectroscopy of the catalyst [647] has
been interpreted as evidence for the associative mechanism [647]. Interestingly,
other studies using field ion mass spectroscopy have shown data incompatible
with the associative mechanism [648] or no reaction [649]. Presumably the
reaction conditions in these experiments are too far removed from conditions
where NH3 may be formed in appreciable amounts.
More direct experimental evidence, such as N-isotopic exchange over the
catalyst, the study of chemisorption of N z, and the synthesis of NH3 over single
crystals, leaves no doubt that the mechanism is dissociative.
While assumption that
(28)
is build into the model by Stoltze and N"JfSkov, the models by Bowker, Parker,
-and Waugh and by Trivino and Dumesic makes no a priori assumption on the
nature of the rate limiting step and actually calculate that the dissociation of
N z* is rate limiting under most reaction conditions.

2.7.4.2 Stability of Intermediates

While the thermodynamic stability may be deduced from experimental deter-
minations of the concentration of intermediates, these concentrations are hard
to determine in situ. In the numerical models the thermodynamic stability of the
intermediates may easily be calculated. Actually, accurate calculations of the
thermodynamic stability of the intermediates is a necessity for approaches
starting at the description of the intermediates at the atomic level.
The calculated enthalpy and Gibbs free energy for the intermediates [396]
are illustrated in Fig. 2.4 and 2.5. One arrives at somewhat different conclusions
on the relative stability of the intermediates depending whether-one bases the
conclusions on the enthalpy or on the Gibbs free energy. As a consequence there
is no simple relation between the heat of formation and the equilibrium concen-
tration.
As the model by Stoltze and N0rskov is based on a description of the
spectroscopic properties of the intermediates, it is straightforward to use data
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 75

Nz(g) .3 Hz(g)
Nz*+3Hz(g)
0r 1 2NH3(g)
0 2N *.3H z(g)
--..E
...., 2NH3*
~
>. -200 I- '\Nz(g)·6H*
2NHz*.2H* -

a.
"0 2NH*+ 4H*
.c
C
w
2N*.6H*
-400 f- -

Fig. 2.4. Calculated enthalpy of the intermediates at 673 K for K-promoted Fe. Reproduced from
[396]

-100

0
E
--..
....,
-" -300
>.
01
Q;
C Nz(g)·6H*
CII
CII
~ -500
III
.0
.0
(5

-700

Fig. 2.5. Calculated Gibbs free energy off the intermediates at 673 K for K-promoted Fe. Repro-
duced from [396]

obtained from quantum-mechanical calculations in this model. This has been

utilized in an investigation of the catalytic activity of metals other than Fe. In
this calculation the trends in chemical binding available from quantum
mechanical calculations were used in an extrapolation using Fe as a reference.
The calculation show, in Fig. 2.6, that the reason why Fe is optimal is the
correlation between the binding of N 2* and the binding of N * Cr and Mn bind
both N 2* and N* more strongly than Fe, and the sticking coefficient for N2 is
close to unity. But the catalytic activity is negligible as N* adsorbs too strongly
on the surface. Co, Ni, and Cu bind N 2* and N* more weekly than Fe, the
coverage by N* is small but the catalytic activity is low as the sticking coefficient
for N2 is extremely low.
76 P. Stoltze

0.30

0.24
'<1
~
C
0
0.18
~
c:
QI
u
c 0.12
0
u Fig. 2.6. Trends in the ammonia produc-
'"
J: tion from 1 m 2 of catalyst as a function of
z 0.06 the number of d-electrons in the substrate.
The reaction conditions are kept constant
I atm, 400 DC, stoichiometric gas). Repro-
0 duced from [724]
4 6 8

2.7.4.3 Coverage by Intermediates

An interesting aspect of the study of the surface chemistry of the catalyst is the
determination of the coverages by intermediates on the surface of the working
catalyst. Attempts have been made to observe the intermediates on the surface of
the working catalyst by various techniques. Laser Raman spectroscopy shows
the following intermediates H* [647], N 2* [473,647], N* [647], NH* [647],
NH2* [647], and NH3* [647].
N 2* has been observed using infra-red spectroscopy [512] and second-
ary-ion mass-spectroscopy [379] on Fe surfaces exposed to N2 + H 2. Argu-
ments in opposition to N 2* being the most abundant reaction intermediate the
weakness of adsorption for this species [650] and the rapid decomposition of
N2H4 to NH3 over Fe [651]
The surface coverages may easily be calculated from numerical models of
ammonia synthesis. From the models [330, 396, 535, 625, 645] it is calculated,
see Fig. 2.7, that N* is the most abundant reaction intermediate at most reaction
conditions. Only at the conditions of vanishing partial pressure ofNH 3 may H*
become the most abundant reaction intermediate [330, 396, 625, 695]. For
a reactor operating at 10.1 MPa, 673 K with no NH3 in the inlet and approxim-
ately 70% of the thermodynamic equilibrium concentration of NH3 at the
outlet, the coverage by * is quite small, approximately 10- 3 and decreasing
through the reactor. The coverage by N 2* is smaller that 10- 5 throughout the
reactor. The coverage by N*, NH* NH2* and NH3* is zero at the inlet and then
quickly increases to approximately 0.80, 0.12, 2· 10- 2 and 4· 10- 4. The coverage
by H* is high, about 0.80, at the inlet and drops to a level of about 0.01 as the
coverage by N* builds up.
The calculations show that there are few free sites and that the surface may
be viewed as a surface nitride with a few vacancies. As the rate of ammonia
synthesis may be expressed as a turnover frequency multiplied by the coverage
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 77

Nli

10- 2 ~ H* -
III

e
Ol

~
0
U
f'-- II
Fig. 2.7. Coverages of intermediates on
10- 4 - - the surface of a catalyst operating at
10.1 MPa, 673 K, The concentration of
NH3 is 0 at the inlet and about 70% of
the thermodynamic equilibrium concen-
tration at the outlet. Reproduced from
-61~2* [645]
10 inlet oullet
Length through catalyst bed

by free sites, and since the calculated rate appears to be correct, it is unlikely that
the calculated coverages should be significantly in error [396].
At conditions where the concentrations of NH3 in the gas phase is signifi-
cant, the coverages by NHx* are high. At these conditions the sequence of
coverages is N* > NH* > NH2* > NH 3*. [396]. This sequence is determined
by the difference in entropy between the intermediates [396].
Evidence for the existence of significant amounts of N * has been found from
interpretation of reaction orders [607,614,617, 652-654J, from a comparison of
work function, electrical resistance and catalyst activity [655J, from laser flu-
orescence [656J, from the observation of N* by electron spectroscopy on the
catalyst [52J or a Fe single crystal [414, 603, 604J after exposure to NH3*
synthesis conditions, and from thermodynamic estimates based on data mea-
sured for the intermediates on single crystal surfaces [550].
Gravimetric measurements of adsorbed N* during NH3* synthesis have
been performed [191, 494, 657]. For a (Fe, AI) catalyst model the coverage by
N* is 0.52-0.69 [191J or 0.11-0.14 [378]. The coverage depends on the operat-
ing conditions [378].
Evidence for the existence of significant amounts of NHx* has been deduced
from interpretations of the reaction orders [512, 614, 658, 659J, laser fluore-
scence [656, 660J, and from the study of the dissociation of NH 3*/Fe(1l0) by
electron spectroscopy [661]. It is a complication in the deduction of the surface
coverages by intermediates that the coverages depend on the operating condi-
tions of the catalyst [396]. These variations may be sufficient to change the
nature of the most abundant intermediate [396]. Experimental evidence has
been found for changes in the nature of the most abundant reaction intermedi-
ates with temperature or promoter concentration [614].
The numerical models of NH3 synthesis predict that the coverages of H* is
quite large if no NH3 is present in the gas phase [396]. At low temperatures and
low conversions a dramatic increase inactivation energy has been observed. The
78 P. Stoltze

variations of the activation enthalpy with the partial pressure of NH3 will be
discussed in details in Sect. 7.6.2.

2.7.4.4 Lifetime of Intermediates

From the numerical models on the synthesis of ammonia the turnover frequency
is readily available. The turnover frequency depends on the operating conditions
[396]. The temperature is a particularly important parameter as the turnover
frequency increases by 5 orders of magnitude from 500 to 1000 Kat 10.1 MPa
and a 28% approach to equilibrium [396].
From the turnover frequency the lifetime of N* may be determined. This
lifetime is about 1 ms at 673 K and 10.1 MPa if no NH3* is present in the gas
phase and about 1 s at thermodynamic equilibrium [396]. The lifetime is
a rapidly decreasing function of temperature and is almost independent of the
pressure [396].
For intermediates other than N*, the model does not contain sufficient
information to calculate the lifetime. However, the model does contain enough
information to calculate an upper limit to the lifetime, see Fig. 8. At 673 K,
10.1 MPa, 10% approach to equilibrium this upper limit to the lifetime is 0.90 s
for N*, 0.18 s for NH*, 24 ms for NH2* and 0.4 ms for NH3* [396].
The estimated lifetime for N 2* is extremely short, around 0.1 ps [396]. This
is further evidence against the associative mechanism.

2.7.5 The Nature of the Rate Limiting Step

One important question about the mechanism of ammonia synthesis is the

nature of the rate limiting step. The question about the nature of the rate

Fig. 2.8. Calculated upper limit to the

1i:i
.0.
lifetime of N-species on the surface at
0. varying conversion. Pressure 10.1 MPa
::J
(solid curve) respective 101 kPa
(dashed curve), N: H ratio 1: 3, tem-
perature 673 K. Reproduced from
10- 4 !---'----+:__'----:::'-:----'-----;~-'-__;::'c:__"'--_::! [396]
o 0.2 0.4 Q6 0.8 1.0
Conversion
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 79

limiting step is closely related to the question whether the mechanism is

associative or dissociative. The single crystal studies leave little doubt that the
mechanism is dissociative that the dissociation of N z* is rate limiting.

2.7.5.1 N z Adsorption as the Rate Limiting Step

For the associative mechanism with N z, adsorption as the rate limiting steps
one finds a kinetic expression of the same form as for the dissociative mechanism
with N z* dissociation being rate limiting. However, this has not prevented the
use of experimental reaction orders for NH3 synthesis [607,662] and for NH3
decomposition [663-666] as arguments for N z adsorption being the rate limit-
ing step.

2.7.5.2 N z Dissociation as the Rate Limiting Step

Evidence for N 2 dissociation as the rate limiting step has been derived from
a large number of experimental studies, such as measurements of isotopic
exchange rates [356, 482, 521, 526, 532, 667], measurements of the isotopic
exchange between NH3 and N* [507, 528, 668], measurements of the
stochiometric number [664] and measurements ofthe rate of dissolution ofN in
bulk Fe from N z and from NH3 [423]. Further evidence has been found by the
agreement between the rates of N2 chemisorption and NH3 synthesis [669], the
agreement between adsorption rates and surface coverages [508], the agreement
between the rate of synthesis for NH3 and the coverage by N* [191], by
int~rpretation of reaction orders for NH3 synthesis [670, 664] and NH3 de-
composition [620], by the value of the activation energy [664] and activation
entropy [597, 668] for NH3 synthesis, and by the agreement between the
calculated density of sites and crystallography [664]. Ultra-high-vacuum single
crystal studies are consistent with Nz(g) + 2* = 2N* being the rate limiting
step for NH3 synthesis [396, 414, 644].
Evidence against the N z dissociation being rate limiting has been postulated
from the higher value of the activation energy for N z dissociation than for NH 3
synthesis [671], from inconsistent rates ofN z chemisorption and NH3 synthesis
[518], from K promoting Nrisotropic exchange but not NH3 synthesis [611],
from the lack of correlation between the rate of N z isotopic exchange and the
catalytic activity [531], and from measurements of the stochiometric number
[672-676]

2.7.5.3. N z Hydrogenation as the Rate Limiting Step

As for the case of N z adsorption as the rate limiting step, the arguments for
hydrogenation as rate limiting step are weak.
Evidence for the hydrogenation of N2 being rate limiting has been found
from interpretations of reaction orders for NH3 synthesis [652,664, 677J and for
NH3 decomposition [678].
80 P. Stoltze

Evidence for the hydrogenation of NHx* being rate limiting has been found
from theoretical considerations [633], from interpretation of reaction orders for
NH3 synthesis [423] and for NH3 decomposition [620, 664, 678-681].

2.7.5.4 Changes in the Rate Limiting Step

Evidence for a shift in the rate limiting step at low H2 partial pressure [508,652],
at high temperatures [664], at low temperatures [663, 666] and far from
equilibrium [652] has been reported. Transients in the rate during NH3 crack-
ing suggest that the rate limiting step for NH3 decomposition is different on iron
surfaces than on iron nitride surfaces [595, 665].
The synthesis of ammonia under transient [682,683] or cyclic [684] opera-
tion has been studied. The rate of the catalytic reaction has been found to exhibit
hysteresis during changes in the H2:NH3 ratio [665] and with changes in
temperature [685]. The occurrence of hysteresis has been assigned to a recrystal-
lisation of the surface [685].
Transients in the rate of ammonia synthesis have been observed when the
flow is changed [418,536,686,687], when the reaction mixture is changed from
N2 + 3H 2 to pure H2 or to pure N2 [536,687,688], and when the temperature
is changed [686]. The behavior under transient operation has been taken as
evidence for auto catalysis [686, 536, 689] or for the existence of parallel
reaction pathways [682, 683]. By modelling the situation of a reactor under
transient operation it has been concluded [690] that limited information can be
derived from such experiments.
A change in reaction order at an extremely high space velocity has been
found [652, 677]. This change was assigned to a change in the rate limiting step
[652, 677]. However, using a numerical model for the kinetics of ammonia
synthesis, Stoltze and N0rskov have found that the measured rates may be
reproduced even with N 2* dissociation as the rate limiting step and that the
unusual reaction orders are found because H* and not N* is the most abundant
reaction intermediate.

2.7.6 The Kinetics of NH3 Synthesis

The kinetics of ammonia synthesis is treated in detail in chapter Q. The

experimental results on the kinetics of ammonia synthesis and decomposition
will be included in the present chapter to the extent that these results illustrate
aspects of the mechanism.
Calculations using a numerical model for the kinetics and mechanism of
ammonia synthesis show that the reaction orders and activation energy in this
model are not constant but show some dependence on the reaction conditions
for the catalyst [396].
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 81

2.7.6.1 Reaction Orders

In the model by Stoltze and N0rskov [396, 645] the reaction orders have
a simple reaction to the surface coverages
aN2 = 1 - 20N2 • (29)
aH2 = 30N• + 20NH• + ONH2' - 0H' (30)
aNH 3 = - 20N • - 20NH• - 20 NH2• - ONH3' (31)
The reaction order for N2 is always very close to 1. Under vanishing partial
pressure of NH3 the reaction order for H2 is aH2 "'" - 1 and the reaction order
for NH3 is aNH 3 "'" O. At these conditions the relation is inhibited by H 2. At
high partial pressures of NH3 the reaction orders may approach their limiting
values aH 2 "'" 3 and aNH 3 "'" - 2 [396, 645]. The calculated reaction orders,
Fig. 2.9, are fairly constant at typical reaction conditions and are in good
agreement with experiments [396, 645].
The reaction orders may easily be calculated if the coverages are known.
However, only in extreme cases may the coverages be calculated from the
reaction orders [396].
At low pressures the synthesis [679] may be treated using pseudo-first-order
kinetics, and the decomposition [675] may be treated using pseudo-second-
order kinetics.
At low pressures the kinetics may be treated using Langmuir-Hinschelwood
kinetics [691-694]. The data for NH3 synthesis indicate that N2 chemisorption
is rate limiting [694] and that N* is the most abundant reaction intermediate
rather than NH* [693]. Deviations from Langmuir-Hinschelwood kinetics have
been explained as the dissolution of N in the bulk phase [691, 692].

",--------------- -----
2{ -
I
~ r'--------------------------~
o
j Or -
)
-2~'····· .......... ··.......... ·............ ·· .... ·...... ··· .... ··· ...... ···· ..- Fig. 2.9. Calculated reaction orders for
N2 (solid curve), H2 (dashed curve),
and NH3 (dotted curve) for NH3 syn-
thesis at 10.1 MPa, N: H ratio 1: 3,
-4~~~~~~L-1~~1~~~1~~~ 673 K. Reproduced from [396]
o 0.2 0.4 0.6 0.8 1.0
Conversion
82 P. Stoltze

However, the model of ammonia synthesis of Stoltze nd N0rskov shows that

even if N* is the most abundant reaction intermediate at typical reaction
condition, N*, NH*, NHz*, H*, and NH3* are all more abundant than *. This
explains why a Langmuir-Hinschelwood expression with only one surface
intermediate is not particularly successful [396, 645].
The Temkin-Pyzhev kinetics [671,695] have been found for NH3 synthesis
[95,617,618,623,624,696-698] and for NH3 decomposition [678,691,692].
The Temkin-Pyzhev kinetics have been interpreted in terms of the existence of
a continous distribution of binding energies on the surface [671, 699-703]. The
reaction orders for NH3 synthesis in the Temkin-Pyzhev kinetics are 1.0 [704,
670] for N z; 0.98 [670], 1.0 [176], 1.8 [704], or 2.l [704] for Hz and - 1.2
[704], - 1.3 [670], or - 1.4 [704] for NH 3. The reaction orders for NH 3
decomposition are 0 [236] for N z - 0.85 [236,662], - 0.64 [678], 1.2 [662] for
Hz and - 0.82 [662], 0.06-0.09 [662], 0.48 [678], 0.6 [236, 662] for NH3
Experimental deviations from the Temkin-Pyzhev kinetics have been as-
signed to pore diffusion [705], to a shift of the rate limiting step at very high
space velocity [677], or to a dependence of binding energy on the dispersion of
Fe [623, 624]
The original derivation of the Temkin-Pyzhev kinetics was later generalized
by Ozaki, Taylor, and Boudart [706]. The Ozaki-Taylor-Boudart kinetics has
been found for NH3 synthesis [613, 614, 617, 618, 620] and for NH3 decomposi-
tion [620, 707]. The Ozaki-Taylor-Boudart kinetics are superior to the Tem-
kin-Pyzhev kinetics [613, 617, 646] for the reproduction of experimental reac-
tion rates over a range of operating conditions. The parameters in the Ozaki-
Taylor-Boudart kinetics has been found to depend somewhat on the tempera-
ture [615] and on the concentration of K [614, 615]

2.7.6.2 Activation Energy

From the model by Stoltze and N0rskov an expression for the activation
enthalpy for ammonia synthesis may be derived [396, 625,645].

Ht = Hl + H~ + 2H 1 8N2 * + 2H 3 8N* + 2H 4 8NH *

+ 2H s 8NH2* + 2H68NH3* + H 7 8H* (32)

where the enthalpies are interpreted as

reaction enthalpy
Nz(g) + *~Nz* Hl
NH3 + *~N* + ~Hz H3
NH3 + *~NH* + Hz H4
NH3 + *~NHz* + 1Hz Hs
NH3 + *~NH3* + H6
H2 + 2*~2H* H7
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 83

and if; is the activation enthalpy for the rate limiting step.
N z* + *~2N* (33)
This expression may be interpreted by noting that Hl + H; is the activation
enthalpy for the dissociative adsorption of N*. The combinations of coverages
and enthalpies'in the expression are all enthalpies for desorption reactions. The
equation thus says that the activation energy for ammonia synthesis equals the
activation enthalpy for the rate limiting step plus the averaged cost of creating
more free sites [396, 625, 645].
The contribution from the rate limiting step is small [396, 625, 645]. At
typical conditions the dominating term originates from creation of free sites by
desorption of N *
(34)
The activation energy is not quite constant, see Fig. 2.10. The calculated
values are in agreement with experimental values, provided the experimental
values used do not span too large a range of operating conditions [396, 625,
645].
At conditions of a vanishing partial pressure of NH 3 , (}N., (}NH. (}NH2" and
(}NH3' will all vanish and the contribution from the reaction

(35)
becomes detectable [396, 625, 645]. This results in a higher activation energy
than under more usual conditions [396, 625, 645].
The reported values for the activation energy for NH3 synthesis are
14 kcaljmole [671], 48 kcaljmole [676], 43.3 kcaljmole [617], 23 kcaljmole
[670], 11.5 kcaljmole [677], 27 kcaljmole [708], 16 kcaljmole [524, 525],
50 kcaljmol [520] and 200 kllmole [696].

Fig. 2.10. Calculated activation enthalpy

- for NH3 synthesis at 10.1 MPa (solid
curve) and 1021 kPa (dashed curve),
N: H ratio 1: 3, temperature 673 K. Re-
O'---'----:"-:---'--::'-:---..1---::-,::-1-,----=-1,-::--'--:-,. produced from [396]
o 0.2 Q4 0.6 0.8 1.0
Conversion
84 P. Stoltze

The rate ofNH 3 synthesis has been measured at 20 atm on single crystals of
Fe [603, 604J in a high pressure microreactor. The activation enthalpy for NH3
synthesis was 19.4 kcaljmole in the absence of K [603, 604J and
18.8 ± 0.5 kcaljmole for KjFe(100) [232J.
The reported values for the activation energy for NH3 decomposition are
0-16.6 kcaljmole [662J, 18.7-26.1 kcaljmol [662J, 31.9 kcaljmole [679J,
44 kcaljmole [236J, and 46 kcaljmole [670]. The value depends on the tempera-
ture [236, 662J and on the bulk composition of the catalyst.

2.7.7 Deuterium Isotope Effect on NH3 Synthesis

The NH3 synthesis is 2.5 [709J, or 3.45 [710J times faster in D2 than in H 2.
The D-isotope effect is identical to the thermodynamic isotope effect if the
most abundant surface intermediate is N* or NH* [693]. This is consistent with
the dissociative mechanisms where H * is not involved in the rate-limiting step.
Based on Laser Raman spectroscopy failing to detect N* or NH*, the
interpretation of the inverse D2 isotope effect as a thermodynamic isotope effect
has been questioned [473].

2.7.8 The Stoichiometric Number for Ammonia Synthesis

The stochiometric number [657, 680, 699, 711J for NH3 synthesis is defined as
the number of turnovers for the rate limiting step, necessary for one turnover of
the total reaction written as N2 + 3H 2 = 2NH 3. If chain reactions are neglect-
ed, the stochiometric number is 1 if the rate limiting step is N z chemisorption
and 2 if the rate limiting step is hydrogenation.
The stochiometric number can be determined experimentally from detailed
rate measurements using a reaction mixture ofN z H2 and NH 3, which is not in
nitrogen isotopic equilibrium.
The stochiometric number for NH3 synthesis is 1 [526,527,615,712, 713J or
2 [672-676]. For NH3 decomposition, the stochiometric number is 1 [527,528J
or 2 [673].
The interpretation of the measurements has caused some polemic [672,674,
677]. The adsorption of NH3 has been found to be important [527, 528].
Neglecting the adsorption of NH3 at high partial pressures could give an
apparent value of 0 for the stochiometric number [527J; this may have been
a problem in some measurements [526].
It has been concluded that some of the earlier measurements were made at
low conversion, where the rate is not inhibited by NH3 and the stochiometric
number is undefined [677]. Other complications are the suggestions that the
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 85

stochiometric number should depend on the presence of K [667J or that some

measurements may have been made after incomplete reduction [714].
From their kinetic model of ammonia synthesis Stoltze and Norskov have
found [396J that even if the rate is proportional to the affinity at equilibrium, the
rate is first order only over a very narrow range of compositions. The usefulness
of assuming proportionality between affinity and rate may be questionable.

2.7.9 Poisoning

A number of compounds including H 2 0, H 2 S and the halogens are strong

poisons for NH3 synthesis. This is reviewed in detail in Sect. D. In the present
section, the literature on poisoning will be reviewed to the extent it illustrates the
mechanism of NH3 synthesis. A common feature of the poisons is that they are
adsorbed at least as strongly on the surface of the catalyst as some of the
intermediates of the reaction.
H 2 0 is a strong poison for NH3 synthesis [194, 715-717]. The loss of
activity during poisoning by H 2 0 is reversible only with mild poisoning [14,18,
718]. The decrease in catalytic activity during poisoning has been found equal to
[719J, or much larger [533J than the expected value due to blockage of sites by
0*.
The poisonous effect of H 2 0 has been included by Stoltze and Norskov in
their numerical models of NH3 synthesis [396, 625J by adding the equation:
(36)
The poisonous effect of H 2 0 is thus simulated by a simple site blocking.
Unfortunately, the binding energy for 0* is not readily available from single
crystal experiments and had to be obtained by fitting experimental data for
ammonia synthesis during partial poisoning by water [396, 625]. The result is
~H = - 117 kJ /mol for reaction 36 [396].
From the model it is found that there is no effect of H 2 0 at a sufficiently low
concentration, see Fig. 2.11. Depending on the reaction conditions, there is
a transition from vanishing coverage by 0* to close to complete coverage.
The reaction orders for the modified model are
aN 2 = 1- 2e N2 * (37)
aH 2 = 3e N * + 2eNH * + eNH2 * - eH * + 2e o* (38)
aNH 3 = - 2e N * - 2e NH * - 2e NH2 * - eNH3 * (39)
aH 2 0 = - 2e o* (40)
At the transition from vanishing coverage of 0* to a significant coverage by
this species, the reaction orders become abnormal as the reaction order for NH3
increases from ~ - 1 to ~ 0, and the reaction order for H 2 0 decreases from
86 P. Stoltze

:g, 0.6
e
~
u 0.4 \
\
\
\ Fig. 2.11. Calculated coverages by N *
\ (solid curve), 0* (dashed curve) and H*
"" (dotted curve) for a catalyst operating at
" ......
......
.... _- 10.1 MPa, N: H ratio 1: 3, 28% conver-
sion, 10 ppm H 2 0. Reproduced from
[396]
900 1000
Temperature (K)

4------------------~----~----~

...................................................
......
......... .
2

----------

-21-------- Fig. 2.12. Calculated reaction orders for

NH3 (solid curve), H 2 0 (dashed curve)
and H2 (dotted curve) for a catalyst oper-
ating at 10.1 MPa, N: H ratio 1 : 3, 673 K,
-4~--~-----l~--~----~~~ 28% conversion. Reproduced from [396]
10- 8 10- 6 10- 4
H20 concentration

'" 0 toward '" - 2 [396J, see Fig. 12. Experimentally, the reaction order for
H 2 0 is - 1.0 [670, 720].
The inclusion of Eq. 36 in the reaction scheme causes an additional term in
the expressions for the activation energy [396, 625].
Ht = Hl + H~ + 2H l 8N2 * + 2H 3 8NH *
(41)
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 87

where the enthalpies are interpreted as

reaction enthalpy
N 2(g) + *;;::=N 2* Hl
NH3 + *;;::=N* + iH2 H3
NH3 + *;;::=NH* + H2 H4
NH3 + *;;::=NH2* + 1H2 Hs
NH3 + *;;::=NH3* + H6
H2 + 2*;;::=2H* H7
H 20 + *;;::=0* + H2 Hs

As before, H; is the activation enthalpy for the rate limiting step

(42)

and the remaining combinations of coverages and enthalpies may be interpreted

as the averaged cost of creating more free sites on the surface.
The activation energy is unchanged as long as the coverage by 0* is
negligible but increases rapidly with increasing coverage by this species, see
Fig. 13. The limiting value under heavy poisoning is close to 200 kJ/mole [396].
Experimentally, the activation energy is unchanged or increased [718].
H 2S is very poisonous for NH3 synthesis [14, 18, 721, 722] and N2 adsorp-
tion [513]. The poisoning is irreversible [14, 590]. The poisonous effect is due to
S* surface blockage [590, 723]. The activation energy of ammonia synthesis is
unchanged during H 2S poisoning [723] After exposure to large partial pressures
of H 2S, FeS may be observed by X-ray powder diffraction [590].

250.----,----,----,----,----,

(5
.§
-,
200
/
,-
,- .... - --
.>t! I
I
I
11: 150 I
"5
.l:
I
/
1:
(l> I
c: 100 I
o /
/
Fig.2.13. Calculated activation enthalpy
~ 50 ----- .....
"."'/ for NH3 synthesis for a partially poi-
soned catalyst operating at 10.1 MPli
~ (solid curve) resp. 101 kPa (dashed
curve), N: H ratio 1 : 3, 673 K, 28 % con-
version. Reproduced from [396]

H20 concentration
88 P. Stoltze

2.8 References

1. Mittasch A (1950) Adv Catal 2: 81

2. Mittasch A (1951) Geschichte der Ammoniak Synthese Verlag Chemie Weinheim
3. Nielsen A (1953) Adv Catal 5: 1
4. Bokhoven C, Van Herden C, Westrik R, Zwietering P (1955) In: Emmett P (ed) Catalysis, Vol 3
p 265
5. Frankenburg W (1955) In: Emmett PH (ed) Catalysis, Vol 3 p 171
6. Vancini CA (1971) Synthesis of Ammonia. McMillan
7. Emmett PH (1975) In: Dragulis E, Jaffee RI (eds) The Physical Basis for Heterogeneous
catalysis. Plenum
8. Ozaki A, Aika K (1979) In: Hardy RWF, Bottomley F, Burns RC (eds) A Treatise on
Dinitrogen Fixation, Wiley
9. Ozaki A, Aika K (1981) In: Anderson JR Boudart M (eds) Catalysis Science Technology, Vol 1
p 87
10. Boudart M (1981) Catal Rev Sci Eng 23: 1
11. Nielsen A (1981) Catal Rev Sci Eng 23: 17
12. Strel'tsov OA (1967) Kinet Katal 3: 140
13. Emmett PH (1940) 12th Report of the Committe on Catalysis. National Research Council
14. Nielsen A (1968) An Investigation on Promoted Iron Catalysts for the synthesis of Ammonia.
Jul Gjelleup Copenhagen, 3rd edition
15. Nielsen A (1950) An Investigation on Promoted Iron Catalysts for the Synthesis of Ammonia.
Jul Gjelleup Copenhagen
16. Nielsen A (1970) Catal Rev 4: 1
17. Nielsen A (1977) Fert Sci Technol Ser 2: 87
18. Nielson A (1986) Chern Age India 37: 267
19. Ert! G (1980) Catal Rev Sci Eng 21: 201
20. Ertl G (1982) CRC Crit Rev Solid State Mater Sci 10: 349
21. Ertl G (1983) In: Anderson JR, Boudart M (eds) Catalysis Science Technology, Vol 4 p 209
22. Strasser G, Grunze M, Golze M (1985) J Vac Sci Technol A 3: 1562
23. Ert! G (1980) NATO Adv Study Inst Ser Ser E 39: 271
. 24. Ertl G (1983) J Vac Sci Technol A 1:1247
25. Ertl G (1989) Stud Surf Sci Catal 44: 315
26. Chen H-C, Anderson RB (1973) J Catal 28: 161
27. Clausen BS, Morup S, Tops0e H, Candia R, Jensen EJ, Baranski A, Pattek A (1976) J Phys
Colloq: 245
28. Malycheva TYa, Rabina PD, Kuznetsov LD (1968) Probl Kinet Katal Akad Nauk SSSR 12:
185
29. Yamaguchi S (1951) J Phys Colloid Chern 55: 1409
30. Aleksic BD, Terlecki-Baricevic A (1973) Rev Roum Chim 18: 575
31. Karibdazhanyan NA, Simulina lA, Lachinov SS, Vorotilina ZI, Mishchenko ShSh (1980)
Khim Prom-st 294
32. Dimitrov M, Slavov S (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 26
33. Slavov S, Dimitrov M (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 19
34. Dimitrov M Slavov S (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 11
35. Dimitrov 1M, Slavov S (1980) Gd Vissh Khim -Tekhnol Inst Sofia 26: 3
36. Sasaki N, Osumi Y (1975) J Chern Soc Japan 73: 808
37. Uchida H, Todo N (1954) Bull Chern Soc Japan, 27: 585
38. Res Group Mossbauer Spectrosc Nanking Nan-ching (1978) Ta Hsueh Hsueh Pao Tzu Jan
K'o Hsueh 48
39. Garbassi F, Fagherassi G, Calcaterra M (1972) J Catal 26: 338
40. Rabina PD, Malysheva TYa, Kuznetsov LD, Batyrev VA (1970) Kinet Katal 11: 1243
41. Saprykina TV, Rabina PD, Chudnilov MG, Alekseev AM, Kuznetsov LD (1976) Kinet Katal
17: 723
42. Norval GW, Phillips MJ (1986) J Phys Chern 90: 4743
43. Peev T, Visokov G, Czako-Nagy I, Vertes A (1985) Appl Catal 19: 301
44. Maksimov YuV, Dumesic DA, Suzdalev IP, Matveev AI (1977) Kinet Catal 18: 499
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 89

45. Peev T, Krylova AV, Bozhinova A (1981) Radiochern Radioanal Lett 47: 307
46. Bleskin 01, Lachinov SS, Mirkin AE (1984) Kinet Katal25: 702
47. Topsoe H, Durnesic JA, Boudart M (1974) J Phys (Paris) 35: C6 411
48. Lachinov SS, Rubinshtein AM, Akirnov VM, Klyachov-Gurvich AL, Konyukhova IN,
Kuznetsov LD, Levitskaya TT, Pribytkova NA, Slink in AA, Chesnokova RA (1964) Kinet
Katal 5: 478
49. Lachinov SS, Torocheshnikov NS, Sirnulina lA, Lyudkovskaya BG, Rakrnat-Zade AG,
Ugnachev VI, Alipur G, Sushcheva AE (1975) Tr Mosk Khirn-Tekhnol Inst 85: 20
50. Ludwiczek H, Preizinger A, Fischer A, Hosernann R, Schoenfeld A, Vogel W (1978) J Catal51:
326
51. Michel A, Pouillard E (1948) Cornptes Rend 227: 194
52. Ertl G, Thiele N (1919) Appl Surf Sci 3: 99
53. Peters C, Schafer K, Krabetz R (1960) Z Elektrochern 64: 1194
54. Wi1chinsky ZW (1949) Anal Chern 21: 1188
55. Westrik R (1953) J Chern Phys 21: 2094
56. Dry ME, Ferreira (1967) J Catal 7: 352
57. Drnitrov M (1979) Geterog Katal 4: 349
58. Klisurski DG, Mitov IG, Petrov KP (1980) Therrnochirn Acta 41: 181
59. Tornov T, Klissurski D, Mitov I (1982) Phys Stat Sol (a) 73: 249
60. Tricker MJ, Vaishnava PP, Whan DA (1982) Appl Catal 3: 283
61. Glodeanu F, Spinzi A, Nicolaescu IV, Galatchi G, Spinzi M (1979) Rev Rourn Phys 24: 161
62. Uchida H, Todo N (1955) Repts Govt Chern Ind Research Inst Tokyo 50: 23
63. Chen HC, Anderson RB (1972) J Coll Interfacial Sci 38: 535
64. Spinzi A, Galatchi G, Spinzi M (1979) Geterog Katal 4: 439
65. Rozin AT, Kornarov VS, Efros MD, Lerneshonok GS (1982) Vestsi Akad Navuk BSSR Ser
Khirn Navuk 31
66. Uchida H, Todo N (1956) Bull Chern Soc Japan 29: 20
67. Krabetz R, Peters C (1965) Angew Chern 77: 333
68. Krabetz R, Peters C (1963) Ber Bunsen-Ges Phys Chern 67: 390
69. Egyhazi T, Scholtz J Beskov VS (1984) React Kinet Catal Lett 24: 1
70. Kornarov VS, Efros MD, Rozin AT, Lerneshonok GS, Erernenko SI (1980) Dokl Akad Nauk
BSSR 24: 1098
71. Rajararn RR, Sermon PA (1985) J Chern Soc Faraday Trans 1 81: 2577
72. Rajararn RR, Sermon PA (1985) J Chern Soc Faraday Trans 1 81: 2593
73. Brown R, Cooper ME, Whan DA (1982) Appl Catal3: 177
74. Youssef AM, Ami NM (1978) Surface Technology 7: 469
75. Marakhovets LN, Sirnulina lA, Lachinov SS, Sobolevskii VS, Lytkin VP, Christozvonov DB
(1972) ObI Neorg Tekhnol 3
76. Chernysheva LA, Tovbin MV, Zabuga VYa, Efirnova NI (1980) Kinet Katal 18: 25
77. Tirnofeev VA, Lachinov SS, Torocheshnikov NS, Sirnulina lA, Lyudkovshaya BG, Rudnitskii
LA, Chudinov MG (1973) Tr Mosk Khirn-tekhnol Inst 738: 33
78. Baranski A, Bielanski A, Blasiak E, Dulski R, Rokosz A (1968) Chern Stosow Ser A 12: 45
79. Bielanski A, Baranski A, Musial U, Rokosz A (1967) Chern Stosow Ser A 11: 365
80. Jensen EJ, Topsoe H, Sorensen 0, Kragh F, Candia R, Clausen BS, Morup S (1977) Sc J Metall
6
81. Lachinov SS, Torocheshnikov NS, Sirnulina lA, Rakhrnat-Zade AG, Ugnachev VI, Alipur G
Garanina EF (1975) Tr Mosk Khirn- Tekhnol Inst 85: 22
82. Ert! G, Prigge D, Schlagl R, Weiss M (1983) J Catal 79: 359
83. Patyi L, Tsarev VI, Krylova AV, Oravets D, Farkas Z, Torocheshnikov NS (1979) React Kine!
Catal Lett 12: 165
84. Maxwell LR, Smart JS, Brunauer S (1951) J Chern Phys 19: 30
85. Maxwell LR, Brunauer S (1949) Phys Rev 76: 175
86. Pererrnan NI, Zubova IE, Rabina PD, Kuznetsov LD, Pavlov a NZ (1978) Tr Mosk Khirn-
Tekhnol Inst irn.D I Mendeleeva 99: 50
87. Pele L (1983) Rev Chirn (Bucharest) 34: 168
88. Aleksic BD, Terelecki-Baricevic A (1973) Bull Chern Soc Beograd 38: 447
89. Medeleanu V, Doca N, Stefanescu M, Bibolaru A (1979) Rev Chirn (Bucharest), 30: 751
90. Tovbin MV, Kuznetsov VA, Popovich ZP (1973) Kinet Katal 10: 55
91. Tovbin MV, Popovich ZP (1969) Ukr Khirn Zh 35: 787
90 P. Stoltze

92. Pennock GM, Flower HM, Andrew SPS (1987) J Catall03: 1

93. Bardik ZN, Loza AN, Rusov MT, Strel'tsov OA (1966) Khim Prom-st (Moscow) 42: 351
94. Tang Ren-Yuan, Zhang Su, Wang Chengyu, Liang Dongbai, Lin Liwu (1987) J Catal106: 440
95. Strel'tsov OA, Antonyuk LF, Kuznetsova EP (1967) Kinet Katal 3: 155
96. Chistozvonov DB, Kirillov IP, Yagodkina GN, Dantsig GA, Lytkin VP, Marakhovets LN
(1974) Izu Vyssh Ucheb Zaved Khim Khim Tekhnol 17: 870
97. Baranski A, Lagan JM, Pattek A, Reizer A (1982) Appl Catal 3: 207
98. Reizer A, Baranski A (1984) Appl Catal 9: 343
99. Baranski A, Lagan M, Pattek A, Reizer A (1980) Arch Hutn 25: 143
100. Strel'tsov OA, Rusov MA, Kokhar LA, Lonza AN (1960) Kinet Katal 1: 597
101. Dvornik OS, Strel'tsov OA, Artyukh YuN (1977) Ukr Khim Zh (Russ ed) 43: 375
102. Strel'tsov OA, Dvornik OS, Lytkin VP (1977) Izv Vyssh Uchebn Zaved Khim Khim Tekhnol,
20: 1662
103. Astakhov NN, Alekseev AM, Minaev DM, Rabina PD, Perel'man PD, Kuznetsov LD (1979)
Ktim Prom-st Ser Azotn Prom-st, p 36
104. Pattek-Janczyk A, Hrynkiewicz AZ, Kraczka J, Kulgawczuk D (1983) Appl Catal 6: 35
105. Pattek-Janczyk A, Baranski A, Kotarba A, Kowalska A, Miczko B, Pyrczak E, Reizer A,
Reubenbauer K, Sepiol B, Spiewak Z (1988) Appl Catal 39: 169
106. Visokov G, Ivanov D (1976) Zh Prikl Khim (Leningrad) 49: 1001
107. Minarev DM, Astakhov NN (1980) Deposited doc no SPSTL 575khp-D80
108. Kock AJHM, Fortuin HM, Geus JW (1985) J Catal 96: 261
109. Mosesman MA (1951) J Am Chern Soc, 73: 5635
110. Popovich ZP, Tovbin MV (1970) Ukr Khim Zh 36: 726
111. Klisurski D, Mitov I (1979) J Phys Colloq 353
112. Topsooe H, Dumesic JA, Boudart M (1973) J Catal 28: 477
113. Hall WK, Tarn H, Anderson RB (1950) J Am Chern Soc 72: 5436
114. Mahapatra H, Ghorai DK, Ganguli NC, Sen SP (1978) Fert Technol 15: 226
115. Baranski A, Lagan M, Pattek A, Reizer A, Christiansen LJ, Topsoe H (1979) Stud Surf Sci Catal
3: 353
116. McCartney JT, Anderson RB (1951) J Appl Phys 22: 1441
117. Uvarova IV, Rusov MT, Sarnchenko NP (1969) Dopov Akad Nauk Ukr RSR Ser B, 31: 439
118. Uvarova IV, Rusov MT, Samchenko NP (1969) Kinet Katal 10: 558
119. Uvarova IV, Rusov MT, Samchenko NP (1969) Porosh Met 9: 51
120. Zhang H-B, Schrader GL (1985) J Catal 95: 325
121. Uvarova IV, Rusov MT, Samchenko NP (1969) Zh Fiz Khim 43: 1423
122. Baranski A, Pattek A, Reizer A (1978) Bull Acad Pol Sci Ser Chim 26: 353
123. Smith PJ, Taylor DW, Dowden DA, Kemball C, Whan DA (1982) Appl Catal 3: 303
124. Baranski A, Fulinski A, Pattek A, Reizer A (1976) Bull Acad Pol Sci Ser Sci Chim 24: 729
125. Baranski A, Bielanski A, Pattek A (1972) J Catal 26: 286
126. Baranski A, Hajduk J (1985) Chemik, 38: 228
127. Pattek-Janczyk A, Hrynkiewicz AZ (1983) Appl Catal 6: 27
128. Pattek-Janczyk A (1988) Wiss Z -Friderich-Schiller-Univ Jena: Naturwiss Reihe, 37: 835
129. Baranski A, Lagan JM, Pattek A, Reizer A (1980) React Kinet Catal Lett 15: 285
130. Yatsimirskii VK, Girenkova NI, Mel'nik PM, Zabyga VYa (1978) Ukr Khim Zh (Russ Ed.)
44: 239
131. Baranski A, Lagan JM, Pettek A, Reizer A (1982) Appl Catal 3: 201
132. Baranski A, Reizer A, Kotarba A, Pyrczak E (1988) Appl Catal 40: 67
133. Ivanov DG, Visokov GP (1972) Zh Prikl Khim (Leningrad) 45: 1456
134. Vink TJ, der Kinderen JM, Gijzeman OLJ, Geus JW (1986) Appl Surf Sci 26: 367
135. Baranski A, Reizer A, Kotarba A, Pyrczak E (1988) Appl Catal 40: 67
136. Baranski A, Reizer A, Kotarba A, Pyrczak E (1985) Appl Catal19: 417
137. Perrichon V, Charcosset H, Barrault J, Forquy C (1983) Appl Catal 7: 21
138. Klisurski D, Mitov I, Tomov T (1979) Geterog Katal 4: 207
139. Klisurski D, Mitov I, Tomov T (1983) Stud Surf Sci Catal 16: 421
140. Zakieva KZ, Rabina PD, Zubova IE, Khaustova LE, Pavlov a NZ, Kuznetsov LD (1973) Tr
Mosk Khim-Tekhnol Inst 73: 120
141. Simulina lA, Lyudkovskaya BG (1969) Tr Mosk Khim-Tekhnol Inst 60: 86
142. Lachinov SS, Bleskin DI, Gandilyan SS (1980) Deposited Doc VINITI 3955
143. Tovbin MV, Zabuga VYa, Chernysheva LA (1977) 12th Tezisy Dokl.- UKr Resp Konf Fiz
Khim 162
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 91

144. Lemeshko ND, Zabuga VYa, Chernysheva LA (1983) Ukr Khim Zh (Russ Ed) 49: 122
145. Tikhonova ON, Zubova IE, Pavlova NZ, Ivanova RF, Lyubchenko YuA, Dmitrenko LM
(1973) Tr Mask Khim -Tekhnol Inst 73: 126
146. Aleksic BD, Klisurski D, Mitov I (1981) Izu Khim 13: 660
147. Aleksic B, Bogdanov S (1985) Thermochim Acta 93
148. Mitov I, Klisurski D, Aleksic B, Gyurova L, Nikolov 0 (1984) Izu Khim 17: 305
149. Wachs IE, Dwyer DJ, Iglesia E (1984) Appl Catal 12: 201
150. Lachinov SS, Krylova AV, Pavlova NZ, Zubova IE (1978) Tr Mask Khim Tekhnol Inst im D I
Mendeleeva 99: 47
151. Lachinov SS, Simulina lA, Tikhonova ON,. Torocheshnikov NS, Alipur G, Strekalova NM
(1975) Deposited Doc 3770
152. Yatsimirskii VK, Kovalenko VN, Ishchenko EV (1982) Ukr Khim Zh (Russ Ed.) 48: 614
153. Peev T (1985) Khim Ind (Sofia) 57: 250
154. Borghard WS, Boudart M (1983) J Catal 80: 194
155. Logan SR, Moss RL, Kemball C (1958) Trans Faraday Soc 54: 922
156. Pavlova NZ, Rogozhina SA, Kuznetsov DA, Zubova IE, Malysheva TYa (1968) Kinet Katal
9: 1390
157. Rogozhina SA, Lachinov SS, Pavlova NZ, Zubova IE, Kuznetsov DA (1966) Tr Mask Khim-
Tekhnol Inst 51: 173
158. Niemann W, Clausen BS, Topsoe H (1987) NATO ASI Ser B 158: 909
159. Niemann W, Clausen BS, Topsoe H (1987) Ber Bunsen-Ges Phys Chern 91: 1292
160. Fierro JLG, Horns N, Ramirez P della Piscina, Sueiras JE (1983) Z Phys Chern (Wiesbaden)
135: 235
161. Rayment T, Schlagl R, Thomas JM, Ert! G (1985) Nature (London) 315: 311
162. Silverman DC, Boudart M (1982) J Catal 77: 208
163. Emmett PH, Brunauer S (1934) J Am Chern Soc 56: 35
164. Nielsen A, Bohlbro H (1952) J Am Chern Soc 74: 963
165. Fagherazzi G, Galante F, Garbassi F, Pernicone N (1972) J Catal 26: 344
166. Hosemann R, Hentschel MP (1986) Yak -Tech, 35: 3
167. Uchida H, Terao I, Ogawa K (1964) Bull Chern Soc (Japan) 37: 653
168. Rudnitskii LA, Demina GN, Makurina NA, Dmitrenko LM, Mazus EI, Alekseev AM (1984)
React Kinet Catal Lett 25: 297
169, Westrik R, Zwietering P (1953) Proc K Ned Akad Wet 56: 492
170. Schafer K (1960) Z Elektrochem 64: 1191
171. Kolbl H, Schottle E (1961) Z Elecktrochem 65: 91
172. Zwietering P, Koks HTL (1954) Nature, 173: 683
173. Fischer A, Hosemann R, Vogel W, Kouteckly J, Pohl J, Ralek M (1981) Stud Surf Sci Catal 7:
341
174. Herbstein FH, Smuts J (1963) J Catal 2: 69
175. Hosemann R, Preizinger A, Vogel W (1966) Ber Bunsen-Ges Phys Chern 70: 796
176. Tamaru K (1964) Bull Chern Soc Japan 39: 771
177. Brunauer S, Emmett PH (1935) J Am Chern Soc 57: 1754
178. Emmett PH, Brunauer S (1937) Trans Electrochem Soc 71: 383
179. Topsoe H, Topsoe N, Bohlbro H, Dumesic JA (1980) Proc 7th Intern Congr Catal (Tokyo) 247
180. Artyukh YuN, Fedun OS, Zyuzya LA (1980) Kinet Katal18: 20
181. Dvornik OS, Kozub GM, StreJ'tsov OA (1977) Teor Eksp Khim 13: 546
182. Dvornik OS, StreJ'stov OA, Lyubchenko YuA (1975) Dokl Akad Nauk SSSR 221: 648
183. Rabina PD, PereJ'man ND, Znbova IE, Kuznetsov LD, Maravskaya GK (1977) Ukr Khim Zh
(Russ ed) 47: 83
184. Dvornik OS, Strel'tsov OA, Lytkin VP (1975) Ukr Khim Zh (Russ ed) 41: 428
185. Dvornik OS, Strel'tsov OA, Maiboroda VP (1975) Ukr Khim Zh (Russ ed) 41: 1329
186. Dvornik OS, Strel'tsov OA, Chernobrivets VL (1975) Ukr Khim Zh (Russ ed) 41: 544
187. Limin Li, Wenxiang Wang (1983) Gaodeng Xuexiao Huaxue Xuebao 4: 763
188. Strel'tsov OA, Fedun OS, Artyukh YuN, Lytkin VP (1977) 12th Tezisy Dokl Ukr Resp Konf
Fiz Khim 159
189. Dvornik OS, Strel'tsov OA, Lytkin VP (1975) Dopov Akad Nauk Ukr RSR Ser B 623
190. Bridger, Pole, Beinlich, Tomson (1947) Chern Eng Progr 43: 291
191. Scholten JJF, Konvalinka JA, Zwietering P (1960) Trans Faraday Soc 56: 262
192. Yatsimirskii VK, Vyaz'mitina OM, Kozlova TP (1971) Teor Eksp Khim 7: 645
193. Podgurski HH, Emmett PH (1953) J Phys Chern 57: 159
92 P. Stoltze

194. Artyukh YuN, Yas'mo VI, Loza AN (1973) Kinet Katal 14: 1599
195. Solbakken V, Solbakken A, Emmett PH (1969) J Catal 15: 90
196. Kuznetsov LD, Rabina PD, Dmitrenko LM, Mischenko ShSh, Zozulya VYu (1979) Khim
Prom-st Ser Azotn Prom-st 19
197. Schultz JM (1972) J Catal 27: 64
198. Emmett PH, Brunauer S (1937) J Am Chern Soc 59: 1553
199. Dry ME, du Pleissis JAK, Leuteritz GM (1966) J Catal 6: 194
200. Dimitrov M, Rusev R (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 120
201. Rabina PD, Kuznetsov LD, Anisimova MI (1969) Khim Prom 45: 350
202. Rozin AT, Komarov VS, Efros MD, Lemeshonok GS (1980) Vestsi Akad Navuk BSSR Ser
Khim Navuk 27
203. Lachinov SS, Simulina lA, Alipur GA, Nefedova NV (1978) Tr Mosk Khim-Tekhnol Inst im D
I Mendeleeva 99: 36
204. Kharchenko EV, Tovbin MV (1970) Khim Prom Ukr 15
205. Yatsimirskii VK, Girenkova NI (1979) Kinet Katal 20: 168
206. Yatsimirskii VK, Tovbin MV, Girenkova NI (1977) 2nd Tezisy Dokl Vses Simp Akt Poverkhn
Tverd Tel 22
207. Brunauer S, Emmett PH (1940) J Am Chern Soc 62: 1732
208. Chudinov MG, Perov VM, Ksenzenko VI, Alekseev AM (1986) Khim Prom-st (Moscow) 92
209. Weiss M, Ert! G (1982) Stud Surf Sci Catal 11: 277
210. Hanji K, Shimizu H, Shindo H, Onishi T, Tamaru K (1980) J Res Inst Catal Hokkaido Univ
28: 175
211. Hanji K, Shimizu H, Shindo H, Onishi T, Tamaru K (1980) J Res Inst Catal Hokkaido Univ 28:
175
212. Pernicone N, Fagherazzi G, Galante F, Garbassi F, Lazzerin F, Mattera A (1972) Proc 5th
Intern Congr Catal 2: 1241
213. Artyukh YuN, Boldyreva NA, Rusov MT (1970) Kinet Katal 11: 1531
214. Sasa Y, Uda M, Toyoshima I (1986) J Mater Sci Lett 5: 470
215. Dmitrov M (1980) Khim Ind (Sofia) 449
216. Chinh Chuong Zang, Krylova AV, Klimova GN, Lachinov SS, Torocheshnikov NS (1975) Tr
Mosk Khim-Tekhnol Inst 85: 24
217. Ivanov MM, Rudnitskii LA, Rabina PD, Kuznetsov LD (1968) Kinet Katal 9: 1239
218. Poniewierski Z, Tchorzewski T (1975) Chern Stosow 19: 161
219. Alekseeva MP, Lachinov SS, Predoiu C, Totocheshnikov NS (1968) Probl Kinet Katal Akad
Nauk SSSR 12: 175
220. Emmett PH, Brunauer S (1937) J Am Chern Soc 59: 310
221. Strongin DR, Somorjai GA (1988) Catal Lett 1: 61
222. Rudnitskii LA, Ivanov MM (1970) Kinet Katal 11: 207
223. Artyukh YuN, Bondarenko RN, Golovatyio VG, Kozub GM (1975) Dopov Akad Nauk Urk
RSR Ser B 121
224. Pennock GM, Flower HM (1984) Conf Ser - Inst Phys 68: 263
225. Van Ommen JG, Bolink WJ, Prasad J, Mars P (1975) J Catal 38: 120
226. Bonzel HP, Broden G, Krebs HJ (1983) Appl Surf Sci 16: 373
227. Bare SR, Strongin DR, Somorjai GA (1986) J Phys Chern 90: 4726
228. Broden G, Bonzel HP (1979) Surf Sci 84: 106
229. Ertl G, Weiss M, Lee SB (1979) Chern Phys Lett 60: 391
230. Seip U, Bassignana IC, Kiippers J, Ert! G (1960) Surf Sci 160: 400
231. Whitman LJ, Bartosch CE, Ho W (1986) J Chern Phys 85: 3688
232. Strongin DR, Somorjai GA (1988) J Catal 109: 51
233. Strongin DR, Bare SR, Somorjai GA (1987) J Catal 103: 289
234. Connell G, Dumesic JA (1985) J Catal 92: 17
235. Berengarten MG, Abdukadyrova SA, Zuvoba IE, Rabina PD, Rudnitskii LA, Pavlova NZ
(1973) Tr Mosk Khim-Tekhnol Inst 72: 14
236. Love K, Brunauer S (1942) J Am Chern Soc 64: 745
237. Rudnitskii LA, Ivanov MM (1969) Dokl Akad Nauk SSSR 104: 139
238. Rudnitskii LA, Ivanov MM (1969) Dokl Akad Nauk SSSR 184: 886
239. Abdukadyrova SA, Rabina PD, Zubova IE, Lyudkovskaya B (1970) Tr Mosk Khim- Teknol
Inst 67: 148
240. Kuznetsov LD, Rabina PD, Lopukhov GA (1971) Tr Nauch-Issled Proekt Inst Azoth Prom
Prod Org Sin 11: 92
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 93

241. Sasa Y, Uda M, Toyoshima I (1982) Chern Lett 2011

242. Enikeev EKh, Krylova AV, Kuznetsov LD, Lachinov SS, Roginskii SZ (1960) Dokl Akad
Nauk SSSR 131: 1126
243. Rudnitskii LA, Ivanov MM (1969) Kinet Katal 10: 349
244. Krylova EV, Kuznetsov LD, Konyukhova IN (1964) Kinet Catal 5: 948
245. Broden G, Gafner G, Bonzel HP (1979) Surf Sci 84: 295
246. Rao CNR, Rao G, Ranga PK (1987) Chern Phys Lett 134: 47
247. Rozin AT, Komarov VS, Efros MD, Lemeshonok GS, Eremenko SI (1980) Vestsi Akad Navuk
BSSR Ser Khim Navuk 35
248. Dimitrov M, Rusev R (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 111
249. Dimitrov M, Rusev R (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 126
250. Dimitrov M, Rusev R (1980) God Vissh Khim -Tekhnol Inst Sofia 26: 103
251. Tikhonova ON, Zubova IE, Ivanova RF, Lyubchenko YuA, Dmitrenko LM, Mishchenko
ShSh (1973) Tr Mosk Khim- Tekhnol Inst 73: 123
252. Berengarten MG, Rudnitskii LA, Zubova IE, Alekseev AM, Dmitrenko LM, Ivanova RF,
Mischenko ShSh, Lyubchenko YuA, Tikhonova ON (1974) Kinet Katal 15: 250
253. Zakieva KZ, Rabina PD, Zubova IE, Batanova AM, Kuznetsov LD (1973) Tr Mosk Khim-
Tekhnol Inst 73: 117
254. Tikhonova ON, Zubova IE, Pavlova NZ, Ivanova RF, Lyubchenko YuA, Dmitrenko LM
(1973) Tr Mosk Khim-Tekhnol Inst 73: 126
255. Takeshita T, Wallace WE, Craig RS (1976) J Catal44: 236
256. Wallace WE, France J, Shamsi A (1982) Rare Earths Mod Sci Technol 3: 561
257. Komarov VS, Dmitrenko LM, Efros MD, Ivanova RF, Karpinchik EV, Rabina PD, Leme-
shonok GS, Mantseva GM (1977) Vestsi Akad Navuk BSSR Ser Khim Navuk 9
258. Komarov VS, Rabina PS, Efros MD, Kuznetsov LD, Rozin AT, Dmitrenko LM, Lemeshonok
GS, Solov'eva AD (1979) Vestsi Akad Navuk BSSR Ser Khim Navuk 26
259. Kiss G (1983) Petrochemia 23: 17
260. Komarov VS, Rozin AT (1977) Dokl Akad Nauk BSSR 21: 1102
261. Lachnikov SS, B1eskin 01, Smirnova SA (1980) Deposited doc no VINITI 1014
262. Zakieva KZ, Rabina PD, Abdukadyrova SA, Zubova IE, Khaustova LE, Pavlova NZ (1973)
Tr Mosk Khim-tekhnol Inst 72: 20
263. Sultanbekov GK, Kondrat'eva NM, Lachinov SS, Torochesnikov NS, Saltanova VP (1973) Tr
Mosk Khim-Tekhnol Inst 72: 12
264. Sergeeva AN, Dovgei VV, Pavlenko LT, Zubritskaya DI, Tkachenko ZhI, Okoskaya AP,
Lyubchenko YuA (1983) Kinet Katal 24: 1473
265. Simulina lA, Lachinov SS, Torocheshnikov NS, Sobolovskii VS, Christozvonov DB, Lytkin
VP (1969) Tr Mosk Khim- Tekhnol Inst 60: 91
266. Chernysheva LA, Zabuga VYa, Pivovarova NS, Tovbin MV (1981) Visn Kiiv Univ (Ser.):
Khim 22: 59
267. Zabuga VYa, Tovbin MV, Chernysheva LA, Drel II, Efimova NI (1975) Ukr Khim Zh (Russ
ed) 41: 818
268. Maksimov YuV, Arents R, Suzdalev IP, Yatsimirskii VK (1980) J Phys Colloq 337
269. Maksinov YuV, Arents RA, Suzdalev IP, Yatsimirskii VK, Chernysheva LA (1980) Kinet Katal
21: 1555
270. Komarov VS, Rozin AT, Lemeshonok GS, Eremenko SI (1985) Vestsi Akad Navuk BSSR Ser
Khim Navuk 30
271. Rudnitskii LA, Rabina PD, Melik'yan AA, Korneev VP, Kuznetsov LD, Saprykina TV,
Alekseev AM (1976) Kinet Katal 17: 989
272. Rudnitskii LA, Rabina PD, Kuznetsov LD, Alekseev AM, Saprykina TV (1976) Dokl Akad
Nauk SSSR 227: 919
273. Aleksic B, Jovanovic N, Terlecki-Baricevic A (1978) 3rd Tr Mezhdunar Simp Geterogennyi
Katal p 147
274. Badik VS, Lyubchenko YuA, Sergeeva AN, Dmitrenko LM (1974) Zhur Prikl Khim 47: 2180
275. Badik VS, Lyubchenko YuA (1976) Zhur Prikl Khim 49: 876
276. Grisnaeva MP, Krylova AV, Marakhovets LN, Kondrat'eva NM, Nazarova IG, Lachinov SS,
Torocheshnikov NS (1972) Tr Mosk Khim- Tekhnol Inst 72: 6
277. Karaslavova K, Anastasov M (1978) 3rd Tr Mezhdunar Simp Geterogenyi Katal 297
278. Krylova AV, Dimitrov M, Karaslavova K, Klimova GN, Lachinov SS, Torocheshnikov NS
(1976) Deposited Doc VINITI 1117
279. Kagan TYu, Krylova AV (1968) Kinet Katal9: 440
94 P. Stoltze

280. Dimitrov M, Karaslavova K (1975) Khim Ind (Sofia) 47: 19

281. Komarov VS, Efros MD, Lemeshonok GS, Rozin AT (1978) Vesti Akad, Navuk BSSR Ser
Khim Navuk 15
282. Artyukh YuN, Zyuzya LA (1980) Kine! Katal 18: 23
283. Artyukh YuN (1977) 12th Tzeisy Dokl - Ukr Resp Konf Fiz Khim p 101
284. Artyukh YuN, Chernobrivets VL, Kuznetasova EP, Lakoza EL, Zyuzya LA (1977) Kine!
Katal 18: 1614
285. Tovbin MV, Vyaz'mitina OM, Silkina SS (1967) Silk ina SS Khim Prom-st Ukr 12
286. Huang KH (1985) Appl Catal 15: 175
287. Uebing C, Viefhaus H, Grabke HJ (1988) Appl Surf Sci 32: 363
288. Ozaki A, Aika K, Morikawa Y (1972) Proc 5th Intern Congr Catal 2: 1251
289. Bleskin or, Lachinov SS (1980) Deposited Doc VINITI 1015
290. Bosch H, van Ommen JG, Gellings PJ (1985) Appl Catal 18: 405
291. Berengarten MG, Rudnitskii LA, Rabina PD, Kuznetsov LD. Zubova IE, Alekseev AM,
Zakieva KZ (1974) Dokl Akad Nauk SSSR 214: 601
292. LeI! SH, Ruckenstein E (1987) J Catal 107: 23
293. Ruckenstein E. Xu SD (1986) J Catal 100: 1
294. Ruckenstein E, Sushumna I (1986) J Catal 97: 1
295. Sushumna I, Ruckenstein E (1984) J Catal 90: 142
296. Sushumna I, Ruckenstein E (1985) J Catal 94: 239
297. Xin-Quan Xin, Xue-Qin Zhang, Long-Gen Zhu, Qing-Jin Meng, Zhao-Xian Wang, An-Bang
Dai (1980) Ts'ui Hua Hsueh Pao 1: 98
298. Sudo M, Ichikawa M, Soma M, Onishi T, Tamaru K (1969) J Phys Chern 73: 1174
299. Mei-Zhi, Bei, Xiu-Zheng Zhang, Shu-Fen Liu, Chang-Ping Shao, Juan Li (1981) Ts'ui Hua
Hsueh Pao, 2: 8
300. Pei-Qun Zhang, Gui-Quan Ji, Shi Li, Wei-Fang Wu, Jian-Xia Shen, Zong-Ce Wang (1981)
Hua Hsueh Tung Pao, 26: 16
301. Jianxia Shen, Zongce Wang, Xianrong Huang (1983) Sci Sin Ser B (Engl ed) 26: 1
302. Guerrero A, Lopez-Gonzales J de D, Moreno-Castilla C, Rodriguez-Reinoso F (1983) Ext Abst
Program 116th Bienn Conf Carbon 349
303. Kalucki K, Morawski W, Arabczyk W (1981) Stud Surf Sci Catal 7 B: 1496
304. Xinquan Xin, Lomggen Zhu, Qingjin Meng, Xuequin Zhang, Peicheng Wu, Anbang Dai (1982)
Gaodeng Xuexiao Hauxue Xuebao, 3: 162
305. Nefed'ev AV, Stukan RA, Alekseev VP, Postnikov VA, Shur VB, Novikov YuN, Vol'pin ME
(1977) Proc Int Conf Miissbauer Spectrosc 1: 259
306. Nitrogen Fixation Group, Chern Dep Nanking Univ (1977) Hua Hsueh Hsueh Pao 35: 141
307. Hsin-Chuan Hsin, Hsueh-Chin Chang, Chin-Chin Meng, Chao-Hsien Wang, Au-Pang Tai,
Yuan-Fu Hsia" Shen-Hao (1981) Yeh K'o Hsueh Tung Pao 26: 93
308. Tovbin MV, Kharchenko EV, Yatsimirskii VK, Bogatova NF, Velichanskaya LA, Zhidkova
TG (1975) Ukr Khim Zh (Russ ed) 41: 22
309. An-Bang Dai, Xin-Quan Xin. Long-Gen Zhu, Xue-Quin Zhang, Qing-Jin Meng (1979) Nan-
ching Ta Hsueh Hsueh Pao 40
310. Bewer G, Wichmann N, Boehm HP (1977) Mater Sci Eng 31. 73:6
311. Nitrogen Fixation Group Chern Dep Nanking Univ (1977) Hsueh Hsueh Hua Pao 35: 153
312. Postnikov VA, Dmitrienko LM, Ivanova RF, Dobrolyubova NL, Golubeva MA, Gapeeva TI,
Novikov YuN, Shur VB, Volpin ME (1975) Izu Akad Nauk SSSR Ser Khim p 2642
313. Rodriguez-Reinoso R, Guerrero-Ruiz A, Moreno-Castilla C, Rodriguez-Ramos I, Lopez-
Gonzales JD (1986) Appl Catal 23: 299
314. Arkhipov IL, Stukan RA. Yunusov SM, Lokshin VB, Ezernitskaya MB, Shur VB, Vol'pin ME
(1988) Metalloorg Khim 1: 314
315. Hegenberger E, Wu NL, Phillips J (1987) J Phys Chern 91: 5067
316. Chen AA, Vannice MA, Phillips J (1987) J Phys Chern 91: 6257
317. Horns N, Rimirez de la Piscina P, Fierro JLG, Sueiras JE (1984) Z Anorg Allg Chern 518: 227
318. Sueiras JE, Horns N, Ramirez de la Piscina P, Gracia M, Fierro JLG (1986) J Catal 98: 264
319. Marchetti SG, Alvarez AM, Mercader RC, Yeramian AA (1987) Appl Surf Sci 29: 443
320. Connell G, Dumesic JA (1986) J Catal 102: 216
321. Stockwell DM, Bertucco A, Coulston GW, Bennett CO (1988) J Catall13: 317
322. Boudart M, Delbouille A, Dumesic JA, Khammouma S, Tops0e H (1975) 1 Catal 37: 486
323. Boudart M, Tops0e H, Dumesic lA (1975) Batelle Inst Mater Sci Colloq 9: 337
324. Dumesic lA, Tops0e H, Khammouma S, Boudart M (1975) 1 Catal 37: 503
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 95

325. Dumesic JA, Topsoe H, Boudart M (1975) J Catal 37: 513

326. Boudart M, Dumesic JA, Topsoe H (1977) Proc Nat! Acad Sci USA 74: 806
327. Topsoe H, Dumesic JA, Derouane EG, Clausen BS, Morup S, Villadsen J, Topsoe N (1979) In:
Preparation of Catalysts II Elsevier Scientific Publishing Co (Amsterdam) p 365
328. Topsoe H, Topsoe N, Bohlbro H, Dumesic JA (1981) Stud Surf Sci Catal 7: 247
329. Ramirez de la Piscina P, Horns N, Fierro JLG, Sueiras JE (1985) Z Anorg Allg Chern 528: 195
330. Dumesic JA, Trevino AA (1989) J Catal 116: 119
331. Beckler RK, White MG (1988) J Catal 112: 157
332. Johnston C, Jorgensen N, Rochester CH (1988) J Chern Soc Faraday Trans 84: 2001
333. Johnston C, Jorgensen N, Rochester CH (1988) J Chern Soc Faraday Trans 1 84: 309
334. Rankin JL, Bartholomew CH (1986) J Catal 100: 526
335. Syrkov AG, Smirnov VM (1987) Kinet Katal 28: 1116
336. Lazar K, Reiff WM, Marke W, Guczi L (1986) J Catal 100: 118
337. Rankin JL, Bartholomew CH (1986) J Catal 100: 533
338. Babb KH, White MG (1986) J Catal 98: 343
339. Conenell G, Dumesic JA (1986) J Catal 101: 103
340. Santos J, Dumesic JA (1982) Stud Surf Sci Catal 11: 43
341. Santos J, Phillips J, Dumesic JA (1983) J Catal 81: 147
342. Tatarchuk BJ, Dumesic JA (1981) J Catal 70: 308
343. Kirch G, Hempelmann R, Schwab E, Zuechner H (1981) Z Phys Chern (Wiesbaden) 126: 109
344. Kirch G, Zuechner H (1982) Springer Ser Chern Phys 19: 398
345. Schwab E, Wicke E (1980) Z Phys Chern (Wiesbaden) 122: 217
346. Kirch G, Schwab E, Wicke E, Zuechner H (1985) In: Proc 8th Inst Congr Catal 4: Verlag
Chemie (Weinheim) 209
347. Biwer BM, Bernasek SL (1986) Appl Surf Sci 25: 41
348. Murell LL, Garten RL (1984) Appl Surf Sci 19: 218
349. Biwer BM, Bernasek SL (1986) Appl Surf Sci, 25: 41
350. Armbruster E, Baiker A, Baris H, Guentherodt HJ, Schlagl R, Walz B (1986) J Chern Soc Chern
Commun 299
351. Walz B, Wiesen danger R, Rosenthaler L, Guentherodt HJ, Dueggelin M, Guggenheim R
(1988) Mater Sci Eng 99: 501
352. Baiker A, Schlagl R, Armbruster E, Guntherodt HJ (1987) J Catal 107: 221
353. Schlagl R, Wiesendanger R, Baiker A (1987) J Catal 108: 452
354. Baiker A, Baris H, Schlagl R (1987) J Catal 108: 467
355. Kharchenko EV, Tovbin MV, Prisyazhnyuk KI, Maznitsa NF, Grion'ko AI (1969) Khim Prim
Ukr 7
356. Taylor OW, Smith PJ, Dowden DA, Kemball C, Whan DA (1982) Appl Catal, 3: 161
357. Arthyukh YuN, Lavrentovich RF, Rusov MT (1968) Sci Selec Catal 214
358. Artyukh YuN, Rusov MT, Boldyreva NA (1967) Kinet Katal 8: 1319
359. Yatsimirskii VK, Ishchenko EV, Maksimov YuV, Arents RA, Suzdalev IP (1985) Teor Eksp
Khim 21: 500
360. Yatsimirskii VK, Ischenko EV, Girenkova NI (1985) Ukr Khim Zh (Russ Ed.) 51: 599
361. Kalucki K, Narkiewicz U (1987) Pr Nauk Po Ii tech Szczecin, 354: 9
362. Yatsimirskii VK, Ishchenko EV, Nazarenko VA, Mischanchuk TB, Girenkova NI (1988) Teor
Eksp Khim 24: 185
363. Nagorny K, Bubert S (1987) J Catal 112
364. Yatsimirskii VK, Girenkova NI, Panyuk EV (1987) Zh Fiz Khim 61: 1516
365. Paryjczak T, Zielinski P (1983) React Kinet Catal Lett 23: 165
366. Rabina PO, Kuznetsov LD, Enikeev EKh, Ivanov MM (1967) Kinet Katal 8: 167
367. Tovbin MV, Zabuga VYa, Perekhod NA, Stupak OP (1979) Kinet Katal17: 49
368. Yatsimirskii VK, Girenkova NI, Ostapyuk VA, Maksimov YuV, Kushkov VD (1979) Kinet
Katal 17: 53
369. Lachinov SS, Krylova AV, Vasser berg VE, Chinh Chuong Zang, Ratnikova LG, Toroches-
hnikov (1976) Deposited Doc VINITI 1241
370. Yatsimirskii VK, Ishchenko EV, Mischanchuk TB, Zo Zen Ho (1988) Teor Eksp Khim 24: 616
371. Matsuyama M, Ashida K, Takayasu 0, Takeuchi T (1986) J Catal 102: 309
372. Tkachenko ZhI, Pavlenko LI, Lyubchenko YuA (1980) Zh Prikl Khim (Leningrad) 53: 2080
373. Lyubchenko YuA, Skolozdra OE, Sergeeva AN (1975) Zhur Prikl Khim 48: 958
374. Kuzora TV, Kozin LF, Sokol'skii DV (1975) Izv Akad Nauk Kaz SSR Ser Khim 25: 15
375. Egawa C, Sawabe K, Iwasawa Y (1988) J Chern Soc Faraday Trans 1 84: 321
96 P. Stoltze

376. Haase G, Asscher M, Linke U (1988) Appl Surf Sci 35:1

377. Zhou X-L, Yo on C, White JM (1988) Surf Sci 203: 53
378. Bozso F, Ert! G, Grunze M, Weiss M (1977) Appl Surf Sci 1: 103
379. Shvachko VI, Fogel YaM, Kolot VYa (1966) Kinet Katal 7: 834
380. Emmett PH, Takezawa N (1978) J Res Inst Catal Hokkaido Univ 26: 37
381. Emmett PH, Harkness RW (1935) J Am Chern Soc 57: 1631
382. Viswanathan B, Srinivasan V, Sastri MVC (1969) Indian J Chern 7: 363
383. Benzinger J, Madix R (1980) Surf Sci 94: 119
384. Kurz EA, Hudson JB (1988) Surf Sci 195: 15
385. Wedler G, Geuss, Colb, McElhiney (1978) Appl Surf Sci 1: 471
386. Wedler G, Borgmann D (1976) J Catal 44: 139
387. Sharnabanger (1985) Surf Sci 150: 451
388. Chornet E, Coughlin RW (1972) J Catal 27: 246
389. Bagg J, Tompkins F (1955) Trans Faraday Soc 51: 1071
390. Porter AAS, Tompkins FC (1953) Proc Roy Soc A217: 544
391. Kwan T (1949) J Res Inst Catal Hokkaido Univ 1: 100
392. Evdokimova ZhA, Valitov NKh, (1985) Zh Prikl Khim (Leningrad), 58: 672
393. Sastri MVC, Srikant H (1961) J Sci Ind Research (India), 20D: 321
394. Koebl H, Eschweiler R, Ralek M, (1973) Z Naturforsch A 28: 54
395. Ostrovskii VE, Igranova EG (1978) Kinet Katal 19: 681
396. Stoltze P (1987) Phys Scr 36: 824
397. Grunze M, Ert! G (1977) Proc 7th Int Vac Congr 2: 1137
398. Kurz EA, Hudson JB (1988) Surf Sci 195: 31
399. Nowicka E, Wolfram Z, Lisowski W, Dus R (1985) Surf Sci 151: L166
400. Vinals J, Gunton JD (1985) Surf Sci 157: 473
401. Arab A, Hudson JB (1987) Appl Surf Sci 29: 1
402. Porter AS, Tompkins FC (1953) Proc Roy Soc A217: 529
403. Artyukh, Yu N (1981) Kinet Katal 19: 14
404. Kai. Huei. Huang, Xiao.Ming.Zeng, Ji.Tao.Li. (1983) J Catal 81: 259
405. Krylova AV, Vokhmyanin NN, Lachinov SS, Torocheshnikov NS (1974) Tr Mosk Khim-
Tekhnol Inst 79: 14
406. Amenomiya Y, Pleizier G (1973) J Catal 28: 442
407. Balint I, Segal E, Bucur I, Chirulescu T (1983) Thermochim Acta 67: 103
··408. Kummer JT, Emmett PH (1951) J Phys Colloid Chern 55: 337
409. Enikeev EKh, Ivanov MM (1973) Kinet Katal 14: 1597
410. Artyukh YuN, Lunev NK, Rusov MT (1970) Dopov Akad Nauk RSR Ser B 32: 605
411. Ert! G, Lee SB, Weiss M (1981) Surf Sci 111: L711
412. Artyukh YuN, Il'yushonok VA (1976) Kinet Katal17: 1078
413. Brunauer S, Emmett PH (1940) J Am Chern Soc 69: 1432
414. Ert! G, Huber M, Lee SB, Paal Z, Weiss M (1981) Appl Surf Sci 8: 373
415. Russo C, Kaplow R (1978) J Vac Sci Technol 15: 479
416. Martsenyuk MG, Rusov MT, Samchenko NP (1968) Kinet Katal 4: 129
417. Ustinov YuK, Urazaev RSh (1979) Kinet Katal 20: 1513
418. Rambeau G, Amariglio H (1978) J Chim Phys-Chim Bioi 75: 333
419. Artyukh YuN, Batychko SV, Roev LM (1972) Katal 9: 72
420. Tret'yakov II, Shub BR, Sklyarov AV, Roginskii SZ (1967) Dokl Akad Nauk SSSR
175: 1332
421. Tret'yakov II, Shub BR, Shlyarov AV (1968) Dokl Akad Nauk SSSR 181: 648
422. Boreskova EG, Kuchaev VL, Temkin MI (1984) Kinet Katal 25: 112
423. Grabke HJ (1970) AGARD Conf Proc 52: 8
424. Kummer JT, Emmett PH (1952) J Phys Chern 56: 258
425. Emmett PH, Harkness RW (1935) J Am Chern Soc 57: 1624
426. Chavalier JC, Chornet E (1976) Surf Sci 60: 125
427. Brundle CR (1978) IBM J Res Develop 22: 235
428. Erley W (1981) J Vac Sci Technol 18: 472
429. Yoshida K, Somorjai GA (1978) Surf Sci 75: 46
430. Moon DW, Cameron S, Zaera F, Eberhardt W, Carr R, Bernazek SL, Gland JL, Dwyer DJ
(1987) Surf Sci 180: L123
431. Rhodin TN, Rsai M-H, Kasowski RV (1985) Appl Surf Sci 22/23: 426
432. Watanabe M, Kadowaki T (1987) Appl Surf Sci 28: 147
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 97

433. Lu J-P, Albert MR, Bemasek SL, Dwyer DJ (1988) Surf Sci 199: L406
434. Kummer JT, Emmett PH (1951) J Am Soc 73: 2886
435. Eischens RP (1952) J Am Chern Soc 74: 6167
436. Wedler G, Ruhmann R (1982) Appl Surf Sci 14: 137
437. Seip U, Tsai C, Christmann, Kuppers J, Ert! G (1984) Surf Sci 139: 29
438. Wedler G, Colb KG, McElhiney G, Heinreich W (1978) Appl Surf Sci 2: 30
439. Beebe, Stevens (1940) J Am Chern Soc 62: 2134
440. Krylova AV, Chechulina GN, Koroleva TL, Lachinov SS, Torocheshnikov NS (1973) Depos-
ited Doc. VINITI 7698
441. Ueda K, Enatsu M (1985) Surf Sci 159: L421
442. Ueda K, Enatsu M (1985) Surf Sci 159: L421
443. Benndorf C, Kruger B, Thieme F (1985) Surf Sci 163: L675
444. Cameron SD, Dwyer DJ (1988) Surf Sci 198: 315
445. Alshorachi G, Wedler G (1985) Appl Surf Sci 20: 279
446. Moon DW, Dwyer DJ, Bemasek SL (1985) Surf Sci 163: 215
447. Huang V-V, Emmett PH (1972) J Catal 24: 101
448. Gafner G (1979) S Afrl J Phys 2: 129
449. Rao CNR, Ranga G (1988) Chern Phys Lett 146: 557
450. Behner H, Spiess W, Wedler G, Borgmann D (1986) Surf Sci 175: 276
451. Pimer M, Bauer R, Borgmann D, Wedler G (1987) Surf Sci 189/190: 147
452. Peev T, Krylova A, Nefedova N, Stoilova T (1986) J Radioanal Nucl Chern 105: 157
453. Krylova AV, Nefedova NV, Peev TM, Stoilova TI (1986) Kinet Katal 27: 520
454. Krylova AV, Ustimenko GA, Nefedova NV, Peev TM, Torocheshnikov NS (1986) Appl Catal
20: 205
455. Krylova AV, Nefedova NV, Torocheshnikov NS (1988) Appl Catal 39: 325
456. Krylova AV, Nefedova NV, Peev TM (1987) Stud Surf Sci Catal 34: 625
457. Peev T, Krilova A, Nefedova N, Stoilova T (1986) Khim Ind (Sofia) 58: 2
458. Ponec V, Knor Z (1968) J Catal 10: 73
459. Hall PG, Hope CJ (1970) J Chern Soc A 1970: 2003
460. Tomanek D, Bennemann KH (1985) Phys Rev B 31: 2488
461. Freund HJ, Bartos B, Messmer RP, Grunze M, Kuhlenbeck H, Neumann M (1987) Surf Sci
185: 187
462. Grunze M, Golze M, Fuhler J, Neumann M, Schwarz E (1985) In Proc 8th Int Congr Catal4:
133
463. Grunze M, Golze M, Hirschwald W, Freund H.-J, Pulm H, Seip U, Tsai MC, Ert! G, Kuppers J
(1984) Phys Rev Lett 53: 850
464. Grunze M, Strasser G, Golze M (1987) Appl Phys A A44: 19
465. Borgmann D, Kroninger F, Steidl P, Wedler G (1987) Surf Sci 189-190: 485
466. Ert! G, Lee SB, Weiss M (1982) Surf Sci 114: 527
467. Ert! G, Lee SB, Weiss M (1982) Surf Sci 114: 515
468. Bozso F, Ert! G, Weiss M (1977) J Catal 50: 519
469. Wedler G, Borgmann D, Geuss J (1975) Surf Sci 47: 592
470. Tsai MC, Seip U, Bassignana IC, Kuppers J, Ertl G (1985) Surf Sci 155: 387
471. Kishi K, Roberts MW (1977) Surf Sci 62: 252
472. Baro AM, Eley W (1980) Surf Sci 112: L 759
473. Liao D, Zhang H, Wang A, Cai Q (1987) Sci Sin Ser B (Eng\. Ed.) 30: 246
474. Bozso F, Ert! G, Grunze M, Weiss M (1977) J Catal 49: 18
475. Whitman LJ, Bartosch CE, Ho W, Strasser G, Grunze M (1986) Phys Rev Lett 56: 1984
476. Toyoshima 1(1981) Shinku, 24: 216
477. Ert! G, Huber M (1980) Z Phys Chern (Wiesbaden), 119: 97
478. Grunze M, Kind In D, Woodruff DP (1982) eds, The Chemistry and Physics of Solid Surfaces
and Heterogeneous Catalysis, Vol 4, p 413
479. Wedler G, Steidl G, Borgmann D (1980) Surf Sci 100: 507
480. Wedler G, Borgmann D, Alter W, Witan K (1986) Ber Bunsen-Ges Phys Chern 90: 235
481. Kummer JT, Emmett PH (1951) J Chern Phys 19: 289
482. Magomedbekov EP, Kasatkina LA (1976) Zhur Fiz Khim 50: 2414
483. Morikawa Y, Ozaki A (1968) J Catal 12: 145
484. Ert! G, Grunze M, Weiss M (1976) J Vac Sci Technol 13: 314
485. Imbihl R, Behm RJ, Ert! G, Moritz W (1982) Surf Sci 123: 129
486. Arabczyk W, Muessig HJ (1987) Vacuum, 37: 137
98 P. Stoltze

487. Dowben PA, Grunze M, Jones RG (1981) Surf Sci 109: L519
488. Wedler G, Borgmann D, Forchhungsber Wehrtech (Bundesrnisist. Verteidigung)., No. BMVg-
FBWT 76-20 Luft-Raurnfahrt Teil. 2: 151
489. Evdokirnova ZhA, Valitov NKh (1985) Zh Prikl Khirn (Leningrad) 58: 2121
490. Vito I EN, Orlova KB (1973) Kinet Katal 14: 1514
491. Kwan T (1953) J Res lust Catal Hokkaido Univ 3: 16
492. Kwan T (1955) J Res Inst Catal Hokkaido Univ 3: 109
493. Brito V, Ralek M (1978) React Kinet Catal Lett 9: 15
494. Zwietering EP, Roukens JJ (1954) Trans Faraday Soc 50: 178
495. Scholten JJF, Zwietering P, Konvalinka JA, de. Boer HJ (1959) Trans Faraday Soc
55: 2166
496. Boreskova EG, Kuchaev VL, Temkin MI (1979) Kinet Katal 20: 147
497. Minachev KhM, Evdokirnova ZhA, Valitov NKh (1982) Dokl Akad Nauk SSSR
266: 296
498. Grabke HJ (1976) Z Phys Chern (Frankfurt am Main), 100: 185
499. Ostrovskii VE, Igranova EG (1978) Kinet Katal 19: 538
500. Paal Z, Ert! G, Lee SB (1981) Appl Surf Sci 8: 231
501. Boheirn J, Brenig W, Engel T, Leuthauser U (1983) Surf Sci 131: 298
502. Rettner CT, Stein H (1987) J Chern Phys 87: 770
503. Rettner CT, Stein H (1987) Phys Rev Lett 59: 2768
504. Rettner CT, Pfnur HE, Stein H, Auerbach DJ (1988) J Vac Sci Technol A 6: 899
505. Durnesic JA, Boudart M (1975) Proc Syrnp Catal Chern Nitrogen Oxides p 95
506. Egawa C, Iwasawa Y Chern Lett 1987: 959
507. Nakata T, Matsushita S (1982) J Chern Phys 76: 6335
508. Tarnaru K (1965) In: Sacht!er WMH, Schuit GCA, Zwietering P (eds), Proc 3,d Intern Conf
Catal 1: 664
509. Takezawa N (1972) J Catal 24: 417
510. Schmidt WA (1968) Angew Chern Int Ed Engl 7: 139
511. Artyukh YuN, Golovatyi VG, Korol EN (1980) Teor Eksp Khirn 16: 714
512. Okawa T, Onishi T, Tarnaru K (1977) Z Phys Chern (Wiesbaden), 107: 239
513. Karpinski W, Palczewska W (1974) Bulll'Acad Polon Sci Chirn 22: 159
514. Kozub GM, Strel'tsov OA, Rusov MT (1973) Kinet Katal 10: 52
515. Paal Z, Ert! G (1982) Kern Kozl 57: 51
516. Arthyukh YuN, Yas'rno VI, Rusov MT (1971) Kinet Katal 8: 51
517. Gay ID, Textor M, Mason R, Iwasawa Y (1977) Proc R Soc London A356: 25
518. Tarnarau K Prep. 2nd Intern Congr Catal (Paris), 1: 325
519. Tarnaru GK (1963) Trans Faraday Soc 59: 979
520. Joris GG, Taylor HS (1939) J Chern Phys 7: 893
521. Boreskov GK, Kolchanova VM, Rachkovskii EE, Filirnonova SN, Khasin AV (1975) Kinet
Katal 16: 1218
522. Takezawa N, Toyoshirna I (1970) J Catal 19: 271
523. Kazusaka A, Toyoshirna I (1981) Z Phys Chern (Wiesbaden) 128: 111
524. Yatsirnirskii VK, Girenkova NI, Maksirnov YuV (1976) Teor Eksp Khirn 12: 263
525. Urabe K, Ozaki A (1978) J Catal 52: 542
526. Tanaka K, Matsuyarna A (1971) J Res lust Catal Hokkaido Univ 19: 63
527. Tanaka K (1966) J Res Inst Catal Hokkaido Univ 14: 153
528. Tanaka K (1965) J Res lust Catal Hokkaido Univ 13: 119
529. Napol'skikh GA, Kasatkina LA (1973) Kinet Katal 14: 653
530. Miyahara K (1956) J Res lust Catal Hokkaido Univ 4: 193
531. Vol'pin ME, Novikov YuN, Postnikov VA, Schur VB, Bayerl B, Kaden L, Wahren M,
Drnitrienko LM. Stukan RA, Nefed'ev AV (1977) Z Anorg AlIg Chern 428: 231
532. Morikawa Y, Ozaki A (1971) J Catal 23: 97
533. Boreskova EG, Kuchaev VL, Temkin MI (1984) Kinet Katal 25: 116
534. Schulz G, Schaefer H (1969) Z Phys Chern (Frankfurt am Main) 64: 333
535. Bowker M, Parker IB, Waugh KC (1985) Appl Catal 14: 101
536. Arnariglio H, Rambeau G (1976) Proc 6th Int Congr Catal 2: 1113
537. Bowker M, Parker I, Waugh KC (1988) Surf Sci 197: L223
538. Hinda F, Hirokawa K (1977) J Electron Spectrosc Related Phenorn 12: 313
539. Benndorf C, Madey TE, Johnson AL (1987) Surf Sci 187: 434
540. Erley W, Ibach H (1983) J Electron Spectrosc Related Phenorn 31: 61
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 99

541. Weiss M. Ert! G, Nietsche F (1979) Appl Surf Sci 2: 614

542. Grunze M, Bozso F, Ert! G, Weiss M (1978) Appl Surf Sci 1: 241
543. Yoshida K (1980) J Res lust Catal Hokkaido Univ 28: 15
544. Okawa T, Onishi T, Tamaru K (1977) Chern Lett 1977: 1077
545. Drechsler M, Hoinkes H, Kaarmann H, Wilsch H, Ert! G, Weiss M (1979) Appl Surf Sci 3: 217
546. Erley W, Ibach H (1981) Surf Sci 119: L357
547. Veselkova AA, Krylova AV, Torocheshnikov NS (1980) Deposited doc no VINITI 3109-80
548. Block J, Schulz-Ekloff G (1973) J Catal 30: 327
549. Al-Haydari YK, Saleh JM, Matloob MH (1985) J Phys Chern 89: 3286
550. Grunze M (1978) Surf Sci 81: 603
551. Ustimenko GA, Krylova AV, Torocheshnikov NS (1978) Deposited doc no VINITI 3611-78
552. Bruker CF, Rhodin TN (1976) Surf Sci 57: 523
553. Pignocco AJ, Pellisier GE (1967) Surf Sci 7: 261
554. Pirug G, Broden G, Bonzel HP (1980) Surf Sci 94: 323
555. Miyano T, Saklisaka Y, Komeda T, Onichi M (1986) Surf Sci 169: 197
556. Vink TJ, Kinderen JM, Gijzeman OLJ, Geus JW, van Zoest JM (1986) Appl Surf Sci 26: 357
557. van Zoest JM, Fluit JM, Vink TJ, van Hassel BA (1987) Surf Sci 182: 179
558. Horgan AM, King DA (1970) Surf Sci 23: 259
559. Vasilevich AA, Blokhina LN, Chesnokova RV, Minaev DM (1988) React Kinet Catal Lett
36: 467
560. Gimzewski JK, Pad alia BD, Affrosman S, Watson LM, Fabian DJ (1977) Surf Sci 62: 386
561. Simons GW, Dwyer DJ (1975) Surf Sci 48: 373
562. Papagno L, Caputi LS, Chiarello G, Delogu P (1986) Surf Sci 175: L767
563. Aleksic BD, Mitov IG, Klissurski DG, Petranovic NA, Jovanovic NN, Boganov SS (1984) Bull
Soc Chim Beograd 49: 477
564. Chudnikov MG, Minaev DM, Zaichko GN, Alekseev AM (1984) Kinet Katal 25: 1205
565. Arents RA, Maksimov Yuv, Baldokhin YuV, Suzdalev IP, Chesnokova RV, Minarev DM
(1983) Poverkhnost no 12: 52
566. Maksimov YuV, Arents RA, Baldokhin YuV, Sudalev IP, Minarev DM, Chesnokova RV
(1982) React Kinet Catal Lett 21: 81
567. Enikeev EZh, Loskutov AI, Rozenfel'd IL (1977) 2nd Tezisy Dokl Vses Simp Akt Poverkhn
Tverd Tel 1977: 23
568. Tsarev VI, Aptekar EL, Krylova AV, Torocheshnikov NS (1980) Izv Akad Nauk SSSR Ser
. Khim p 1404
569. Morozov VV, Vokhmaynin NN, Krylova AV, Pograbennyi YuV, Kasatkina LA, Toroches-
hnikov NS (1979) Tr Mosk Khim-Tekhnol Inst im D I Mendeleeva 107: 144
570. Vasilevich AA, Chesnokova RV, Minaev DM, KUznetsov LD (1979) Kinet Katal 20: 984
571. Burnett JA, Allgood HY, Hall JR (1953) Ind Eng Chern 45: 1678
572. Krylova AV, Morozov VV, Lachinov SS, Torocheshnikov NS (1978) React Kinet Catal Lett
9: 125
573. Tsarev VI, Aptekar EL, Krylova AV, Torocheshnikov NS (1980) React Kinet Catal Lett
14: 279
574. Tsarev VI, Aptekar EL, Krylova AV (1981) Izu Akad Nauk SSSR Ser Khim p 1658
575. Krylova AV, Tsarev VI, Peev T, Kushnarenko TI, Torocheshnikov NS (1986) React Kinet
Catal Lett 30: 229
576. Krylova AV, Spirina LN, Rakhmat-Zade AG, Mokhova VN, Lachinov SS, Torocheshnikov
NS (1974) Tr Mosk Khim-Tekhnol Inst 79: 24
577. Dorfeld WG, Hudson JB, Stuhr R (1976) Surf Sci 57: 460
578. Brundle CR (1979) Surf Sci 66: 581
579. Krylova AV, Vokhmanin NN, Koroleva TL, Lachinov SS (1973) Deposited Doc no VINITI
7699
580. Sakisaka Y, Komeda T, Miyano T, Onichi M, Masuda S, Harada Y, Vagi K, Kato H (1985)
Surf Sci 164: 220
581. Hall GK, Mee CHB (1971) Surf Sci 28: 598
582. Alshorachi G, Wed;er G (1985) Appl Surf Sci 20: 279
583. Kasatkina LA, Krylova AV, Morozov VV, Holu A (1977) Kinet Katal 18: 1603
584. Krylova AV, Morozov VV, Kasatkina LA (1979) Kinet Katal 20: 806
585. Patyi L, Tsarev VI, Krylova AV, Orovac D, Farkas Z (1978) Deposited doc no VINITI 3096
586. Heras JM, Albano EV (1983) Appl Surf Sci 17: 220
587. Heras JM, Albano EV (1983) Appl Surf Sci 17: 207
100 P. Stoltze

588. Baer DR, Thomas MT (1986) Appl Surf Sci 26: 150
589. Nakanishi S, Sasaki K (1988) Surf Sci 194: 245
590. Brill R, Schafer H, Zimmermann G (1968) Ber Bunsen Ges Phys Chern 72: 1218
591. Zubova IE, Rabina PD, Pavlova NZ, Kuznetsov LD, Chudinov MG, Le. Congo Shang (1974)
Kinet Katal 15: 1261
592. Tovbin MV, Psheichnaya OV (1966) Kinet Katal 2: 149
593. Nwalor JU, Goodwin JG Jr, Biloen P (1989) J Catal117: 121
594. Ertl G, Huber M, Thiele N (1979) Z Naturforsch 34A: 30
595. LoeftIer DG, Schmidt LD (1976) J Catal 44: 244
596. Brill R (1970) J Catal 19: 236
597. Yatsimirskii VK (1978) Zh Fiz Khim 52: 584
598. Kuznetsova EP, Samchenko NP, Rusov MT (1974) Kinet Katal 11: 92
599. Mahapatra H, Chhabra DS, Puri VK, Sen SP (1981) Fert Technol 18: 160
600. Samchenko NP, Rusov MT, Strel'tsov OA (1966) Kinet Katal2: 96
601. Vasilevich AA, Rabina PD, Alekseev AM, Dmitrenko LM, Mosolova E, Lyubchenko YuA,
Muravskaya GK, Kuznetsov LD (1973) Izu Otd Khim Mauki Bulg Akad Nauk 6: 225
602. Brill R, Kurzidim J (1970) Colloq Int Cent Nat Rech Sci 187: 99
603. Spencer ND, Schoonmaker RC, Somorjai GA (1982) J Catal 74: 129
604. Spencer ND, Schoonmaker RC, Somarjai GA (1981) Nature (London) 294: 643
605. Strongin DR, Carrazza J, Bare SR, Somorjai GA (1987) J Catal 103: 213
606. Yatsimirskii VK (1982) Izv Sib Otd Akad Nauk SSSR Ser Khim Nauk 1982: 131
607. Rambeau G (1987) Bull Soc Chim Fr 1987: 815
608. Artyukh YuN, Zyuzya LA (1981) Kinet Katal 19: 17
609. Tornqvist E, Chen AA (1991) Catal Lett 8: 359
610. Fierro JLG, Horns N, Ramirez P, Sueiras J (1984) React Kinet Catal Lett 24: 179
611. Magomedbekov EP, Kasatkina LA (1978) Tr Mosk Khim-Tekhnol Inst im DI Mendeleeva
99: 60
612. Chirstozvonov DB, Lytkin VP, Sobolevskii VS, Kozlov LI, Markovets LN, Kirillov IP,
Korbutova ZV (1969) Izv Vyssh Ucheb Zaved Khim Khim Teknol12: 1388
613. Miki Y, Terao I, Uchida H (1967) Tokyo Kogyo Shikensho Hokoku 62: 64
614. Altenburg K, Bosch H, Van. Ommen JG, Gellings PJ (1980) J Catal 66: 326
615. Scholten JJF, Zwietering P (1957) Trans Faraday Soc 10: 1363
616. Parker IB, Waugh KC, Bowker M (1988) J Catal 114: 457
617. Nielsen A, Kjaer J, Hansen B (1964) J Catal 3: 68
618. Krabetz R, Peters C (1963) Ber Bunsen-Ges Phys Chern 67: 381
619. Bosch H, van Ommen JG, Gellings PJ (1985) Appl Catal 18: 405
620. Takezawa N (1966) Shokubai, 8: 390
621. Rudnitskii LA, Berengarten MG, Alekseev AM (1973) J Catal 30: 444
622. Rudnitskii LA, Berengarten MG (1972) Kinet Katal 13: 115
623. Samchenko NP, Golodets GI (1986) Kinet Katal 27: 378 (russ) 324 (eng)
624. Samchenko NP, Golodets GI (1986) Kinet Katal 27: 384 (russ) 329 (eng)
625. Stoltze P, Nl'Jrskov JK (1987) J Vac Sci Technol A 5: 581
626. Nl'Jrskov JK (1981) J Vac Sci Technol 18: 420
627. Holloway S, Lundqvist BI, Nl'Jrskov JK (1984) Proc 81h Intern Congr Catal (Berlin) 4: 85
628. Nl'Jrskov J, Holloway S, Lang ND (1984) Surf Sci 137: 65
629. Rudnitskii LA, Berengarten MG (1971) Dokl Akad Nauk SSSR 201: 396
630. Markert K, Wandelt K (1985) Surf Sci 159: 24
631. Schlagl R, Schoonmaker RC, Muhler M, Ertl G (1988) Catal Lett 1: 237
632. Konoplya MM, Gorlov Yul, Yatsimirskii VK (1982) Teor Eksp Khim 18: 398
633. Kai-Hui.Huang. (1981) Sci Sin (Engl. ed) 24: 800
634. Kai Huei Huang (1981) Stud Surf Sci Catal 7 A: 554
635. Tomanek D, Kreuzer HJ, Block JH (1986) J Phys Colloq 1986: 139
636. Tomanek D, Kreuzer HJ, Block JH (1985) Surf Sci 157: L315
637. Ortoleva E, Simonetta M (1985) Croat Chern Acta 57: 1387
638. Gagarin SG, Chuvylkin ND (1981) Zh Fiz Khim 55: 3094
639. Guzikevich AG, Chuiko AA, Yatsimirskii VK (1988) Teor Eksp Khim 24: 263
640. Tomanek D, Kreuzer HJ, Block JH (1985) Surf Sci 157: L315
641. Holloway S, Hodgson A, Halstead D (1988) Chern Phys Lett 147: 425
642. Boudart M (1988) Catal Lett 1: 21
643. Boudart M, LoftIer DG (1984) J Phys Chern 88: 5763
 2 Structure and Surface Chemistry of Industrial Ammonia Synthesis Catalysts 101

644. Stoltze P, Norskov JK (1985) Phys Rev Lett 55: 2502

645. Stoltze P, Norskov JK (1988) J Catal 110: I
646. Kjaer J (1958) Measurement and calculation of temperature and conversion in fixed bed
catalytic reactors. Gjellerup
647. Zhang HB, Schrader GL (1986) J Catal 99: 461
648. Lui W, Tsong TT (1986) Surf Sci 165: L26
649. Li Y, Drachsel W, Block JH, Okuyama F (1986) J Phys Colloq 1986: 413
650. Wedler G, Borgmann D (1971) Angew Chern Internat Ed 10: 562
651. Aika KI, Ohhata T, Ozaki A (1970) J Catal 19: 140
652. Temkin MI, Morozov NM, Shapatina EN (1963) Kinet Katal 4: 565
653. de Bruijn H (1950) Disc Faraday Soc 1950: 69
654. Igranova EG, Ostrovskii VE, Temkin MI (1976) Kinet Katal 17: 1257
655. Kozub GM, Turchaninov AM, Zyuzya LA, Voroshilov IG (1977) Katal 15: 64
656. Melton FCE, Emmett PH (1964) J Phys Chern 68: 3318
657. Mars P, Scholten JJF, Zwietering P (1960) In: de Boer HJ (ed) The Mechanism of Heterogen-
eous Catalysis p 66
658. Brill R, Jiru P, Schulz G (1969) Z Phys Chern (Frankfurt am Main), 64: 215
659. Nakata T, Matsushita S (1968) J Phys Chern 72: 458
660. Selwyn GS, Lin MC (1982) Chern Phys 67: 213
661. Bhattacharya AK, Chesters MA (1987) J Catal 108: 484
662. Love K, Emmett PH (1941) J Am Chern Soc 63: 3297
663. Takezawa N, Toyoshima I (1966) J Res Inst Catal Hokkaido Unlv 14: 41
664. Takezawa N, Toyoskima I (1968) J Res Inst Catal Hokkaido Univ 15: 111
665. Toyoshima I, Horiuti J (1958) J Res Inst Catal Hokkaido Univ 6: 146
666. Ert! G, Huber M (1980) J Catal 61: 537
667. Tanaka K. Yakamoto 0, Matusuyama A (1965) Proc 3,d Intern Congr Catal 1965: 676
668. Nakata T, (1982) J Chern Phys 76: 6328
669. Gauchman SS, Roiter WA (1938) Zh Fiz Khim 11: 569
670. Kwan T, (1956) J Phys Chern 60: 1033
671. Temkin M, Pyzhev V (1940) Acta Physiochem URSS 12: 327
672. Kodera T, Takezawa N (1960) J Res Inst Catal Hokkaido Univ 8: 157
673. Enomoto S, Horiuchi J (1953) J Res Inst Catal Hokkaido Univ 2: 87
674. Horiuti J, Takezawa N (1960) J Res Inst Catal Hokkaido Univ 8: 127
675. Horiuti J, Toyoshima I (1957) J Res Inst Catal Hokkaido Univ 5: 120
676. Enomoto S, Horiuti J, Kobayashi H (1955) J Res Inst Catal Hokkaido Univ 3: 185
677. Temkin MI, Morozov NM, Shapatina EN (1963) Kinet Katal4: 260
678. Takezawa N, Toyoshima I (1966) J Phys Chern 70: 594
679. Horiuti J, Takezawa N (1960) J Res Inst Catal Hokkaido Univ 8: 170
680. Horiuti J, Toyoshima I (1958) J Res Inst Catal Hokkaido Univ 6: 68
681. Takezawa N, Toyoshima I (1966) J Catal 6: 145
682. Li C, Hudgins RR, Silveston PL (1985) Can J Chern Eng 63: 795
683. Li C, Hudgins RR, Silveston PL (1985) Can J Chern Eng 63: 803
684. Rambeau G (1988) Bull Soc Chim Fr 1988: 450
685. Richard MA, Vanderspurt TH (1985) J Catal 94: 563
686. Rambeau G, Amariglio H (1978) J Chim Phys Phys-Chim Bioi 75: 110
687. Rambeau G, Amariglio H (1978) J Chim Phys Phys-Chim Bioi 75: 397
688. Jain AK, Hudgins RR, Silverston PL (1982) Can J Chern Eng (1982) 60: 809
689. Rambeau R (1988) Bull Soc Chim Fr 1988: 941
690. Jain AK, Li C, Silveston PL, Hudgins RR (1985) Chern Eng Sci 40: 1029
691. Horiuti J, Kita H (1956) J Res Inst Catal Hokkaido Univ 4: 132
692. Schwab GM, Krabetz R (1956) Z Electrochem 60: 855
693. Aika K, Ozaki A (1970) J Catal 19: 350
694. Brill R (1970) J Catal 16: 16
695. Temkin M, Pyzev V (1939) Zh Fiz Khim 13: 851
696. Panov GI, Kharitonov AS (1985) React Kinet Catal Lett 29: 267
697. Shapatina EN, Kuchaev VL, Temkin MI (1988) Kinet Katal 29: 603 (russ) 520 (eng)
698. Kuchaev VL, Shapatina EN, Temkin MI (1988) Kinet Katal 29: 610 (russ) 526 (eng)
699. Boudart M, Djega-Mariadassou G (1984) Kinetics of Heterogeneous Catalytic Reactions.
Princeton University Press
700. Brunauer S, Love KS, Keenan RG (1942) J Am Chern Soc 64: 751
102 P. Stoltze

701. Tamaru K (1964) Bull Chem Soc Japan, 37: 1087

702. Temkin MI (1967) Kinet Katal 8: 1005
703. Boreskova EG, Kuchaev VL, Pen'kovoi BE, Temkin MI (1972) Kinet Katal 13: 358
704. Brill R (1951) J Chem Phys 19: 1047
705. Peters C, Krabetz R (1957) Z Electrochem 60: 859
706. Ozaki A, Taylor H, Boudart M (1960) Proc Roy Soc 258: 47
707. Takezawa N, Toyoshima I, Kazusaka A (1972) J Catal 25: 118
708. Anokhin VN, Menshov VN, Zuev AA (1975) Zhur Prikl Khim 48: 489
709. Shapatina EN, Kuchaev VL, Temkin MI (1972) Kinet Catal 12: 1307
710. Kita K, Ozaki A (1969) J Catal 13: 232
711. Horiuti J. (1957) J Res lust Catal Hokkaido Univ 5: 1
712. Bokhoven C, Gorgels MJ, Mars P (1959) Trans Faraday Soc 55: 315
713. Yamamoto 0, Tanaka S, Tanaka K (1974) Sci Pap lust Phys Chern Res (Jap) 68: 8
714. Enomoto S, Horiuti J (1952) Proc Jap Acad 28: 493
715. Baranski A, Reizer A, Kotarba A, Pyrczak E (1985) Appl Catal 19: 417
716. Brill R, Hensel J, Schaefer H (1969) Ber Bunsen-Ges Phys Chern 73: 1003
717. Smirnov lA, Morozov, NM, Temkin MI (1965) Kinet Katal 6: 351
718. Brill R, Hensel J, Schaefer H (1969) Ber Bunsen-Ges Phys Chern 73: 1003
719. Royen P, Langhans GH (1962) Z Anorg Allgem Chern 315: 1
720. Kiperman SL (1954) Zjur Fiz Khim 28: 389
721. Le Cong Shang, Kuznetsov DA, Rabina PD, Zubova IE, Semenov GM (1967) Tr Mosk Khim-
Tekhnol lust 56: 174
722. Sarkisyan AA, Kuznetsov LD, Rabina PD, Vavilov NS (1980) Khim Prom 1980: 92
723. Brill R, Tauster S (1963) Ber Bunsen-Ges Phys Chern 67: 390
724. N"lfSkov JK, Stoltze P (1987) Surf Sci 189/190: 91
Chapter 3
Ammonia Synthesis over Non-Iron Catalysts
and Related Phenomena

Ken-ichi Aika 1 and Kenzi Tamaru 2

1 Department of Environmental Chemistry and Engineering,
Interdisciplinary Graduate School of Science and Engineering,
Tokyo Institute of Technology, Yokohama, Japan
2Department of Chemistry, Faculty of Science,
Science University of Tokyo, Tokyo, Japan

Contents

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

3.2 Ammonia Synthesis Activity of Elements and Promoter Effects ... 105
3.2.1 Properties of the Elements in the Activation of Dinitrogen . . 105
3.2.2 Properties of the Elements in Ammonia Synthesis . . . . . . . 109
3.2.3 Alloying Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.2.4 Support and Promoter Effects . . . . . . . . . . . . . . . . . . . 114
3.2.4.1 Electron Donation to the Active Center . . . . . . . . 115
3.2.4.2 Structure Sensitivity . . . . . . . . . . . . . . . . . . . . 118
3.2.5 Preparation, Activation and Nitridation . . . . . . . . . . . . . 119

3.3 Mechanism of Ammonia Synthesis over Metals . . . . . . . . . . . . . 124

3.3.l Kinetics of Ammonia Synthesis and Decomposition. . . .. 124
3.3.1.1 Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.3.1.2 Decomposition . . . . . . . . . . . . . . . . . . . . . . . 129
3.3.2 Isotopic Equilibration of Dinitrogen on Various Metals ... 131
3.3.3 Dinitrogen Adsorption Study and Retarding Species. . . 132
3.3.3.1 Ruthenium and Osmium. . . . . . . . . . . . . . 132
3.3.3.2 Tungsten, Molybdenum and Rhenium. . . . . . . . . 133
3.3.3.3 Nickel, Platinum, Rhodium, Palladium, Iridium
and Copper. . . . . . . . . . . . . . . . . . . . . . .. 137
3.3.3.4 Alkali Metals, Alkaline-Earth Metals, Scandium
and Lanthanides. . . . . . . . . . . . . . . . . . . " 137
3.3.4 Absorbed State of Nitrogen . . . . . . . . . . . . . . . . . . . . 139
3.3.4.l Infrared Absorption Spectroscopy . . . . . . 139
3.3.4.2 Electron Spectroscopy . . . . . . . . . . . . . . . . . . 142

3.4 References.. 143

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
104 K. Aika and K. Tamaru

3.1 Introduction

Industrial ammonia synthesis, now known as the Haber-Bosch process, began

in 1913. For this process, the doubly promoted iron catalyst (Fe-AI203-K20)
was synthesized in 1909 and the preparation concepts are still applied today [1].
During the research of that time, most elements other than iron were also
examined. Osmium was found to have the excellent activity, giving as high as
8% NH3 at 550°C and 19 MPa for a long time. Continuous production ofNH 3
(80 gjh) was first demonstrated in July 1909 using Os [2]. However, osmium was
too expensive and has not produced in quantities sufficient for commercial use.
Uranium as its carbide was also very active, but it was irreversibly poisoned by
traces of O 2 of water vapor. These examples suggested to us that we might be
able to develop catalyst containing active elements other than iron. In this
chapter, various catalyst systems other than iron, which have not been applied
commercially, will be reviewed. Several elements, for example ruthenium and
rhenium, are quite interesting as active elements which might lead to second
generation ammonia catalysts.
In the second section, attention is generally focused on catalyst materials
such as the elements, alloys, supports, promoters and precursor materials. How
these materials are related to nitrogen activation and ammonia synthesis will be
shown through the reaction mechanism or creation of active centers. The
developments of material science and preparation techniques lead to new
catalysts for ammonia synthesis, and will be reviewed in this section. The
promoter effect and structure sensitivity are important concepts for catalyst
preparation.
In this third section, classical kinetic studies will be surveyed first. Here,
attention is again focused on the active elements with respect to the important
step, dinitrogen chemisorption. Every element has different characteristics for
ammonia synthesis, because N2 adsorption, NH3 retardation, H2 retardation
and nitridation are each different among the elements. Thus activation condi-
tions and proper reaction conditions may differ. Catalyst materials can be
characterized through instrumental techniques such as TEM, SEM, XPS, AES,
LEED and EXAFS. These techniques, which have been developed by the
advances in surface science, contributed to the understanding of the catalysis
and the detailed mechanism of N2 activation.
Ammonia is used as fertilizer, chemical reagents, on both large and small
scales, and recently as a reagent for NO x removal. The production conditions
can differ depending on the final use. This may demand the availability of
various catalysts, each with different characteristics. Related phenomena with
respect to ammonia synthesis are not discussed in detail in this monograph.
Although the kinetics of ammonia decomposition is described in Sect. 3.3.1, the
surface science study is not reviewed in detail. Ammonia decomposition is easily
studied and is often used as a model reaction in the field of surface science [3,4].
N2 coordinated metal complexes were studied recently in great detail [5, 6].
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 105

Photosynthesis and plasmasynthesis of ammonia could be important in the

future [7-13]. There are many good reviews on ammonia synthesis, mainly
based on an iron catalyst [1, 14-24], as well as other various catalysts
[22, 24- 26].

3.2 Ammonia Synthesis Activity of Elements and Promoter Effects

3.2.1 Properties of the Elements in the Activation of Dinitrogen

The ammonia synthesis reaction from N2 and H2 is composed of several steps,

namely: 1) dissociation of the N == N bond, 2) dissociation of the H-H bond and
3) formation of a N-H bond. Generally speaking, step 1 (dissociation of the
N == N bond) is the most difficult step (rate-determining step) because the bond
energy of N z is the highest among the diatomic molecules (941 kljmol). Thus,
one of the important roles of the catalyst is to dissociate the N == N bond on the
surface (dissociative adsorption or surface nitride formation). A reaction path in
which N2 is hydrogenated prior to N == N splitting is not known in the field of
metal catalysis. Thus, it is important to study the nature of dissociative adsorp-
tion of nitrogen (surface nitride formation) and its reactivity with hydrogen,
which must be related to the catalytic properties of ammonia synthesis. Before
discussing surface nitrides the properties of bulk nitrides are described. Table 3.1
is aperiodic presentation of nitrides and their properties. Most elements form
nitrides. Elements of groups IA and IIA form ionic nitrides. Lithium and group
IA elements readily react with N2 to form stable nitrides, but the heavier alkali
metals do not react directly with N 2. Group IlIA elements also react directly
with N 2, forming stable covalent nitrides. Nitrides of Group IV A to VIII are
interstitial compounds. The affinity toward N2 is still large for group IVA, but it
decreases toward group VIII, where only Fe, Co and Ni form nitrides, but only
from NH 3. Both Fe2N and Fe4N are prepared by reaction with NH3 at
673-773 K [27]. Although Mo 2N can be formed from N 2, nitride formation is
enhanced by Hz [28]. Elements of groups IB and lIB are less reactive toward
N2 and their nitrides, prepared only by indirect methods, are unstable. Group
IIIB elements form stable covalent nitrides.
Those elements active as NH3 synthesis catalysts are found in groups IVA
through VIII, which form interstitial nitrides. Interstitial nitrides have expanded
metal lattices in which nitrogen occupies interstital positions. They are referred
to as "metallic nitrides" because of their resemblance to the metals [29].
Chemisorption is generally easier than the formation of a nitride from N 2,
because the clean surface atoms are more active (having an available bond) than
the bulk metal atoms. Chemisorption of N 2 takes place on some metals even at
room temperature if it is a vapor deposited film. Those metals that can
106 K. Aika and K. Tamaru

Table 3.1. Reactivity of the elements with N2 and properties of their nitrides

Element Reactivity Nitride Standard heat Decomposition

with N 2a of formatiion temperature
(kJjN-atom 25°C) ('C)

IA
Li + + Li3N - 197 (S)
Na Na3N - 151 150
K K3N + 84 L
Rb Rb 3N + 180 L
Cs Cs 3N + 314 L
IIA
Be + Be 3N2 - 285 > 220
Mg + Mg3N 2 230 700
Ca +.+ Ca 3N2 - 213 H
Sr + + Sr3N2 - 197 H
Ba + + Ba3N2 - 184 H
iliA
Sc + ScN -285 H
Y + YN - 301 H
La + LaN - 301 H
IVA
Ti + TiN - 305 H
Zr + ZrN - 343 > 3000
Hf + HfN - 326 H
VA
V + VN -172 > 2300
Nb + NbN - 247 > 2300
Ta + TaN - 243 > 3000
VIA
Cr + CrN - 121 H
Mo + Mo 2N -71 H
W + WN - 71 H
U + UN - 335 H
VilA
Mn + MnsN2 -117 > 1200
Tc TcN
Re Re2N
VIII
Fe Fe4 N -12 440
Co Co 3N
Ni Ni3N +0
Ru
Rh
Pd
Os
Ir
Pt
IB
Cu CU3N + 75 450
Ag Ag3 N + 285 Ex
Au AU3N Ex
liB
Zn Zn3N2 -12 H
Cd Cd 3N2 + 79
Hg Hg3N 2 +8
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 107

Table 3.1. (Continued)

Element Reactivity Nitride Standard heat Decomposition

with Nz" of formatiion temperature
(kJ/N-atom 25°C) (0C)

II1B
B + BN - 134 > 3000
Al + Al - 243 2000
Ga GaN - 105 H
In InN - 21 H
TI Ti3N + 84
IVB
C + (CN), + 155
Si + Si3 N4 - 188 H
Ge Ge3N4 - 17 450
Sn Sn3N4 < 360
Pb
VB
P + PN - 84 750
As
Sb
Bi BiN
VIB
0 + NO +92
S S4N4 + 134 178 Ex
Se Se4N4 + 176 U
Te Te3N4
VIIB
F NF3 -109 S
Cl NCI 3 + 230 Ex
Bi NBr3 + 335 U
I NI3 +272 Ex

a + +: Reacts with N 2 directly below 300 'C, +: Reacts with N 2 directly with N 2 above 300°C,
- : Made from nitrogen compounds, - -: nitride unknown
b L: Low temperature, H: High temperature, S: Stable, U: Unstable, Ex: Explosive

chemisorb N z at room temperature are: IlA (Ca, Sr and Ba), IVA (Ti, Zr and
Hf), VA (V, Nb and Ta), VIA (Cr, Mo and W), VIlA (Re), and VIII (Fe) [30, 31].
Note that they are mostly found in Group IV A through VIII, which are known
to form interstitial nitrides. It is interesting to note that IIA metals chemisorb
N z and these metal "nitrides" are quite effective catalysts for the isotopic
equilibration of N z. However, they are inactive for ammonia synthesis because
they react to form hydrides in an Nz-H z mixture [4, 32, 33J, which will be
shown in Sects. 3.3.2 and 3.3.3.4.
A similar chemisorption can take place on other metals which do not form
a nitride from N z. The much lower ability of other metals to chemisorb
N z seems to come primarily from the difficulty in activating the N z molecule.
Even a copper surface can chemisorb N z when the copper surface is activated by
ion bombardment [34J, even though copper nitride, CU3N, is unstable.
Chemisorption ofN z was found on reduced cobalt oxide with a potassium oxide
promoter at room temperature [35, 36] and even on noble metals (Ru, Rh,
108 K. Aika and K. Tamaru

Os and Ir) promoted with alkali metals. Promoter action will be shown in
Sect. 3.2.4.
The second reason for an inability for N z chemisorption may come from an
insufficient metal-nitrogen bond energy. The heat of chemisorption of N z has
been measured over metals and is shown in Table 3.2. The differential heat of
chemisorption generally decreases with an increase in the surface coverage [37].
Although the initial heat of chemisorption is much larger on vapor deposited
films than on powder or supported metal, the value on the vapor deposited films
decreases more rapidly, presumably because of a larger degree of disorder in
crystallinity [38]. The data shown in Table 3.2 are the initial heat of adsorption
and the desorption energy.

Table 3.2. Observed and calculated heats of dissociative adsorption of N2

Metals Heat of adsorption - 2 x (Heat of

kJjmol nitride formation)'
Calc." Obs, b.,.d., kJjN-atom

IVA
Ti 481 610
Zr 657 686
Hf 816 652
VA
V 469 344
Nb 582 > 502' [39] 494
Ta 732 590 b [40] 486
VIA
Cr 410 439 b [38] 242
Mo 335 263' [41], 289' [42], 259" [43] 142
W 536 397 b [38], 385' [44], 314' [45], 142
389' [46], 334-372c [47]
VIlA
Mn 465 234
Tc 126
Re 167 284-313 c [48]
VIII
Fe 205 293 b [49] 24
Co 134
Ni 138 0
Ru -117 92-167" [50]
Rh -146
Pd - 209 - 159' [51]
Os - 67
Ir -109 242d [52]
Pt -142 25 c [53], 92 d [53]

"Ref. [54]
b Initial heat of adsorption on film at room temperature
c Initial heat of adsorption on filament at room temperature.
d Desorption energy on filament
, Estimation from the heat of dissociative adsorption of NO and O 2
, From Table 3.1
g Ru or RujAl 20 3 with K
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 109

The initial heat of chemisorption of N 2 can be evaluated by a semiempirical

approach [54]. This approach is based on correlations between the initial heat
of chemisorption and the heat of formation of the corresponding compound as
demonstrated for hydrides, nitrides, and oxides by Sachtler and van Reijen [55]
or others [56-58]. The calculated results are shown in Table 3.2 together with
the observed data shown in a review [26, 59]. The calculated values roughly
correspond with the observed value. The negative value of the heat of nitride
formation per N 2 is also shown in Table 3.2. These values are always lower than
the heat of chemisorption. However, the two values are parallel [38, 40, 60]. It is
noteworthy that the initial heat of chemisorption of N2 on Fe is 293 kJ/mol,
whereas the negative heat of formation of Fe4N is only 12.5 kJ/mol. This
indicates that chemisorption evolves a large amount of excess energy due to the
the bonding ability of surface available bonds.
A recent adsorption study on single crystal planes disclosed that there can be
many binding states of adsorbed species on a given surface and that the heat of
adsorption many differ by as much as 80 kJ/mol. Thus, it is not possible to
identify one value for the heat of chemisorption on a given transition metal
unless the binding state is specified or unless it is certain only one binding state
exists [59].
It is notable that the heats ofN 2 chemisorption on noble metals are negative
as shown in the Table. The negative values are caused by the large value of DN2
(bond energy) as compared with the metal-nitrogen bond energy. If the heat of
chemisorption is really negative, the adsorption state of nitrogen atoms on noble
metals should be unstable. Mimeault and Hansen showed that nitrogen atoms
(desorded from a hot tungsten filament (2000 K)) can be adsorbed on an iridium
or rhodium filament at 300 K, as demonstrated by the desorption of dinitrogen
upon flashing [52]. Since contamination of the filament with tungsten vapor
was carefully avoided, the nitrogen atoms must be held by the iridium surface. It
seems that the nitrogen atoms stay on the surface because of a very slow rate of
the second order desorption at lower temperatures (with its high activation
energy). Stimulated nitrogen adsorption by electron bombardment on polycrys-
talline Pd [61], Pd (331) [62] and Pd (110) [63] has been reported. Nitrogen
accumulation on polycrystalline Pd by a reaction of NO with H2 or CO has
been observed [51]. Use of the same reaction also enabled the adsorption of
nitrogen on Pd (100), where a C(2 x 2}-N surface structure was confirmed [64].
Active elements for N 2 activation are summarized in Fig. 3.1 in the form of the
periodic table.

3.2.2 Properties of the Elements in Ammonia Synthesis

Systematic studies of the catalytic activity of single metals ofNH 3 synthesis were
first made by Haber as described in the preceeding paragraphs. In these studies
not only readily reducible metals were tested but also less reducible ones as well,
110 K. Aika and K. Tamaru

IIa rna :szIa Wa :szm

Mn Fe Co Ni

Sr Mo Te Ru Rh

Sa Ce W Re Os Ir

Fig. 3.1. Active elements for N 2 activation

such as cerium which was reduced with magnesium. The results of these early
studies are summarized in Fig. 3.2 [65]. In addition to the metals shown, Re
[66], Cr [67], V [68], Rh [69], Ir [69] and Tc [70] act as NH3 catalysts.
Platinum was also used [69, 71], but has a poor activity. Some of the above
metals, such as Mo, V and U are transformed into nitrides during the reaction.
The reverse reaction, NH 3 decomposition, has been studied over a series
of vapor deposited metal films [72]. The catalytic activities are also shown in
Fig. 3.2.
It is obvious that osmium and iron are the most effective elements under the
conditions studied by Haber. On the other hand, ruthenium is the most active
metal in ammonia decomposition. Since the two reactions are forward and
backward steps of the same reaction, the most active metals should be the same,
at least near equilibrium. The difference disclosed above is probably caused by
some discrepancy in the reaction conditions. It is obvious that the H2/NH3 ratio
is much larger in the synthesis reaction than in the decomposition reaction.
Adsorbed nitrogen and hydrogen often become a retarding species, which
depends upon the H2/NH3 ratio. The effect of such variables will be discussed
later.
In the 1970s, a catalyst system promoted by metallic potassium [73, 74] was
studied. The ammonia synthesis rates at 80 kPa and 588 K over transition
metals supported on active carbon and promoted by metallic potassium are
given in Fig. 3.2 [69]. The activity of isotopic equilibration ofN 2 over the same
series of catalysts at 30 kPa of N2 and at 588 K are shown in Fig. 3.3 [75]. The
same reaction over Raney metals are also shown in this figure [76]. In these
cases ruthenium is the most active metal. There is a common belief that Fe, Ru
and Os are the most active elements in ammonia synthesis, ammonia decompo-
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 111

.
l' ~ ___
15
'Ill
N 14
\
decomposition
'E
u
u
13- ~~I~\'
Q.o
12

'1\
(5

\
E
11 ,, 3
a: ,,
~ 10 ,
9 synthesis 2
5%M - KIA. C. 588K ~

'01
"-o-.. ~ ..
0;-
~

(5
synthesis \
E
2 M powder 773 K ::1.

"0 0 a::
a; 01
':;' 9- Fig. 3.2. Rate of ammonia syn-
M
J: thesis and decomposition over
Z -1
0 various metal catalysts: De-
~
composition at 0.2-0.8 kPa [72],
0 synthesis at 80 kPa (5%
M-K/AC) [69,74] and 5 M Pa
(M powder) [65]

equilibration at 588 K on
Raney M(o). 5% M-K/AC (.)
and SrNo.s(o)
o
4

3
'01

~ 2
(5
E
::1.

a:
01
9-
0

-1
Fig. 3.3. Rate of isotopic equilibration of
dinitrogen e
S N 2 + 30N2 = 2 29 N 2) over

various metal catalysts at 588 K and a

pressure of 20 kPa (150 Torr) [75, 76, 87]
112 K. Aika and K. Tamaru

sltlOn and isotopic equilibration of N z . These elements are known to have

a medium metal-nitrogen bond energy among the elements listed in Table 3.l.
A radioactive element, technetium (Tc, fission product of U), was also proved
to be one of the most active ammonia catalysts. Per-weight activities of Tc
powder, 5.9%Tc/BaTi0 3 , 5.3%Tc-4.1 %BaO/Al z0 3 were comparable with that
of Fe-KzO-AlzOrCaO-SiO z. The results are shown in Table 3.3 [70]. The
supported Tc catalyst was prepared by impregnating with NH4Tc04 and
reducing at 593 to 773 K for 5 h by a Nz-H z mixture. It was reduced almost
completely to the metal with only 0.01 % of the oxide of the technetium
remaining. The catalytic activity is stable over several months. A constant
radiation from 99Tc (E (max) = 0.29 MeV, half life = 2.12 x 10 5 year) emits
counts of 4-8 x 10 3 rad/day. The radiation is thought to create a defect on the
support which may stabilize the technetium metal cluster or create an active
center on technetium which activates dinitrogen [70].

3.2.3 Alloying Effect

As will be described in the section on kinetics, the rate of ammonia synthesis is

a function of the rate of N z chemisorption, the amount of adsorbed nitrogen
(retardation), and the amount of adsorbed hydrogen (retardation). These factors
in turn depend on the reaction conditions (temperature, pressure and flow rate)
and the nature of the element. Thus, we might change the reaction rate or
kinetics on a new active center which is composed of two elements (ensemble
effect). A support and a promoter also influence such characteristics and will be
described in the next section. However, if an alloy is used as a starting material,
and the two elements are separated and turned into an active metal and an inert
oxide, then the resulting activity should be classified as a support effect.

Table 3.3. Ammonia synthesis at 673 K with N2 + 3H 2 = 101 kPa on techne-

tium catalysts· [70]

Catalyst Surf. area k400 b XC

(m2/g) (atmo. s/h)

Tc powder 0.9 6050 0.54

5.9% Tc/BaTi0 3 9.0 4868 0.50
13.8% TC/1'-AI 2 0 3 109 2182 0.36
5.3% Tc-4.1 % BaO/l' - AI 20 3 110 6772 0.56
CA-ld 10.2 5124 0.51

• Space velocity 15000 h - 1, catalyst volume 0.5 ml, catalyst weight 0.5-1 g,
Tc; radioactive element
b Rate constant from Temkin equation (ex = 0.5, k 400 = 15000x2(1 - X2)-1).
Activation energy of k400 = 40-50 kcal/mol
C Relative concentration of ammonia compared to the value at equilibrium

d 31.5%FeO-61.0% Fe 20 3 - 0.81 %K 20 - 3.92%AI20 3 - 2.4%CaO-

0.33%Si0 2
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 113

A number of bimetallic catalysts have been examined for their activity in

ammonia synthesis. An Fe-Mo (1/1) catalyst gives a high activity, although it
decreases remarkably during a prolonged synthesis run unless the Mo content is
a higher than 80 per cent [77]. Ni-Mo, and Co-Mo systems behave similarly.
The catalysts are prepared by calcination of metal nitrate-ammonium molyb-
date mixtures, followed by a reaction with N r H 2 gas [77]. These catalysts
absorb N 2 at the beginning of the operation to an extent that increases with Mo
content, whereas the absorbed N2 is gradually desorded during the operation
resulting in the decrease in activity. X-ray analysis of the catalyst indicates that
the desorption of N2 is accompanied by the formation of Mo-metal (Ni, Co or
Fe) mixed crystals. It appears that the real active component in this system is
molybdenum. The catalytic activity of either iron or molybdenum seems to be
lost upon the formation of a mixed crystal. The results of other metals mixed
with Mo show that Cu behaves as a simple diluent. Mn gives a favorable effect,
and both Cr and Ware unfavorable [67].
The activity of Fe is decreased by addition of Ni [78, 79], and a Mn-Fe alloy
of 30--40 percent Mn gives a high activity [80]. The conventional promoted iron
catalyst is further promoted by the addition of Co. The catalysts are prepared by
burning a Fe-Co alloy in O 2 followed by the addition of promoters [81]. The
alloying effect of Fe-Co, and Fe-Ni was studied in detail [82]. The addition of
3-7 wt% RU02 to a catalyst composed of Co ferrite (25-35%), Mg ferrite
(20-25%), K 20 (0.5-2%), and Fe oxide (rest%) increases the ammonia activity
and heat resistance of the catalyst [83].
Transition metals are also activated by alloying with electropositive metals.
Raney Ru prepared from Ru-Al alloy is more active than pure Ru in terms of
specific activity per surface of Ru, suggesting promotion by residual Al remain-
ing after leaching. It is further activated by the addition of metallic potassium to
give a highly active catalyst which works even at 373 K [84]. CsN0 3 promoted
Raney Ru is also very active [85, 86].
Rare earth-transition metal intermetallics such as CeC0 3 , CeCo s , CeRu2
and CeFe2 are claimed to be more active than doubly promoted iron catalyst on
a unit (BET) area basis for ammonia synthesis under elevated pressures (5 MPa)
and at temperatures up to 873 K, although the actual activity is lower because of
the relatively small surface area of the intermetallics. The rates at 588 K under
a total pressure of 5 MPa were re-calculated and are shown in Table 3.4. Cerium
intermetallics with Fe, Co and Ru were active. Praseodymium intermetallics
with Co and ErFe3 were also active. GdFe3, Gd 2Fe 17, Th 2Fe 17, Er2Fe17,
HoFez, TbMnz, HoMnz, CeCus, LaCus, Y6Mnz3, Tb6Mn23, CeNi, CeNi s ,
CeCo z, CeIn, CeOsz, CeRez, HoCo z and DyC0 3 had lower activities than those
shown in Table 3.4. The intermetallics are decomposed into finely dispersed
transition metals and cerium nitrides during the reaction [88, 89]. The high
specific activities observed suggest an electron donation from the rare earths to
the transition metals which are the active materials [88]. Recently, CeRuz,
CeCo z and CeFez were studied by an in situ powder X-ray diffraction technique.
These intermetallic alloys were found to be converted to cerium hydride
114 K. Aika and K. Tamaru

Table 3.4. Rage of ammonia synthesis on rare earth intermetallics at

N2 + 3H 2 = 5 MPa and Space velocity of 120000 h -I [88]

Catalyst' Surf area Rate at 588 K Act. energy

(m2/g) (mmol NH3/g/h) ( kcal/mol)

Fe-AI 20 r K 20 b 5.21 3.50 20.5

CeFe2 0.45 0.41 13.2
Ce 2Fe!7 1·01 8.0
TbFe, 0.04 23.0
DyFe 3 0.10 7.4
HoFe3 0.02 18.6
ErFe, 1.24 (4.3)
ThFe3 0.60 0.36 10.6
CeRu2 0.33 1.41 10.3
Ce24CoII 1.06 5.6
CeCo 2 0.61 1.45 9.2
CeCo 3 0.12 1.33 8.7
Ce2Co7 0.11 0.30 13.6
CeCo s 0.21 0.73 9.3
PrCo 2 0.38 0.13 14.4
PrCo, 0.12 0.17 11.9
PrCo s 0.20 0.51 11.9

a Catalyst weight ca. 3.3 g,volume ea. lem'

b Catalyst 416 (Fe - 0.97% AI 20, - 0.65% K 20)

(CeH2 +x) and transition metal phases during the synthesis condition
(N2 + 3H 2 = 50 bar, 450 to 550 QC). These phases are considered to be the active
component. However, these states are quite sensitive to oxygen containing
compounds (air, H 20 and CO) forming an inactive phase Ce02/transition metal
[272].
TiFe alloy turned out to be an active ammonia catalyst which is composed of
the mixture of Fe, TiN and TiO x surface phases mounted on a TiFe bulk phase
[30]. However, TiRu has no activity because TiN and Ti0 2 cover the surface of
TiRu and Ru bulk phase [90]. Fe91Zr9 alloy is found to be an active catalyst,
Fe-ZrOx, under the reaction conditions [91]. A Re-Pt bimetallic cluster is
thought to be formed on an Al 20 3 support [92]. It does not seem that alloying
induces drastic effects.

3.2.4 Support and Promoter Effects

Precious metals which can be easily reduced are usually supported on non-
reducible oxides. Starting metal compounds may be reduced, forming metal
cluster particles. The activity is a function of the number of exposed metal
atoms. The particles have various crystal planes on the surface, each of which
may have different activities. The chemical and physical properties of the
support, concentration of metal compounds, and the reduction temperature
may influence the cluster forming process, which in turn influences the size of
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 115

metal particles (structural effect of the support). Supports and the third additives
(promoters) may have some electronic interaction with the metal clusters, which
may influence the activity (chemical effect).
Common metals often form mixed oxides with the support compounds. For
that reason common metals are usually used as massive metal catalysts. In the
case of massive metal catalysts, a few weight percent of a promoter is added.
Some promoters make mixed oxides with the active element and influence the
reduction process and the surface area (structural promoter). Others are depo-
sited on the metal surface and have an electronic interaction with the surface
(chemical effect). Ammonia activity on Fe is known to be enhanced by adding
Al 20 3 and K 20. It is believed that Al 20 3 stabilizes the high surface area of Fe
(structural effect) and K 20 promotes the ammonia activity per Fe surface area
(chemical effect). The structural effect is well studied on Fe single crystal
surfaces, where Fe(lll) is the most active plane and Fe(110) is the next and
Fe(lOO) is the least active plane [93]. Such studies have been expanded to other
catalysts such as Re, and will be reviewed in Section 3.2.4.2.
Various alloy catalysts have been studied as was shown in the previous
section. Raney metal catalyst which has small amounts of aluminum, is classified
as a massive catalyst. However, some intermetallic are transformed to the
supported metal catalysts when one of the components is separated and oxi-
dized or nitrided.

3.2.4.1 Electron Donation to ,he Active Center

A chemical promoter such as K 20 in an Fe catalyst is thought to be an electron
donor to Fe. In the case of Ru, the electron donating properties of the support
and the promoter are quite important for the activity. Table 3.5 and Fig. 3.4
summarize the effect of the electron donor. The compounds' electronegativities,
which are represented as the geometric average of the elements electronegativi-
ties, are 0.8 (K), 1.20 (Cs 2 0), 1.31 (K 2 0), 1.73 (CsOH), 1.80 (KOH), 1.87 (CaO),
2.05 (MgO), 2.2 (Ru), 2.29 (BeO), 2.49 (AI 2 0 3), and 2.5 (C), respectively. Either
the support or the promoter which interacts more strongly with Ru particles is
arranged based on the order of the compounds' electronegativity. The turnover
frequency (TOF) of ammonia synthesis and that of isotopic equilibration of
N2 obeys this order as is shown in Table 3.5. Although the promoter interaction
should be discussed on the atomic scale, a rough criteria of the promoter can be
estimated from the intrinsic chemical properties. It should be noted that ruthe-
nium has quite a low activity when the metal is supported on Al 2 0 3 and has
almost no activity when it is on AC (active carbon). This is ascribed to the acidic
nature of Al 2 0 3 or the electron-accepting properties of AC.
Since the alkali "metal" is a superior promoter, the effect has been studied in
detail. Fig. 3.5 shows the effect ofthe kind and amount of alkali metal promoter.
Unpromoted Ru/AC is not active, however, the activity increases with the
amount of added alkali metal. The maximum activity obtained at 4 mmol-alkali
metaljg-catalyst was 3.30 for Cs, 1.28 for K, and 0.12 mmol NH3/g/h. for Na
116 K. Aika and K. Tamaru

Cs2 0(1.20). CsOH (173)

(5
E K20(1.31). KOH (180)
~

:.:: AI (15)
co
co
III

15 100
!:
o
15
E
~ Ru powder
J:"" (2.2)
Z 10
Fig. 3.4. Rate of ammonia synthesis
at 588 K and N2 + 3H 2 = 80 kPa on
Ru (2 or 5 wt %) catalysts with vari-
ous supports and promoters as a
function of average e1ectronegativi-
ties of compounds [94, 96]

Electronegativityof compounds

70>

~
3
-0
E
E
:.::
co
co
III

15 2
!:
.Q
15

-
E
5
J:
Z
'"
'5
<lI
15 Fig. 3.5. Effect of alkali metal addi-
0::
tion on the rate of ammonia synthesis
No at N2 + 3H 2 = 80 kPa on 2.5%
Ru-AC (1.0 g) at 588 K [73]
4 6
Alkali metal (mmol/g)
Table 3.5. Support and promoter 'effect in ammonia synthesis (N z + 3H 2 = 80 kPa) and N z isotopic equilibration (N2 = 20 kPa) on Ru catalysts
[94,96]

Main promoter Catalyst Ru dispersion NH3 Rate at 588 K N z equilibration at 588 K

(%) (TOF x 104 ) (mmoljh/g)b (TOF x 104 ) (mmol/h/g)b

K(metal) 4.7% Ru-K/AC 10.3 205 3.3

5% Ru-K/AC 12.7 24 0.51
5% Ru-K/BeO 8.6 204 2.9 175 2.5
4.6% Ru-K/MgO 2.9 197 0.87 147 0.66 w
2.0% Ru-K/Al z0 3 5.0 100 0.34 >
0.5% Ru-K/Al z0 3 22.1 24 0.094 S
4.2% Ru-K/CaO 2.5 75 0.26 139 0.49
S
0
i:l
Ru powder-K 0.30 29 0.31 p;'
Ru powder-K 0.25 6.9 0.061 rJl
'<
Raney Ru-K 3.1 17.5 1.76 i:l
Er-
Cs+(Cszo)a 1.96% Ru-Cs+ /MgO 8.5 19 0.083 ~.
1.96% Ru-Cs+ /MgO 4.1 11 0.022 0
1.96% Ru-Cs +/MgO 35 27 0.47 <
2% Ru-Cs+ /A1 2 0 3 20 7 0.073 ""'
Z
0
K+(K 2 O) 2% Ru-K +/Al z0 3 21 3.8 0.046
2% Ru-K +/ Al 20 3 34 7.5 0.146 :t
2% Ru-K +/Al z0 3 18 4.2 0.052 "'i:l0
(")
CaO 4.2% Ru/CaO 2.5 18.7 0.066 00
S'
MgO 4.6% Ru/MgO 2.9 14.3 0.063 -<
1.96% Ru/MgO 8.4 7.2 0.041 ~
00
1.96% Ru/MgO 12.5 3.0 0.026 i:l
P-
Al and/or AI3+ Raney Ru 3.1 1.40 0.144 ~
!l.
None Ru powder 0.30 0.54 0.006 ~
Ru powder 0.25 0.020 0.0002 "P-
"ti
Ru powder 0.18 0.41 0.0025 ::r
i:l
"
BeO 5% Ru/BeO 8.6 1.63 0.023 0
S
Al z0 3 2% Ru/AI 2 0 3 5.0 0.67 0.0023 "i:l00
0.5% Ru/AI 2 0 3 22.1 0.12 0.0005
AC 4.7% Ru/AC 10.3 Neg Neg
118 K. Aika and K. Tamaru

promoted catalysts, respectively. This follows the order of electronegativity (0.7

for Cs, 0.8 for K, 0.9 for Na) or ionization potential of the elements (90 for Cs,
100 for K, 118 kcaljmol for Na). XPS chemical shifts suggest that there is an
electron transfer from the promoter to the Ru surface [94]. Ammonia synthesis
is observed at a steady rate even at room temperature and atmospheric pressure
over 4.5%Ru-K/AC. Since the activity (0.45 Jlmoljh/g) was higher than the
extrapolated value from the kinetic data above 473 K, a different reaction
mechanism was suggested to occur at room temperature [95].
The effect of added metallic potassium is similarly observed with respect to
the isotopic equilibration of N z on Ru/AC [97] as well as on pure ruthenium
[50]. The addition of metallic potassium results in about a 500 fold increase in
the activity of Ru at 673 K, while the increase in the ammonia synthesis rate is
less extensive (25 fold), due to the inhibition by the presence of hydrogen (cf.
Sections 3.3.2, 3.3.3.1) [98]. The inhibition by hydrogen is accounted for by the
competitive adsorption of hydrogen as well as by a decreased adsorption
constant for N z [86, 98].
An analogous catalyst system including graphite has been reported [99]. It is
known that transition metal chlorides are intercalated into the layered lattice of
graphite [100]. On reduction of these compounds, transition metals remain
intercalated as metal atoms giving no activity, although they can become
activated for ammonia synthesis by the addition of metallic potassium [101].
These studies originated from the finding that ammonia was catalytically
produced over electron donor-acceptor complexes of graphite 6r metal
phthalocyanines with alkali metals [102].
The detailed mechanism of the electronic promoter action is not well
understood. Since the apparent activation energy is not changed by applying the
promoter, several mechanisms are suggested; 1) molecularly adsorbed N z,
which is the intermediate species for N(a), is stabilized (increases of the sticking
probability); or 2) generation of active sites by an electronic promoter. The first
mechanism is shown in Fig. 3.6.
Various transition metal/active carbon (AC) catalysts are also promoted by
the addition of an alkali metal [73, 74]. The synthesis activities of transition
metals in this system are shown in Fig. 3.2.

3.2.4.2 Structure Sensitivity

Since Re is not easily reduced completely, the activity has not been reported in
detail. This element becomes active when promoted with metallic potassium.
Rhenium also become quite active when activated in an ultrahigh vacuum
chamber [103], the activity promotion of platinum by a vacuum treatment is
known with respect to the hydrogenation reaction, where chlorine evaporation
may playa role [103, 104]. Polycrystalline Re is reported to be nearly 10 times
more active than Fe(111) at low conversion under 2 MPa and at a temperature
between 603 and 713 K. The activation energy was 16.2 kcaljmol. Although Re
becomes reversibly poisoned by product ammonia (1-2 Torr), exposure to water
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 119

Ru surface

Fig. 3.6. A model of the alkali promoter effect

in the activation of N2 on Ru surface
Ru surface

Table 3.6. Structure sensitivity of ammonia synthesis on

Re single crystals

Face Ratea TOP

(NH3 molec./cm 2 s)

(1121) 314
(1120) 8.5 x 10 16 98
(1010) 2.5 x 1015 6
(0001) 5 X 10 13 0.03

aActivity at 870 K at 20 atm (2 MPa) of N2 + 3H 2 [105,

106]

vapor or presulfidation has little effect on its activity. This is contrast to the
behavior of iron. Ammonia synthesis rates were measured on Re single crystals
at 870 K under 2 MPa of N2 + 3H 2 . The results are shown on Table 3.6. The
basal plane (0001) was inactive. The open and rough planes (1010), and espe-
cially (1120) or (1121) were quite active for this reaction. The activity on (1121)
was ca. 104 times greater than that on (0001). The activation energy was
19.4 kcaljmol for all of the surfaces. The structure sensitivity of Re surfaces was
verified for ammonia synthesis [105, 106].

3.2.5 Preparation, Activation and Nitridation

As shown in the previous section, ammonia catalysts are classified as two types:
The supported metal catalyst and the massive metal catalyst. For the supported
120 K. Aika and K. Tamaru

catalyst, metal salts such as chlorides or nitrates are impregnated on the support
materials and usually reduced with H2 or N2/H2 mixture. Promoter salts,
usually nitrates, are also impregnated on the supports. Mixed metal compound
salts, which are composed of metal elements and promoter elements, can also be
used as the starting materials. If the chloride ion may possibly remain after the
reduction, and its poisoning is not negligible, compounds free of chlorine should
be used. The starting materials and preparation methods may influence the
activity through the degree of metal dispersion or metal-support interaction as
was shown in the previous section.
Massive metal catalysts are usually made from a mixture of both metal and
promoter oxides by reduction with H2 or N2/H2 gas. Many studies of activation
processes have been carried out on Fe catalysts which are written elsewhere
[107]. The commercial iron catalyst needs to be reduced for a long period
because it is in a massive form and the oxide is not easily reduced completely.
The iron catalyst is activated under the ammonia synthesis condition [108],
during which the surface structure probably changes to form the active sites
[109]. On the other hand, supported precious metal catalysts such as Ru catalysts
need a short reduction time because their starting compounds are easily reduced.
Generally, they have no induction period with respect to the activity.
Alkali metals, especially potassium, were proved to be excellent promoters
for Ru catalysts. However, since metallic potassium is very reactive to water
vapor or oxygen-containing gases, the catalyst system has several problems,
such as a handling problem or deactivation. Several other preparation methods
were developed. K4Ru(CN)6 was prepared by alkali fusion of Ru metal and
KN0 3 or by adding KCN to RuCl 3 [110]. Although the hydrogenation of
K4Ru(CN)6 alone was very slow even at 773 K, alumina supported K 4Ru(CN)6
was readily hydrogenated at 653 K. The reaction stoichiometry was suggested
as follows:

It should be noted that K 4Ru(CN)6 gives metallic potassium and ruthenium on

an alumina support, although KCN cannot be reduced to K by hydrogen alone.
Activated K4Ru(CN)6/AI203 catalysts gave catalytic activities for ammonia
synthesis higher than RuCI 3 ' 3H 20/AI 20 3, and the activities correspond to 1/6
to 1/2 of that of Ru-K/AI 20 3, one of the most active catalysts [110]. This high
activity of K4Ru(CN)6/AI203 was irreversibly poisoned by water vapor, which
suggested the existence of metallic potassium [110]. Various alkali metal or
alkaline earth metal salts of hexacyanometal (metal: Fe, Os, Co) were activated
on Si0 2 or other supports and showed ammonia activities [111, 112]. Alkali
salts such as Na2C03 were effectively added to such complex salts [113].
CsN0 3 was also found to be an excellent promoter [110, 114]. Hydro-
genolysis of CsN0 3 only occurs in the presence of Ru metal as hydrogen
activator.
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 121

Thermal gravimetric analysis suggested that Cs 2 0 was partly converted to

CS 2 0 2 and Cs, and finally to CsOH. The activity of Ru-Cs +/MgO (Cs + means
Cs 2 0 and/or CsOH), which is shown in Table 3.5, is quite high. It is second to
a Ru-K/Support catalyst. This promoter is stable in the presence of water vapor.
The activity can be controlled by the amount of nitrate addition, which depends
on the surface area of the support. In the case of RU3(COhz supported on
a graphite containing carbon support, RbN0 3 was effective as a promoter [115,
116]. CsN0 3 promoted RU3(COhz/MgO was also quite active [117]. Careful
studies showed that a small amount of CsN0 3 was enough to give the highest
activity for Ru/MgO as is shown in Fig. 3.7. 5% Ru-Cs+ /MgO was as active as
4.7%Ru-K/AC as is shown in Fig. 3.8 [117]. Since Ru-Cs+ /MgO is not
poisoned irreversibly by oxygen containing compounds, it is hoped to be one of
the second generation catalysts after the iron catalyst. Barium compounds as
well as an alkali salt were also useful promoters in a RuCl 3/carbon catalyst
[118]. RbN0 3 and Ba(N0 3h are effective for Ru/AC which is prepared from
RuCl 3 [119]. Lanthanide nitrates were found to be effective promoters for
RU3(CO)u!Alz03 [120, 121].
CsN0 3 was also proved to be an excellent promoter for Raney Ru catalyst.
The activity was compared with other alkali metal nitrate-promoted catalysts
and is shown in Fig. 3.8. CsN0 3 was suggested to remain mainly as Cs + (CszO,
CsOH) and partly as Cs metal because water vapor decreased the activity. The
activity sequence is in the order of electropositivity (Cs > Rb > K > Na). Since
Raney Ru contains aluminium (mainly Alz0 3, some AI), CsN0 3-Raney Ru
quite resembles the Fe-AlzOrKzO catalyst with respect to the composition.

F', ,
I
,\

-;- \\
0>
It
~ \
\
"0
E
\
::l.
....
:>::
a)
a)
Ln

"0 400
~
0:: o
Fig. 3.7. Effect of CsN0 3 addition on am-
200 o monia synthesis at 588 K on 2 wt % Ru cata-
lysts; Ru3(CO)12/support is treated with Hz
at 623 K for 4 h. Ru-CsOH/MgO (0 and e);
Ru-CsOH/AI 2 0 3 (0)
12
Cs/Ru mol ratio
122 K. Aika and K. Tamaru
Rate at 573K(mmol ' h-I .g-I)
....
o
Fe
Z
N
Fe-AI 2 O:3-K2 0 •
eN
Ru N
:I:

....ZI
4.7"10 Ru-A.C.- K
Raney Ru
:I:
Raney Ru-K Co>

Raney Ru-NaNO:J
Raney Ru- KN0:3
Raney Ru-Rbt)lO:J
Raney Ru-CsN0 3
5.0"10 Ru-Cs+'MgO

Fig. 3.8. Rate of ammonia synthesis at 573 K and 80 kPa on various Fe and Ru catalysts. Fe [123J,
Fe-AI 2 0 3 -K 2 0 [108, 124J, Ru and 4.7% Ru-A.C.-K [73J, Raney Ru and Raney Ru- K [84J,
Raney Ru-alkali nitrate [85, 86J, 5.0% Ru-Cs + j MgO [117J

CsN0 3 -Raney Ru is more active than the commercial iron catalyst and is one
of the most active ammonia catalysts under atmospheric pressure conditions
[85, 86J (see Fig. 3.8). However, the Ru catalysts might not be highly active at
high pressure because of hydrogen inhibition. Nevertheless, a recent work
suggests that hydrogen promotes the activation of N2 on Ru catalysts if
a lanthanide promoter is present [122].
Elements between groups lIlA and VIlA usually become nitrides under
ammonia synthesis conditions. Nitrides of uranium and molybdenum are
known to be active catalysts. When Mo0 3 is reduced with H 2- CH 4 mixture,
a high surface area M0 2C is formed. The oxide is first heated to a temperature
75 K below the temperature at which reduction of the oxide and reaction with
CH 4 occur. The temperature was then raised at a rate of 0.1- 5 K/min until the
reaction ceased (1041 to 1273 K). The quenched sample (M0 2C) had a surface
area of 51 m 2/g, which was higher than the sample (13 m2/g) made by the usual
method. The activity of the M0 2C was higher than 5% Ru/ Al z0 3 , but was lower
than Fe- Al z0 3-K 20 catalyst [125, 126]. Oxycarbonitride of molybdenum
(MoxCyN z ) was also active for ammonia synthesis [127]. The rate of ammonia
synthesis on reduced molybdenum (Mo), its carbide (MozC), and the oxycarbide
(MoOxCy) was measured with simultaneous gravimetric determination of the
nitrogen uptake of the catalysts. On prereduced M0 2C and MoOxCy the activity
increased with time as reduction proceeded and reached a constant value when
the catalyst had taken up about one monolayer of nitrogen, corresponding to
a surface stoichiometry of MoN. On Mo the steady state was reached after
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 123

taking up the same monolayer of MoN plus two layers of M0 2N. In all three
cases the steady-state rates were found to be very similar. The small amounts of
nitrogen taken up to reach similar rates suggested that the bulk plays no role in
determining the catalytic activity [128].
With respect to ammonia synthesis at 773 K, tungsten carbide (WC, 2m 2/g)
also seems to behave like tungsten metal (W), although their physical properties
differ considerably. 390 Pa of ammonia was produced at a contact time of 0.10 s
under N2 + 3H 2 = 101 kPa and at 773 K. H 2S adsorption on only a fraction of
the surface of WC inhibited the ammonia synthesis. This suggested that the
three side planes (l010), (1100) and (0110) of hexagonal WC crystal exposed
tungsten atoms in the uppermost layer. These faces would not hold carbon
atmos strongly and the "metallic" face was considered to be the active sites for
ammonia synthesis [129].
Atomically dispersed low-valency Mo, W, V and Zr ions can be deposited on
surface oxides with the use of n-allyl (Mo, W, and Zr) and cr-benzyl (V)
compounds and subsequent hydrogen treatment. Mo(C 3Hs)4 reacted with
surface hydroxyl ions and formed Mo(II). Such heterogenized metal ion cata-
lysts showed some ammonia activity, although the levels were low. Some
examples are listed in Table 3.7. Among four ions, Mo(II) ions were the most
active. TOF (NH3 molec./Mo ions/s) did not change when the Mo ion concen-
tration was increased from 10 17 to 10 18 ions/m 2-Si0 2. The activation energy
(25 kcal/mol) was also constant. The activity of the deposited Mo ion was shown
to depend on the nature of the oxide carrier as is shown in Table 3.7 [130]. The
reaction mechanism on transition metal ions might be different from that on
reduced metals [5].

Table 3.7. Activity of atomically dispersed transition metal ions (Me) in ammonia synthesis
N2 + 3H 2 = 5 MPa [130]

Me/wt.% Support/m2 g-l Reduction Reaction Contact Conversion TOF x 104

temp. (K) temp. (K) time (%)
(104 h)

Zr/4.3 A1 20 3/200 823 823 1 0.002 0.46

V/2.0 Si0 2/250 823 723 1 0.007 0.16
W/1.5 Si0 2/250 873 723 1 0.015 2.2
Mo/4.3 Si0 2 /250 823 723 1 0.7 320
Mo/4.3 Si0 2 /250 823 673 1 0.135
Mo/7.1 A1 2 0 3/200 823 723 1 0.012
Mo/2 Cr 20 3/19 673 673 0.6 0.024
Mo/0.5 Fe203/21 483 483 0.6 0.0040
- /0 Fe203/21 483 483 0.6 0.0
Mo/l Co 30 4 /11 473 473 1.9 0.0013

TOF: NH3 molec./Me ions

124 K. Aika and K. Tamaru

3.3 Mechanism of Ammonia Synthesis over Metals

3.3.1 Kinetics of Ammonia Synthesis and Decomposition

3.3 .1.1 Synthesis

According to Temkin and Pyzhev [131], the rate of ammonia synthesis is
controlled by the rate of N2 adsorption (step 1), whereas the H2 adsorption
(step 2) and the hydrogenation of adsorbed nitrogen (step 3) are fast enough to
be in equilibrium.
N2 -> 2N(a) (1)

H2 ~ 2H(a) (2)

N(a) + 3H(a) ~ NH3 (3)

The rate of adsorption of N 2(Vs ) is expressed as follows:
Vs = k[N 2] exp( - ge N) (4)
where [N 2] is the N2 pressure, eN is the coverage of adsorbed nitrogen, and
k and g are constants, eN is a function of the fictitious N2 pressure [N 2]* which
is in equilibrium with H2 and NH 3.
eN = (1jf)ln(A[N 2]*)
(5)
where A is a constant and K is an equilibrium constant of ammonia decomposi-
tion. Thus the synthesis rate is expressed as follows:

Vs = k[N2](AK[NH 3]2j[H2]3)-g/f
= k s [N 2] ([H2rj[NH 3]2)g/f (6)

The corresponding rate ofN 2 desorption which is assumed to be rate-determin-

ing in ammonia decomposition is similarly given by
(7)

The Temkin and Pyzhev equation expressed above was in agreement with
a number of kinetic measurement made on various catalysts as summarized in
Table 3.8 for the synthesis and in Table 3.9 for the decomposition. One
characteristic feature of the ammonia synthesis rate is the retardation by the
product ammonia, and this is reasonably explained by the Temkin theory. The
basic assumption of the rate-determining step, N 2 chemisorption was also
supported.
Although Eq. (4) was successful in described a number of experimental data,
the experimental values of gjf are not always the same. Sometimes it ranges from
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 125

Table 3.8. Kinetic parameters of NH3 synthesis according to the temkin-phzhev equation

Metal Temperature Pressure glf Eact 3 Ref.

(K) (MPa) (kJ/mol)

VIA
Mo 720-823 0.1 0.5 178 [132]
W 855-951 0.1 0.5 190 [133]
W 673-773 0.1 0.5 (203) [129]

VII
Tc 623-698 0.1 0.5 190 ±20 [70]

VIII
Fe( - AI 2 0 3-K 2 O) 673-723 0.1 0.5 167 [131]
723-773 30 0.5 176 [134]
Fe( - A1 2 0 3) 645-651 0.1 0.5-0.6 [135]
Fe( - MgO) 570-680 0.1 0.5 (156) [136]
Ru 881-1013 0.1 0.5 249 [137]
Os 673-723 0.1 0 164 [138]
823-873 0.1 0.5 174 [138]

a Activation energy of NH3 decomposition (kd)

0.5 to 0.8 depending on temperature. This is partly natural because Eq. (4) is
valid only for an intermediate range of coverage (eN = 0.2-0.8). Thus, when
[NH3J in the gas mixture is very low or very high, the results tend to give lower
or higher value of g/f [124, 163]. Even with moderate coverage, the rate
constants given by the Temkin equation are dependent on the H2/N2 ratio,
and/or total pressure [21, 135, 164, 165]. Thus, the value of g/f obtained from
rate data can be an average value under the experimental conditions, and is not
a constant as postulated in the original theory. However, this rate expression is
widely used in commercial operations at high pressure with an Fe catalyst.
If the nitrogen adsorption is described by the Langmuir equation:
V = k [N2J (1 - e N )2, the rate of ammonia synthesis is described as follows:
(8)
This equation can be applied even under low NH3 pressure. When various
catalysts are examined below a total pressure of 1 atm, the activity (or NH3
pressure on the catalyst) is sometimes quite low. Under these conditions, the
Langmuir type equation is quite often used.
Equation (8) is based on the mechanism (1 to 3), where N(a) is the main
retarding species. If NH(a) is the main retarding species, the equation can be
written as follows:
(9)
In the case of Temkin expression, the equation is written as follows:
Vs = ks[N2J([H2J 2/[NH 3J 2)g/f (10)
 ......
tv
a-.

?"<:

~
::s
''""
Q..

?"<:
--l
Table 3.9. Kinetic parameters of NH3 decomposition according to the equation V = k(PNH3)m/(PHZ)" 8'"
....
'"s::
Metal Temperature (K) Pressure m n n/m E..I Ref.
(kPa) (kJ/mol)

VA
V 673-753 ca. 100 (Low P HZ ) 0.5 0 0 138 [139]
673-753 High P HZ 1.0 1.5 1.5 138 [139]

VIA
Mo 1027-1273 13.3 0 + 134-180 [140]
W 548-873 2-20 0-1 205 [141]
903-1023 4.7-20 0 146 [142]
953-1153 0.1 0 0 176 [143]
904-1214 7-26 0 0 163 [144]
1073 - 1523 2-35 0 + 146-197 [140]
950-1150 1-5 0 0 113-130 [145]
1073 10- 9 -10- 6 0.8 [146]

VIlA
Re 653-713 ca. 100 0.53 0.89 1.68 134 [147]
728-843 1.3-7 0.7 1.4 2.0 205 [72]

VIII
Fe 608-743 1-4.4 0.5 0.7 1.4 163 [148]
653-773 1.3-10.6 0.9-1.0· 1.4-1.5 1.5 176-209 [149]
765-797 ca. 100 0.9 1.5 1.67 226 [150]
Fe( - Al 2 0 3 608-703 100 0.6 0.85 1.42 192 [151]
-K 2 O) 693 13 0.48 0.72 1.5 [152]
752 13 0.75 0.38 0.5 [152J
CO 643-753 1.3-7 0.85 1.42 1.67 188 [72J
Ni 573 0.2-5.3 1.0 1.5 1.5 [153]
663-773 1.3-7 0.96 1.53 1.59 180 [72]
623-723 4xlO- 3-O.17 1 [154] w
Ru 543-738 1.3-7 1.2 2.0 1.67 188 [72] >
623-673 80--106 0.6 0.9 1.5 130 [155] 3
825-1009 1.0 1.75 1.75 247 [137]
30
::l
Rh 693-773 1.3-7 1.35 2.45 1.81 239 [72] ~.

Pt 763-833 0.001 1.5-2.0 2.2-3.0 1.5 205 [156] C/l

'<
793-883 1.3-7 0.96 1.63 1.7 247 [72] ::l
978 10- 5 0.63-0.84 0.25 0.34 17 [157] ;.
~
1045-1131 Low PH2 1 1 1 167 [158] tn·
1045-1131 High PH2 1 0 0 167 [158] 0
<0
1100--1485 0.03-0.5 1 1 1 184 [159] ~
'"
1100--1485 1.3-39 1.4 2.3 1.64 419 [159] Z
0
1206-1488 13-26 1 1 1 586 [144]
1173-1623 ca. 10- 3 1 0 0 21 [160] ~
~
0
::l
IB (")
Cu 768-893 5--67 1 1 192 [161] '"
§:.
993-1183 100 1.5 1.5 243-259 [162] '<
~
::l
'"
0.
)1:1
!!.
'0.ft"
'1:1
::r
::l
'"
0
3
::l
'"
'"
,...
IV
-.l
128 K. Aika and K. Tamaru

Practically, the power rate expression is convenient. This is evident when the
mechanism is complicated.
Ru catalysts for ammonia synthesis were found to be retarded by hydrogen
adsorption. In this case both N(a) (or NH(a)) and H(a) may retard the reaction.
Vs = k[NzJ (1 - eN - eH)Z
(11)
Since this expression is complicated to use in analysis, it can be changed to
the power rate form although it is not correct algebraically.
Vs = k[NzJ(K 1 [NH3J [Hzr 1.5)-ZY (K z [H zJO. 5 )-2z
(12)
Table 3.10 shows the kinetic parameters of ammonia synthesis over Ru
catalysts. On Ru powder and Ru-alkali/support, hydrogen has a negative order
and the ammonia order is close to zero. This suggests strong hydrogen inhibi-
tion and weak nitrogen inhibition. On the other hand, it is suggested that the
nitrogen inhibition is extensive and hydrogen inhibition is weak on the support
Ru catalysts [122, 166]. The extent of ammonia inhibition also depends on the
kind of promoter as is shown in Table 3.10 [92, 167, 170].
When the synthesis is carried out far from equilibrium as happens by
increasing the space velocity or by decreasing the reaction temperature, two
steps can be the rate-determining. Under such a condition where the reverse rate
is negligible, the overall rate (Vs) in the steady state is equal to the rate of

Table 3.10. Kinetic parameters* in ammonia synthesis on various Ru catalysts

Form of Temp. Press. n h a Ref.

Catalyst (K) (kPa)

Ru Powder 598 80 -0.72 -0.15 [166]

553-581 101 -1 0** [192]
Ru/Si0 2 825-1009 101 1.25 -1 [137]
574-671 101 - 0.56 [167]
Ru/Celite 619-683 80 0 0 [168]
Ru/Al z0 3 610-691 101 - 1.0 [167]
638 80 1 0.30 - 0.74 [166]
500-725 150 1.1 0.6 [169]
Ru/MgO 593 80 1 0.66 -0.78 [166J
500-725 150 0.9 0.2 [169]
Ru-Cs+ /Al z0 3 595 80 1 -0.55 0 [166]
Ru-K+/AI 20 3 500-725 150 1 - 0.3 [169]
Ru-Cs+ /MgO 548 80 1 - 0.43 - 0.12 [166]
Ru-K+/MgO 500 -725 150 1 - 0.5 [169]
Ru-powder-Cs + 598 80 1 - 0.90 - 0.07 [166]
Ru-K/AI 20 3 500-725 150 0.9 -1.1 [169]
Ru-K/MgO 500-725 150 1.2 - 0.5 [169]

* V = k[N2]n[H2]h[NH3J"
** High flow rate (100-1000 mlfmin) condition
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 129

nitrogen chemisorption (V n), and possibly to the rate of hydrogenation of

adsorbed nitrogen, step 3, (Vh).
If Vn and Vh are given by:

Vn = k n[N 2](1 - e N)2 = k n[N 2](1 - 2e N)

Vh = k h [H 2 ]eN (13)

and ifVs = Vn = Vh
Vs is given by eliminating eN

Vs = k nk h[N 2] [H 2] (2k n[N 2] + kh[H 2])-1

= k'[N2]m[H 2]n (15)
where m and n are apparent orders of reaction with m < 1, n < 1, and
m + n = 1. Temkin et al. found kinetics corresponding to Eq. (15) with
m = n = 0.5 under reaction conditions far from equilibrium with very high
space velocities of about 106 h- 1 at 627-723 K on an iron catalyst [171]. The
rates of each elemental step can be better elucidated by a dynamic approach,
measuring the reaction rates as a function of adsorption amounts during
catalysis. At a sufficiently low pressure of H2 and high temperature, the rate of
hydrogenation was found to be lower than the rate of N 2 chemisorption on
doubly promoted iron catalyst by measuring the adsorption during the reaction
[172].

3.3.1.2 Decomposition
The rate equation of ammonia decomposition was expressed in Eq. (7) accord-
ing to the Temkin-Pyzhev mechanism (Eq. 1-3). The observed parameters in
Table 3.9 show that n/m values are mostly 3/2, which indicates that the
Temkin-Pyzhev mechanism is applicable on most catalysts (Re, Fe, Co, Ni, Ru
and Rh) under normal ammonia decomposition conditions.
However, the hydrogen order (n) is often zero for the reaction on Wand Pt
catalysts, for which a different mechanism should be applied [173].
Ammonia decomposition on a clean tungsten surface was studied by direct
and simultaneous measurements of the reaction at 773-1473 K under an ammo-
nia pressure of 10- 6 -10- 3 Pa [174]. The order of the reaction rate with respect
to ammonia pressure changed with temperature from first order at 1478 to 2/3
order at 773 K. No hydrogen was adsorbed on the surface in any form above
973 K and an increase in the hydrogen partial pressure (P H2 ) during the reaction
had no effect on either the reaction rate or on the amount of surface nitrogen
(eN)·
Under higher ammonia pressures, up to 100 Pa, thick surface nitride layers
were formed during the decomposition, which were decomposed at the same
rate in vacuo as in the steady state NH3 decomposition, provided that the
uptake of nitrogen was the same.
130 K. Aika and K. Tamaru

The rate of nitrogen desorption from the surface depended only on eN,
being faster with increasing eN.
(16)
whereas the rate of nitrogen uptake decreased with increasing eN and increased
with ammonia pressure (P NH3 ).
(17)
On the basis of these data, it was concluded that the overall reaction proceeds
through a reaction mechanism of "dynamic balance" between two consecutive
"rate-determining steps: 2NH3 --+ 2N(a) + 3H 2(V a ) and 2N(a) --+ N2 (V d). The
rate of the overall reaction (Vs) is obtained as follows:

(18)

Here no hydrogen order is reasonably explained [174, 175]. The isotope

effect observed with NH3/ND3 is consistent with the above theory [146]. Such
kinetics are obtained when the rate of hydrogenation of N(a)
(2N(a) + 3H 2 --+ 2NH 3; Vh) is negligibly lower than Yd' This mechanism (tung-
sten-type mechanism) is seen at high temperatures and low pressures on W, Pt,
and even on Fe [176, 177].
The kinetics of ammonia decomposition on platinum wire under pressures
below 1.3 Pa are found to conform to that of the rate-determining dissociation
of NH3 at around 1273 K, while to that of desorption of N z at around 773 K
[156, 157]. Details of ammonia decomposition were studied over a poly crystal-
. line Pt wire in a continuous flow microreactor [53, 176, 177]. Kinetics were
analyzed by the steady-state intermediate method. For the high temperature
and/or the low pressure (for example, 550-1200 K at 10- 3 Torr or 550-1400 K
at 10- 6 Torr), a surface reaction involving the dissociation of the first N-H bond
controls the rate of ammonia decomposition (V h« V d, tungsten-type mechan-
ism). The activation energy in this regime is 4.2 kcal/mol, and the rate of
decomposition exhibits a primary isotope effect with a first order dependence of
ammonia pressure. For the low temperature and/or the high pressure (for
example, 480-550 K, at 10- 3 Torr or 550-800 K at 0.5 Torr), the surface is
nearly saturated with nitrogen, and the rate of nitrogen desorption controls the
rate of reaction (Vh » V d ).
Consequently, the reaction rate is independent of ammonia pressure, and the
observed activation energy, 22 kcaljmol, is equal to the activation energy for
desorption of nitrogen. Including several other works on Pt, the activation
energies (Ea) and desorption energies (Ed) of the elementary steps are estimated
as follows [53]:
NH3(g) ±:+ NH 3(a) ilH = - Ed = - 12 kcaljmol (19)
NH3(a) --+ NHz(a) + H(a) Ea = 16 kcaljmol (20)
Ea = 17.2 kcaljmol (21)
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 131

2N(a) -+ Nz(g) Ea = 22 kcal/mol (22)

N z -+2N(a) Ea = 16 kcal/mol (23)
2H(a) ~ Hz(g) Ed = 19 kcal/mol (24)
This value can provide a one-dimensional potential energy diagram of the
reaction. Ammonia is assumed to be adsorbed with a sticking probability of
unity [53].
Over the Ru (1,1,10) surface, the tungsten-type mechanism is operative on
its terrace sites, whereas the Temkin and Pyzhev type is operative on the step
site under similar condition [175]. Sebba and coworkers investigated ammonia
decomposition over the metal nitrides TiN, VN and Cr3NZ, finding VN the
most active [118]. The kinetics of ammonia synthesis [68] and decomposition
[139] on VN at around 773 K follow the Temkin-Pyzhev equation, while the
VN catalyst was contaminated with oxygen. At lower pressures of hydrogen the
decomposition rate on VN has a zero order for hydrogen [139], suggesting
a tungsten-type mechanism. The kinetics of ammonia synthesis over uranium
nitride (U ZN 3-UN z) at 648-823 K and 3 MPa [179], as well as over molyb-
denum nitride (MozN) at 720:.-828 K [132], also obey the Temkin-Pyzhev
equation.

3.3.2 Isotopic Equilibration of Dinitrogen on Various Metals

If the chemisorption or dissociation of dinitrogen is rate-determining in the

ammonia synthesis as has been demonstrated, the isotopic equilibration of
dinitrogen
(25)
must be a slow process. The two reaction rates must not be exactly the same
because the surface in isotopic equilibration is more strongly covered with N,
and the surface in ammonia synthesis must be partly covered with hydrogen.
Kummer and Emmett confirmed that the rate of equilibration is comparable
with the rate of nitrogen desorption from the same Fe catalyst at 773 K [180].
Catalysts which are active for nitrogen equilibration are also active for ammonia
synthesis. Known active metals include Fe [181-183] Os [184], Co [185],
Raney Ni [185], and others [76]. The equilibration on Os is measurable at
a temperature as low as 473 K, and is rapid at 573 K [184]. The equilibration
requires higher temperatures over Re [147], W [186-187] and Mo [188-190].
The activities of transition metals on carbon promoted by potassium for
ammonia synthesis [21] and for isotopic equilibration [75] are shown in Figs.
3.2 and 3.3. The relative activities run parallel in the two reactions. Generally,
ammonia activities are higher than the isotopic equilibration. The reasons are
suggested to be: a) the adsorption step is in a dynamic state under ammonia
132 K. Aika and K. Tarnaru

synthesis, whereas it is in the adsorption equilibrium under isotopic equilibra-

tion, and 2) possibly, incomplete reduction of the surface in isotopic equilibra-
tion. The rates of nitrogen isotopic equilibration on Raney metals are not high
except for Raney Ru, probably due to the second reason. However, the rate is
inversely related to the desorption energy of atomic nitrogen. Adsorbed states
on Raney-type metals are studied by the TPD method. Atomically adsorbed
species, which is related to isotopic exchange, is believed to be the active
intermediate species for ammonia synthesis [76].
The rate of isotopic equilibration is the rate of forward and backward
reaction of step (1). If the rate-determining step of ammonia synthesis is step (1),
the rate of isotopic equilibration (backward reaction) reaches zero at ammonia
synthesis conditions. In this way the rate-determining step was determined on
Ru-K/AC [98J and on M0 2N [191].

3.3.3 Dinitrogen Adsorption Study and Retarding Species

3.3.3.1 Ruthenium and Osmium

N2 adsorption and hydrogenation of absorbed nitrogen was studied in detail on
Ru powder [192]. One g of Ru powder (0.1 m2/g) gives an ammonia activity of
0.011 mmol/h/g at 573 K under 1 atm ofN 2 + 3H 2. Under this condition, 1 g of
Fe powder (0.1 m2/g) gives the activity of 0.040 mmol/h/g when the ammonia
. retardation is extrapolated to zero. Thus, Ru seems to be less active than Fe.
However, when N2/H2 ratio is increased, the activity of Ru surpasses that of Fe
because the rate of nitrogen adsorption (the rate-determining step) increased
with the decrease of hydrogen retardation. The rate of nitrogen adsorption at
[N2J = 0.027 atm and zero coverage of nitrogen in the absence of H2 was
estimated to be 40 mmol/h/g at 573 K. This value is 10 3 times higher than the
ammonia activity (N2 adsorption rate) of Fe. Thus, the intrinsic activity of
N2 activation of Ru is 10 3 times higher than Fe. The former suffers from
hydrogen inhibition, while the later does not. The rate of hydrogenation of
adsorbed nitrogen is so fast on Ru powder above 431 K that it is controlled by
the flow rate of hydrogen and the equilibrium ammonia pressure. The true rate
is so fast that it is measurable only at low temperature (203 to 298 K). Both the
rate of N2 adsorption and the rate of hydrogenation of adsorbed nitrogen are
much faster on Ru than those on Fe. However, the rate of ammonia synthesis on
Ru is comparable with that on Fe because of the hydrogen inhibitation on Ru
[192]. An ammonia process in which N2 and H2 is supplied alternatively was
proposed [193, 194].
The kinetic analysis of this reaction was also studied on 1 g of Os powder
(7.5 /lmol of sites/g) [195J, the rate is expressed as follows:

(26)
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 133

The reaction is therefore limited by N2 adsorption and inhibited by H2.1t is also

inhibited by NH 3 , but to a lesser degree. The high rate of hydrogenation of
adsorbed nitrogen even at 473 K exemplifies the rate-determining step of
N2 adsorption. The situation is quite similar to Ru powder. The rate of
ammonia synthesis on a bare site of Os (rate of N2 adsorption) is 100 times
higher than that of Fe but 100 times lower than that of Ru at 473 K. However,
under 1 atm of a stoichiometric mixture and at 673 K, Os is 100 times less active
than Fe powder [195].
The rates of isotopic equilibration are retarded by hydrogen on Ru catalysts:
Ru-K, Ru-K/AI 2 0 3 , Ru-K/active carbon [98] and Raney Ru-CsN0 3 [86]. This
corresponds to the hydrogen retardation for N2 adsorption on Ru powder [192]
or to the negative order of hydrogen pressure in the ammonia synthesis rate
expression (see 3.3.1). In the case ofCsN0 3 promoted Raney Ru (one ofthe most
active ammonia catalysts, the rate of isotopic equilibration is expressed as being
proportional to (1 + b[H 2]o.S)-2 [86]. The reaction order in the isotopic
equilibration of N2 on Ru catalysts increases with temperature as follows:
from 0.26 at 553 K to 0.45 at 593 K on Ru-K/AI 2 0 3 , from 0.44 at 593 K to 0.79
at 653 K on Ru-K, and first order on Ru/AI 20 3 at 653-713 K [50]. The kinetics
on Ru-K and Ru-K/AI 2 0 3 are represented by a Langmuir type equation for
dissociative adsorption:
(27)
The heat of chemisorption of nitrogen on Ru-K is estimated from the adsorption
constant in the above equation to be 167 kllmol. It is to be noted, however, that
both isotopic equilibration and N2 chemisorption are enhanced by hydrogen on
potash-promoted iron catalysts [26]. No enhancement by hydrogen was found
on pure iron [181, 182]. Recently, N2 isotopic equilibration was found to be
enhanced by hydrogen on rare earth-promoted Ru catalysts [122]. The hydro-
gen effect on Ru seems to depend on the kind of support and promoter. The
guiding principle of this phenomena has not been solved, yet, the study is quite
important to construct an effective Ru catalyst under high pressure.
It is interesting to note that a clean single crystal of Ru (0001) does not
adsorb nitrogen effectively at a high temperature. It has a very small sticking
coefficient for dissociatively adsorbed nitrogen. Exposing the Ru crystal to be
a does of 10 L of N2 at temperatures between 400 and 760 K showed no
N2 peaks in the desorption spectrum [196, 197]. At lower temperatures (78 K),
Ru (001) can adsorb molecular species with a high sticking coefficient (ca. 0.4).
Two molecular forms are suggested from the unusual coverage dependence of
a sticking coefficient at 78 K [198, 199].

3.3.3.2 Tungsten, Molybdenum and Rhenium

Adsorbed nitrogen on metal films is desorbed by increasing the temperature at a
constant rate. This gives TPD spectra or thermal desorption spectra, and the
adsorbed states were analyzed on W, Fe, Mo or Re. Three peaks of adsorbed
134 K. Aika and K. Tamaru

nitrogen have been observed in TPD on W. The typical feature of the three types
of nitrogen on a W surface are summarized in Table 3.11. The first order kinetics
with respect to the nitrogen coverage, as well as the lower energy of desorption
observed with y and a types, suggest a molecular or dinitrogen nature of the
adsorbed site. This was confirmed by the absence of isotopic mixing in dini-
trogen desorbed from the surface on which a mixture of 28N 2 and 30N 2 was
preadsorbed [200,201]. On the other hand, the f3 type was regarded as an
atomic state on the basis of the second order desorption kinetics, the higher
energy of desorption, and the occurrence of isotopic mixing [42, 47, 201-204].
The TPD spectra were studied on single crystal surfaces ofW, and the results
are summarized on Table 3.11. As expected from the weak interaction, the y
state is found on any of three planes, W(100) [203], (110) [205,206], and (111)
[207]; and a state is only found on W(lll) as confirmed by FEM [203] or
probe-hole emission [207]; and the f3 state is predominant on (100) as demon-
strated by the large differences in sticking probability of nitrogen between
W(100) (0.3-0.6) [208-210] and W(1lO) (0.01-0.005) [203,210, 211J or W(111)
(0.08) [209,21OJ (see also Table 3.12). The desorption energies from single
crystalline planes have been measured and are shown in Table 3.11 [59]. Adams
and Germer demonstrated by LEED techniques that the f3 nitrogen atoms on
W(100) are located in an interstitial position surrounded by 4-square W atoms
as is shown in Fig. 3.9 [208,212].
The sticking probability s (rate of adsorption divided by the collision flux)
was studied as well on the clean W surface as a function of the coverage e. The
initial values of s (on a bare site) are as high as 0.3 to 0.4 on four-fold symmetry
sites such as (100) (see Table 3.12 and Fig. 3.9). This situation is contrary to the
case of Fe single crystals. Molecular beam experiments suggest that the dissocia-
tion preferably takes place at a vacant pair of (100) sites [73,210]. On W(100)
polycrystalline foil, s decreases gradually at lower coverage and then more
steeply with increasing e showing a convex curve. A constant value of s,
independent of coverage, indicates the presence of a precursor state of suffi-

Table 3.11. Adsorbed states of nitrogen on tungsten surfaces

y.state a-state fJ-state

Desorption temperature (K) ca. 150 ca. 300 ca. 1,000

Desoprtion energy (kJjmol) 8-40 80 300-320
Desorption kinetics 1st order 1st order 2nd order
Isotopic mixing of N 2 no no yes
Suggested form molec. molec. atomic
Plane selectivity (110) formed no a little
Desorp. energy (kJjmol) < 38 [203] 330 [211]
(111) formed formed a little
38 [203] 67 [203] 314 [203]
(100) formed no formed
38-44 314-334
[213, 214] [203,213,214]
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 135

Table 3.12. Sticking probability at zero converage

(so) at 300 K on various W planes [215J

Plane Reference

(110) < 10- 3 [211J

(211) < 10- 2 [203J
(321) < 10- 2 [212J
(111) ca. 10- 2 [208J
(310) 0.28 [208J
(411) 0.31 [212J
(210) 0.25 [208J
(311) 0.29 [212J
(100) 0.4 [203J

0.00.0
00.00
0.00.0
0000 Fig. 3.9. Surface Structure of W(l00) with
nitrogen atoms

ciently long life such that there is a high probability of transformation to the
chemisorbed state before desorption occurs. For the chemisorption of nitrogen
on W(100), the problem precursor is the y-state with a heat of adsorption of
43 kJ/mol (see Table 3.11). The value of s as a function of 8, s = aZ/(1
+ kd/kf(8)), indicates the following scheme:
N2 (g) ~Nia) kf(8) '2N(a)
kd
where a is a condensation coefficient, Z is the collision flux, kd is a rate of
desorption, k is a conversion rate constant. The experimental value of s is about
0.4 between 190 and 600 K, and the activation energy of k is 39 kJ/mol [213].
On the other hand, for N 2 on W(111), s decreases linearly with 8, indicating the
absence of a precursor state and direct molecular chemisorption (a-state) on
specific sites [73]. The adsorption and decomposition of ammonia on a clean
W(100) surface has been studied by photoemission spectroscopy (XPS and
UPS), and the results have been compared with those on c(2 x 2) -N ordered
W(lOO) [216].
Although the fresh surfaces of W have a strong affinity with N 2, the
ammonia activity is not high. The W surface is partly turned to nitride (WN)
136 K. Aika and K. Tamaru

under the ammonia synthesis condition, so that the rate is determined by a

dynamic balance of nitride formation and its hydrogenation (see Sect. 3.3.1.2).
Analogous types of adsorbed nitrogen have been found or suggested by
thermal desorption studies on Mo [188,214, 217-219J and Re [48J in addition
to W. Desorption energies were measured to be 41 kllmol for the )I-state on
Mo(100), 364 kllmol for the f3-state on Mo(100) [214J, and 339 kllmol for the 13-
state on Mo(110). Ammonia activity on Re single crystals is reviewed in Section
3.2.4.2. f3-type nitrogen is suggested as the intermediate in ammonia synthesis.
Adsorption of N 2 has also been studied on Mo single crystals. Mo( 111) was
changed to the (433) face by faceting at 850 K. The nitrogen atom is adsorbed on
this surface up to 0.35 of the coverage. The sticking coefficient at this state is
about 0.05, which is less than that on Mo(100). The coverage independence of
the sticking coefficient has been explained by an island formation of superstruc-
ture on the surface in this case [220]. The reaction rate of isotopic equilibration
of N 2 was measured over M02N(5 m Z Ig) at a temperature from 725 to 774 K.
The rate at 748 K and a N z pressure of 21 Torr is 0.85 x 1011 molec. Nz/cmz/s.
The ammonia synthesis rate was calculated to be 0.68 x 10 12 molec.Nz/cmz/s by
assuming the same intermediate state of adsorbed nitrogen. The good agreement
with the experimental value (0.75 x 10 1Z molec.N 2 /cm 2 /s) at 748 K and 760 Torr
of N2 + 3H 2 suggests the rate-determining step is N2 dissociative adsorption
and desorption [189].
At lower temperatures (80 K), Re(OOOl) and Re(1120) adsorb molecular
nitrogen with a high sticking coefficient ca. 0.9 [221]. The adsorbed dinitrogen is
suggested as being bound with its molecular axis perpendicular to the surface
[222]. At higher temperatures (218 K), the dissociation proceeds with a prob-
ability of 10- 2 to 10- 6 depending on the surface defect concentration [222J, or
with a probability of 4 x 10- 4 (1120), 9 x 10- 6 (0001), or 0.25 (polycrystalline foil)
[229]. The stabler form of the two molecular species is considered to be the
precursor to the dissociation [223].
The nitride-forming metals, such as vanadium and molybdenum in particu-
lar, seemingly give rise to a faster rate of N2 chemisorption and a slower
hydrogenation rate for the adsorbed nitrogen. In fact, the initial rate of N 2
chemisorption on pure molybdenum is very fast even at 583 K. However, the
rate decreases rapidly with an increase in N 2 uptake, or the extent of nitride
formation. Thus, the rate ofN 2 chemisorption on M0 2N is much slower than on
Mo. Even with iron, the rate of ammonia decomposition on iron nitride (Fe 4 N)
was reported to be two orders of magnitude slower than on pure iron [224].
Since both vanadium and molybdenum are transformed into bulk nitrides under
the reaction conditions of ammonia synthesis, the mechanism should be referred
to a nitride surface.
Kinetic studies suggest that the chemisorption of N 2 is the rate-determining
on nitrides. This was also confirmed by observing the absence of isotopic mixing
in N2 on M0 2N [191]. In this experiment, the 15N content of product ammonia
collected in a trap was always lower than that in N 2, indicating that the bulk
nitrogen in M0 2N undergoes an exchange reaction with adsorbed nitrogen or
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 137

ammonia. Moreover, the synthesis rate is as slow as expected [191]. The

ammonia synthesis rate can be calculated from the isotopic equilibration rate
assuming the same rate-determining step. Calculated data corresponds well with
the observed data on M0 2N [189].

3.3.3.3 Nickel, Platinum, Rhodium, Palladium, Iridium and Copper

The rate-determining step of ammonia synthesis and decomposition on Ni
powder was studied using tracer isotopes (determination of the stoichiometric
number) at 743 K. Dissociative adsorption and desorption were the slowest steps
[225]. Dissociative adsorption was studied in few cases. The slight extent of
chemisorption of N 2 at 523 K to 573 K was speculated to be atomic species on
Ni powder [226]. A spectroscopic study of surface "nitride" on poly crystalline Ni
was carried out under a N2 pressure of 200 Torr [227].
Thermal desorption peaks at 470 (low) and 800 K (high) were attributed to
dissociatively adsorbed nitrogen on Ni(110). This nitrogen was adsorbed at
591 K and gives Auger peaks at 380 eV. The sticking probability was very low
[228].
Very few works have been reported on the dissociative adsorption ofN 2 on
Pt, Rh, Pd and Ir. N 2 is inferred to be adsorbed dissociatively on poly crystalline
platinum ribbon with an apparent sticking coefficient of ca. 10- 2. Desorption of
dissociatively adsorbed nitrogen occurs around 650 K with second-order kin-
etics and an apparent activation energy of 80 kllmol [229]. Dissociative
adsorption of nitrogen has not been observed on Pt by other investigators. The
use of a high-frequency discharge or the use of a hot W filament supplies
activated N2 on Pt [230,231], on Pd [61], and on Cu [231]. The dissociative
adsorption ofN 2 on a Rh field emitter was suggested to occur with a significant
rate at a temperature of 500 K in addition to the molecularly adsorbed state
[232]. Atomically adsorbed nitrogen on Pd was formed by electron bombard-
ment [61-63] or by reduction of adsorbed NO [51,64] (see also Sect. 3.2.1).
Nitrogen is adsorbed in the molecular form (y-state) on Ir only at lower
temperatures (below 300 K). Dissociative {Horm was produced only by the use
of electron beams on Ir [233]. Dissociatively adsorbed nitrogen was observed on
Raney Rh, Ir and Pt in addition to Raney Fe, Co, Ni and Ru [76].

3.3.3.4 Alkali Metals, Alkaline-Earth Metal, Scandium and Lanthanides

Isotopic equilibration of N2 has been studied over nitrides of alkali metals,
alkaline-earth metals, and rare earth metals [32, 87, 234, 235]. The kinetic
results including U, Fe and Mo are summarized in Table 3.13. Many nitrides are
characterized by the mobility of the nitrogen atom in the nitride. More than
several monolayers of nitrogen atoms can take part in the isotopic exchange
with N 2 on group IlA or IlIA elements, and the entire bulk nitrogen in the
solid phase takes part in the exchange even at 573 K [235]. However, it is
138 K. Aika and K. Tamaru

interesting to point out that the mobility is not directly related to the equilibra-
tion activity. The activity seems, rather, to be related to the surface M-N bond
strength, that is, the heat of adsorption. Since the isotopic equilibration reaction
generally takes place through the dissociative adsorption and desorption, the
activation energy is equal to that of the adsorption step when the coverage is low
and is equal to the desorption step when the coverage is high. In the case of
nitrides, the coverage is high. Thus, the activation energies in Table 3.13
represent the activation energies of desorption, which may be comparable to the
heat of adsorption. Nitrides with low heats of activation (or low heat of
adsorption) are active for isotopic equilibration as is seen in Table 3.13.
It should be pointed out that nitrides of Sr and Ba are quite active. The
activity per unit surface may be the highest among those reported. However, the
estimated weight-base activity of SrN o.s is lower than Raney Ru and is higher
than 5% Ru-KjAC (see Fig. 3.3 and Table 3.13). The high catalytic activity ofBa
nitride is suggested as being due to the presence of various nitrogen-containing
complexes (Ba 2N, BaN 2, Ba 3N 4 ) on the Ba3N2 surface, which may give the best
route of bond expansion into the dissociation [32].
The rates of ammonia synthesis have been measured on several nitrides, and
the results are shown on Table 3.14 [87]. If there is no hydrogen effect, ammonia
synthesis rate can be estimated from the rate data of isotopic equilibration (see
Table 3.14 foot note). The experimental values are mostly lower than the
calculated values. In case of Ba, Sr, Ca, and Li, the nitride is changed to the
hydride and the activity decreases. Sr, the most active equilibration element,

Table 3.13. Isotopic equilibration of N2 on nitrides or metals'

Nitrides Exchangeable React. Act. log Vb Surf. V'

or N(a) order Energy (molecjcm2js) Area (Ilmol/hjg)
Metals (Monolayers) (kJjmol) (m2jg)

BaN o.62 50 0.7 50 12.4 O.1 f 550

SrN o.s 60 0.8 55 12.5 1.1" 12500
UN1.70 bulk 0.5 105 10.7
Ru d 0.7 125 8.6
Fee 0.5 130 10.1 5e 82
CaN o.s 0.7 0.7 130 9.0 2.2; 7.5
CeN 3 0.5 130 8.8 0.9 h 0.74
Mo 2N e bulk 0.5 155 7.0 se 0.075
Li3N bulk 0.4 175 7.5
oc-U 2N 3 bulk 0.5 226 5.1 0.6" 0.00021

a Ref. [87], LaN, PrN, and ErN have very low activities, and MgN o.6 is inactive
b Rate under 1.3 kPa of N2 at 573 K
'Estimated rate under 20kPa (150 Torr) ofN 2 at 588 K
d Ref. [236]
e Ref. [189]
f Ref. [32]
g Ref.

h Ref. [237]
; Ref.
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 139

Table 3.14. Ammonia synthesis on nitrides'

Nitrides Temp. NH3 yield NH3 synthesis rate Phase composition

(K) (mole%) log (NH3 molec./cm 2/g) after synthesis
expo calc.

Ba 573 0.32 13.8 14.5 Hydride

Sr 873 Inactive Nit.-Hydirde
U N J.70 623 0.24 11.6 12.9 Nitride
CA-lc 673 0.27 11.8
Ca 773 0.08 12.3 13.3 Nit.-Hydride
Ce 673 0.28 11.4 11.5 Nitride
Li 673 0.03 11.3 12.0 Hydride
Mg 873 Inactive Nitride
La, Pr, Nd 773 0.01-0.07 9-10 Nitride
Tb, Ho

• N2 + 3H 2 = 98 kPa [87]
b calculated as V([N 2]/[N 2]e - 1) where [N 2] = 25 kPa, [N 2]e is a ficticious N2 pressure under the
equilibrium with NH3 produced, and V is isotopic equilibration rate at [NZ]e
cPromoted Fe catalyst

turns out to be inactive, and Ba does not activate N 2 in the presence of H 2. The
activity of Ba comes only from the hydrogenation of the nitride which is only
formed by N2 without H 2. For UN1.7o, the rate of isotopic equilibration has a
reaction order of - 0.2 with respect to H2 [235]. Recalculation of the ammonia
activity using this data gives values closer to the experimental value of the
synthesis rate [87,235]. For CeN, hydrogen has no effect on the rate of isotopic
equilibration of N2 [237].
N2 is inferred to be adsorbed dissociatively on Dysprosium at 115 K,
yielding two XPS peaks in the N is region at 396.2 and 398.2 eV, corresponding
to a nitride and chemisorbed N(a), respectively. No peaks corresponding to
molecularly adsorbed N2 (400.2 eV) were observed. Upon heating the sample,
the N(a) is converted into nitride species. At a warm-up temperature of 300 K,
the N(a) species accounts for only ca. 10% of the total surface [238].
The rate of nitrogen sorption was studied at 10- 4 to 10- 6 Torr and 298 to
473 K on continuously renewed scandium films at a condensation rate of
4.8 x 10 13 atom cm - 2 S - 1. The rate of adsorption is limited by the rate of
physical adsorption [239].

3.3.4 Adsorbed State of Nitrogen

3.3.4.1 Infrared Adsorption Spectroscopy

The adsorbed state of nitrogen was studied by various instrumental and
spectroscopic techniques. These adsorbed species observed at room temperature
may be des orbed at higher temperatures, however, they may be a short-lived
140 K. Aika and K. Tamaru

intermediate of ammonia synthesis. As was shown before, N 2 must generally be

dissociated before the reaction with hydrogen. Thus N(a), the main adsorbed
species, is more important than N 2 (a) in the mechanism. However, it gives
important information, for example, how N2 is activated by the surface and
turned to the dissociated form N(a). Although Ni had been known to be inactive
in the chemisorption of N 2 , an IR-active adsorbed dinitrogen was found on
Ni/Si0 2 which disappeared on evacuation at 303 K. The adsorbed state was
confirmed to be molecular as evidenced by the isotope shift (2195, 2160 and
2123 cm- 1 for 28N2 29N2 and 30N2, respectively) and was assigned to
Ni-N == N+ [240]. It was later found that the IR-active species is formed only
when the size of the nickel particles on silica is in a range of 1.5 to 7.0 nm. The
effect of particle size was interpreted as that the concentration of "B5" sites
(inverted square pyramid) was at a maximum when the particle diameter ranged
from 1.5 to 7.0 nm [241]. Later, analogous IR-active species were found on
other metals as summarized in Table 3.15. These bands were also shown to be
diatomic by the isotope shift [240, 242, 243] as well as the first order kinetics of
desorption [244]. The high extinction coefficient associated with the adsorption
bands, comparable to the value of chemisorbed CO, has given evidence for an
end-on type of coordination rather than a side-on type [245].
The weak adsorption was formerly interpreted as an interaction with the
electric field on the metal surface [241]. However, later findings disclosed a
chemical nature of adsorption as suggested by: 1) a fairy high initial heat of
adsorption on Ni, 50 kllmol, 2) the fact that the shift in IR band frequency was
large and 3) the effect of hydrogen which causes a shift toward higher wave
numbers [240, 242, 243]. The deviation from gaseous N 2 (2331 cm -1) is a kind
of measure of the M - N bond strength in the case of a metal surface. On oxides,
on the other hand, the N-N strength is not changed with a very weak M-N
interaction, which gives an adsorption band at a wave number higher than for
the normal dinitrogen molecule. Such cases are seen on ZnO [246], zeolite
[247], Ti0 2 [248] and Zr0 2 [249].
Although iron is well known as a dinitrogen-activating metal, analogous IR-
active species observed on iron are formed by treatment with ammonia [250] or
a N2-H2 mixture [251] at elevated temperatures. The adsorption strength,
shown as the deviation of the wave number from that of gaseous N2 (Table 3.15),
is influenced by the co adsorbed hydrogen on Co/Al 2 0 3 [252], by the oxidized
site of Rh on Al 20 3 [253], or by the diluent element (Ag) in Pt on Si0 2 [254].
A very stable and IR-active dinitrogen species which can be converted to
ammonia by hydrogen treatment has been found on Ru-K/AI 2 0 3 at 2020 cm- 1
by N2 treatment above 473 K [255]. 15N2 gives rise to an isotope shift to
1997 cm -1. Since the IR absorption at 2020 cm -1 was also observed when
ammonia was decomposed on AI 20 3-K (without Ru) at 623 K, it was tentatively
assigned to a novel K-N2 surface complex [256]. Thus, it does not seem to be
the intermediate but rather a side product on the surface. The spectra which
might be due to the "intermediate" complex was observed recently on active Ru
catalysts. The isotope shift confirms these species are end-on adsorbed species
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 141

Table 3.15. IR data of adsorbed dinitrogen on metals

Sample Support Wave No Deviation Heat of Ref.

(cm-I) from adsorption
gaseous (kllmol)
N 2 (cm- l )

Fe Al 2 0 3 2200 131 [250]

MgO 2200 131 [250]
MgO 2050 281 [250]
Co Al 2 0 3 2214 117 [252]
Ni Si0 2 2202 129 50 [240]
Si0 2 2202 129 [241]
Si0 2 2195 136 [242]
Si0 2 2202-2220 129-111 [257]
(evaporated) 2202 129 [258]
Ru K-AI 2 0 3 2020 311 [255]
1935 396 [256]
Si0 2 2240 111 [259]
Al 2 0 3 2214 117 [260]
MgO 2168 163 [260]
CsOH/MgO 1910 421 [260]
Rh Si0 2 2236 95 [243]
2255 76 [261]
Al 2 0 3 2257 74 9 [253]
2252 79 [261]
Ti0 2 2254 77 49 [261]
Pd Al 2 0 3 2260 71 [241]
Ir 2185 146 17 [262]
Pt Si0 2 2230 101 [241,254]
Si0 2 2230 101 15 [244]
(as Pt black) 18 [264]
ZnQ 2337 -6 19 [246]
Na-A-Zeolite 2339 -8 [247]
I]-AI 2 0 3 2360 - 29 46 [263]
Ti0 2 2337 -6 11 [248]
2348,2340 - 17, -9 25 [265]
Zr0 2 2341 - 10 [249]
Si0 2 2329 2 [265]
BeO 2350,2337 - 19, - 6 [265]
Si0 2 -AI 2 0 3 2357 - 26 31 [265]
H-mordenite 2356,2335 - 25, - 4 [265]

(Ru-N == N). The wave numbers (cm -1) are 1910 for Ru-CsOHjMgO, 2268 for
RujMgO, 2214 for RujAI 20 3 , and 2240 for RujSi0 2 (Table 3.15). The deviation
from the gas phase N2 molecule is 421,163,117 and 111 cm - \ respectively. This
number is a measure of N2 activation, N-N bond looseness, and the extent of
electron donation by the Ru atom which is interacting with the support or the
promoter. Interestingly, the ammonia synthesis activity is ranged in this order
(Fig. 3.4). Although the life time of these species might be short at the synthesis
reaction conditions, the species might be a precursor of the dissociated state. A
mechanism of ammonia synthesis and the role of the support and promoter is
shown in Fig. 3.10.
142 K. Aika and K. Tamaru

acidic support

basic support

Fig. 3.10. A model of the support and pro-

moter effect on Ru catalyst in ammonia syn-
thesis mechanism

3.3.4.2 Electron Spectroscopy

Electron spectroscopy of adsorbed species has been developed in recent years.
Both AES (Auger electron spectroscopy) and XPS (X-ray photoemission spec-
troscopy) can give information about the electronic state of atoms in the
outermost layer of a solid surface, thus permitting the study of adsorbed species.
When dinitrogen is adsorbed at 83 K on Mo or at 373 K on Fe, at which
temperatures molecular adsorption is expected, a single peak is observed at
around 385 eV. While, when it is adsorbed at 273 K on Mo or 673 K on Fe, three
peaks are observed. Since M0 2N gives three similar peaks, the adsorbed species
obtained at higher temperatures are assigned to atomic nitrogen [218].
Analogous Auger spectra were observed on W(llO) for N2 adsorbed at
100 K [206]. The amount of adsorbed nitrogen can be monitored by the peak
height of the Auger spectra as was done on W by the 390 eV peak [202] and on
Fe by the 380 eV peak [266].
Dinitrogen adsorbed on W, Ni or Fe at 100 K or lower gives two peaks in
XPS at around 400 and 406 eV, which were interpreted in different ways:
1) N2(a) for 406.6 eV and N(a) for 397.6 eV on W because the latter remained
while the former disappeared at 300 K [267]; 2) Two types ofN2(a) [268], end-
on and bridge [269]; and 3) two nitrogen atoms of an end-on type dinitrogen,
W == N == N [206].
There has been agreement that the atomically adsorbed nitrogen gives rise to
a peak of around 400 eV, e.g. 397.2 and 398.0 eV for N(a) on W at 300 K [267];
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 143

397.2 eV for N(a) on Fe at 290 K [269] and 397.5 eV for N(a) on promoted iron
at 673 K [270]. Surface reconstruction by nitrogen adsorption has been ob-
served on various surfaces [219]. Surface science reviews have been published
[3,4,271].
The valuable comments of Professor Emeritus A. Ozaki are acknowledged
by one of the authors.

3.4 References

1. Timm B (1984) The Ammonia Synthesis and Heterogeneous Catalysts, A Historical Review in
"Proc. 8th Intern. Congr. Catal.," DECHEMA ed., Verlag Chemie. Weinheim, Vol 1 p 7
2. Mittasch A (1951) Geschichte der Ammoniak Synthese, Verlag Chemie, Weinheim
3. Grunze M (1982) In: King DA, Woodruff DP (eds) Chern Phys Solid Surf Heterog Catal,
Elsevier, Amsterdam, Vol 4 p 143
4. ErtI G (1983) In: Anderson JR, Boudart M (eds) Catalysis, Science and Technology, Springer-
Verlag, Berlin, Vol 4 p 209
5. Shilov AE (1989) In: Bottomley F (ed) A Treatise on Dinitrogen Fixation, John Wiley & Sons,
New York, Sec. 1, p 31
6. Bottomley F (1989) In: Bottomley F (ed) A Treatise on Dinitrogen Fixation, John Wiley & Sons,
New York, Sec. 1, p 109
7. Khan F, Yue P-L, Rizzuti L, Augugliaro V, Schiavello M (1981) J Chern Soc Chern Commun
1049
8. Endoh E, Leland JK, Bard AJ (1986) J Phys Chern 90: 6223
9. Khader MM, Lichtin NN, Vurens GH, Salmeron M, Somorjai GA (1987) Langmuir 3: 303
10. Uyama H, Uchikura T, Niijima H, Matsumoto 0 (1987) Chern Lett 555
11: Kalman J, Varga TA, Hajos R (1983) Proc 6th Intern Symp Plasma Chern Vol 3: 686
12. Sugiyama K, Akazawa K, Oshima M, Miura H, Matsuda T, Nomura 0 (1986) Plasma Chern
Plasma Process 6: 179
13. Miyahara K (1983) Chern Lett 1971
14. Nielsen A (1970) Review of Ammonia Catalysis, In: Heinemann H (eds) Catalysis Rev, Marcel
Dekker, New York., Vol 4 p 1
15. Nielsen A (1977) Fert Sci Tehcnol Ser 2: 87
16. Ammoniak NH3-Bildung und Zerfall (1935) In: Gme1in Handbook of Inorganic Chemistry,
Vol 4 Nitrogen p 320
17. Mittasch A (1949) Early Studies of Mulicomponent Catalysts, In: Frannkenburg WG et al. Adv
Catalysis, Academic Press, New York, Vol 2 p 81
18. Ammonia In: (1968) Encyclopedia of Chemical Technology (R.E. Kirk, D.F. Othmer), Hark
HF et al. (eds) John Wiley, New York, p 258
19. Ammoniak, Synthetisches, In: Ullmann, Vol 3 p 544
20. Vancini CA, (1971) Synthesis of Ammonia, Macmillan, London
21. Nielsen A (1968) An Investigation on Promoted Iron Catalysts for the Synthesis of Ammonia,
Jul Giellerups Forlag, Copenhagen
22. Nielsen A (1981) Catal Rev -Sci Eng 23: 17
23. Boudart M (1981) Catal Rev -Sci Eng 23: 1
24. Jennings JR (1991) Catalytic Ammonia Synthesis, Fundamentals and Practice Plenum Press,
New York
25. Ozaki A, Aika K (1979) The Synthesis of Ammonia by Heterogeneous Catalysis, In: Hardy
RWF et. al. (eds) A treatise on Dinitrogen Fixation Sec. I and II, John Wiley, New York, p 169
26. Ozaki A, Aika K (1981) Catalytic Activation of Dinitrogen, In: Anderson JR, Boudart M (eds)
Catalysis-Science and Technology, Springer-Verlag, Berlin, Vol 1 p 87
27. Emmett PH, Love KS (1933) J Am Chern Soc 55: 4043
28. Aika K, Ozaki A (1968) Bull Chern Soc Jpn 41: 2818
144 K. Aika and K. Tamaru

29. Jolly WL (1964) The Inorganic Chemistry of Nitrogen, Benjamin, New York, p 36
30. Trapnell BMW (1953) Proc Roy Soc A218: 566
31. Bond GC (1962) Catalysis by Metals, Academic Press, London New York
32. Panov GI, Boreskov GK, Kharitonov AS, Moroz EM (1981) React Kinet Catal Lett 16: 247
33. Panov GI, Boreskov GK, Kharitonov AS (1982) Kinet Ketal 23: 438
34. Schlier E, Farnsworth HE (1950) Phys Rev 78: 316
35. Toyoshima I, Takezawa N, Suzuki H (1973) J Chern Soc Chern Commun 270
36. Toyoshima I, Takezawa N, Suzuki H (1977) Proc 6th Intern Congr Catal, Bond GC et al. (eds)
Chern Soc London, p 708
37. Scholten JJF, Zwietering P (1957) Trans Farad Soc 53: 1363
38. Beeck 0 (1950) Advan Catal 2: 151
39. Ko SM, Schmidt LD (1974) Surf Sci 42: 508
40. Beeck 0, Colle WA, Wheeler A (1950) Discuss Faraday Soc 8: 314
41. Oguri T (1964) J Phys Soc Jpn 19: 77
42. Oguri T (1963) J Phys Soc Jpn 18: 1280
43. Parry AA, Pryde JA (1967) Br J Appl Phys 18: 329
44. Kisliuk P (1959) J Chern Phys 31: 1605
45. Hickmott TW, Ehrlich G (1958) J Phys Chern Solids 5: 47
46. Hill MP, Lecchini SMA, Pethica BA (1966) Trans Farad Soc 62: 229
47. Madey TE, Yates JT Jr (1966) J Chern Phys 44: 1675
48. Yates JT Jr, Madey TE (1969) J Chern Phys 51: 334
49. Bagg J, Tomkins FC (1955) Trans Faraday Soc 51, 1071
50. Urabe K, Aika K, Ozaki A (1975) J Catal 38: 430
51. Obuchi A, Naito S, Onishi T, Tamaru K (1982) Surf Sci 122: 235
52. Mimeault VJ, Hansen RS (1966) J Phys Chern 70: 3001
53. Vajo JJ, Tsai W, Weinberg WH (1985) J Phys Chern 89: 3243
54. Miyazaki E, Yasumori I (1976) Surf Sci 55: 747
55. Sachtler WMH, Van Reijen LL (1962) J Res Inst Catal Hokkaido Univ 10: 87
56. Brenann D, Hayward DO, Trapnell BMW (1960) Proc Roy Soc A256: 81
57. Roberts MW (1960) Nature 188: 1020
58. Tanaka K, Tamaru K (1963) J Catal 2: 366
59. Toyoshima I, Somorjai GA (1979) Catal Rev -Sci Eng 19: 105
60. Frankenburg WG (1955) The Catalytic Synthesis of Ammonia from Nitrogen and Hydrogen,
.. In: Emmett PH (ed) Catalysis, Reinhold Pub, New York, 1955, Vol 3 p 171
61. Kunimori K, Kawai T, Kondow T, Onishi T, Tamaru K (1976) Surf Sic 59: 302
62. Davies PW, Lambert RM (1981) Surf Sci 110: 227
63. Kuwahara Y, Fujisawa M, Jo M, Onchi M, Nishijima M (1987) Surf Sci 180: 421, ibid 188: 490
64. Matsuo I, Nakamura J, Hirano H, Yamada T, Tanaka K, Tamaru K (1989) J Phys Chern
93: 7747
65. Mittasch A (1950) Adv Catal 2: 81
66. Stathism E (1937) Osterr Chemiker Ztg 40: 80
67. Mittasch A, Keunecke E (1931) Z Phys Chern 574
68. King DA, Sebba F (1965) J Catal 4: 253
69. Aika K, Yamaguchi J, Ozaki A (1973) Chern Lett 161
70. Spitsyn VI, Mikhailenko IE, Pokrovskaya OV (1982) Dokl Akad Nauk SSSR, 263: 656
71. Jost F (1908) Z Anorg Chern 57: 414
72. Logan SR, Kemball C (1960) Trans Farad Soc 56: 144
73. Aika K, Hori H, Ozaki A (1972) J Catal 27: 424
74. Ozaki A, Aika K, Hori H (1971) Bull Chern Soc Jpn 44: 3216
75. Urabe K, Oh-ya A, Ozaki A (1978) J Catal 54: 436
76. Ogata Y, Aika K, Onishi T (1989) Bull Chern Soc Jpn 62: 642
77. Keunecke E (1930) Z Elektochem 36: 690
78. Brill R, Osumi Y (1966) Bull Chern Soc Jpn 39: 1678
79. Zabuga V Ya, Markova GP (1965) Kataliz i Katalizatory, Akad Nauk Ukr SSR, Resp Mezhved
sb 110 (CA64:4312d)
80. Chalenko VG, Tovbin MV (1964) Ukr Khim Zh 30: 1128
81. Artyukh Yu. N, Rusov MT, Boldyreva NA (1967) Kinet Katal 8: 1319
82. Smith PI, Taylor DW, Dowden DA, Kemball C, Taylor D (1982) Appl Catal 3: 303
83. Komarov VS, Efros MD, Rabina PD, Dmitrenko LM, Rozin AT, Kuznetsov LD, Lemeshonok
GS, Khachaturyan IG, Mantseva GM, Eremenko SI (1983) SU 988327 (CA98:1l4516b)
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 145

84. Urabe K, Yoshioka T, Ozaki A (1978) J Catal 54: 52

85. Hikita T, Aika K, Onishi T (1990) Catal Lett 4: 157
86. Hikita T, Kadowaki Y and Aika K (1991) J Phys Chern 95: 9396
87. Panov GI, Kharitonov AS (1984) Proc 9th Intern Congr Catal DECHEMA (eds) Verlag
Chemie, Veinheim, Vol 3 p 371
88. Takeshita T, Wallace WE, Craig RS (1976) J Catal 44: 236
89. Wallace WE, France J, Shamsi A (1982) Rare Earth Mod Sci Technol 3: 561
90. Kirch G, Schwab E, Wicke E, Zuchner H (1984) Proc 8th Intern Congr Catal DECHEMA (ed)
Verlag Chern, Weinheim, Vol 4 p 209
91. Baiker A, Schlogl R, Armbruster E, Guntherodt HJ (1987) J Catal 107: 221
92. Shiflett WK, Dumesic JA (1982) J Catal 77: 57
93. Spencer ND. Schoonmaker RC, Somorjai GA (1982) J Catal 74: 129
94. Aika K, Ohya A, Ozaki A, Inoue Y, Yasumori I (1985) J Catal 92: 305
95. Aika K (1986) Angew Chern 25: 558
96. Ohya A, Aika K, Ozaki A (1984) J Chern Soc Chern Commun 321
97. Urabe K, Aika K, Ozaki A (1974) J Catal 32: 108
98. Urabe K, Aika K, Ozaki A (1976) J Catal 42: 197
99. Ichikawa M, Kondo T, Kawase K, Sudo M, Onishi T, Tamaru K (1972) J Chern Soc Chern
Commun 176
100. Croft RC (1956) Austr J Chern 9: 184
101. Vol pin ME, Novikov Yu N, Postnikov VA, Shur VB, Bayerl B, Kaden L, Wahren M,
Dmitrenko LM, Stukan RA, Nefedev AV (1977) Z Anorg AlIg Chern 428: 231
102. Sudo M, Ichikawa M, Soma M, Onishi T, Tamaru K (1969) J Phys Chern 73: 1174
103. Nishiyama S, Matuura S, Morita H, Tsuruya S, Masai M (1985) Appl Catal 15: 185
104. Nishiyama S, Yoshioka K, Yoshida T, Tsuryua S, Masai M (1988) Surf Sci 33: 1081
105. Asscher M, Somorjai GA (1984) Surf Sci 143: L389
106. Asscher M, Carraza J, Khan MM, Lewis KB, Somorjai GA (1986) J Catal 98: 277
107. Stoltze P, Structure and Surface Chemistry ofIndustrial Ammonia Synthesis Catalysts, chapter
2 in this book
108. Amariglio H, Rambeau G (1977) Proc 6th Intern Congr Cat aI, Bond GC et al (eds) Chern Soc
London, p 1113
109. Strongin DR, Bare SR, Somorjai GA (1987) J Catal 103: 289
110. Ozaki A, Urabe K, Shimazaki K, Sumiya S (1979) Preparation of Catalysts II, Elsevier,
Amsterdam, p 381
111. Lyubchenko YuA, Sergeeva AN, Dmitrenko LM, Tkachenko Zhl, Pavlenko LI (1981) SU 484
718 (CA 96:037844M)
112. Johnson MM, Tabler DC, Nowack GP (1982) US 4309311
113. Sydney A, Smith P, Yarm C (1986) EP 201205
114. Aika K, Shimazaki K, Hattori Y, Ohya A, Ohshima S, Shirota K, Ozaki A (1985) J Catal 92: 296
115. Lear AM, McCarroll n, Pippard DA, Tennison SR (1982) GB 2087746
116. McCarroll n, Tennison SR (1983) GB 2109361
117. Aika K, Takano T, Murata S (1992) J Catal 136: 126
118. McCarroll JJ, Tennison SR, Wilkinson NP (1986) US 4600571
119. Aika K, Kawahara T, Murata S, Onishi T (1990) Bull Chern Soc Japan 63: 1221
120. Murata S, Aika K, Onishi T (1990) Chern Lett 1067
121. Murata S, Aika K (1992) J Catal 136: 118
122. Kadowaki Y, Murata S, Aika K (1993) New Frontiers in Catalysis, Proc, 10th Intern Congr
Catal Budapest, Guczi L et al eds, Elsevier p 2055
123. Aika K, Ozaki A (1969) J Catal 13: 232
124. Ozaki A, Taylor HS, Boudart M (1960) Proc Roy Soc (London) 258: 47
125. Boudart M, Oyama ST, Leclercq L (1981) Proc 7th Intern Cogr. Catal 1980, Seiyama T,
Tanabe K (eds) Elsevier, Tokyo, p 578
126. Boudart M, Oyama ST (1985) US 4515763
127. McCandlish LE, Turaew LW, Wrisht FJ, Kugler EL (1982) EP 53018
128. Oyama ST, Boudart M (1980) J Res Inst Catal Hokkaido Univ 28: 305
129. Schulz-Ekloff G, Barsesel D, Sarholz W (1976) J Catal 43: 353
130. Kuznetsov BN, Kuznetsov VL, Ermakov Yu I (1975) Kinet Katal 16: 915
131. Temkin MI, Pyzhev VM (1940) Acta Physicochim (USSR) 12: 327
132. Kiperman S, Temkin MI (1946) Acta Physicochim (USSR) 21: 267
133. Kiperman S, Temkin MI (1946) Zh Fiz Khim 20: 623
146 K. Aika and K. Tamaru

134. Sidrov IP, Livshits VD (1947) Zh Fiz Khim 21: 1177

135. Peters C, Krabetz R (1956) Z Elektrochem 60: 859
136. Dumesic JA, Topsoe H, Khammouma S, Boudart M (1975) J Catal 37: 503
137. Kiperman S (1947) Zh Fiz Khim 21: 1435
138. Kiperman S, Granovskaya V (1951) Zh Fiz Khim 25: 557
139. McGill WJ, Sebba F (1963) J Catal 2: 104
140. Kunsman CH (1928) J Am Chem Soc 50: 2100
141. Tamaru K (1961) Trans Faraday Soc 57: 1410
142. Jungers JC, Taylor HS (1935) J Am Chem Soc 57: 679
143. Barrer RM (1936) Trans Faraday Soc 32: 490
144. Hinshelwood CN, Burk RE (1925) J Chem Soc (London) 127: 1105
145. Hailes HR (1931) Trans Faraday Soc 27: 601
146. Shindo H, Egawa C, Onishi T, Tamaru K (1979) Z Naturforsch 34a: 96
147. McGeer JP, Taylor HS (1951) J Am Chem Soc 73: 2743
148. Logan SR, Moss RL, Kamball C (1958) Trans Faraday Soc 54: 922
149. Schwab GM, Krabetz R (1956) Z Electrochem 60: 855
150. Sidrov IP, Livshits VD (1952) Zh Fiz Khim 26: 538
151. Love KS, Emmett PH (1941) J Am Chem Soc 63: 3297
152. Takezawa N, Toyoshima I (1966) J Phys Chem 70: 594
153. Tamaru K, Tanaka K, Fukasaku S, Ishida S (1965) Trans Faraday Soc 61: 765
154. Friedlander AG, Courty PhR, Montarnal RE (1977) J Catal48: 322
155. Amano A, Taylor H (1954) J Am Chem Soc 76: 4201
156. Apel'baum LO, Temkin MI (1959) Zh Fiz Khim 33: 2697
157. Robertson AJB. Willhoft EMA (1967) Trans Faraday Soc 63: 476
158. Dixon JK (1931) J Am Chem Soc 53: 2071
159. Schwab GM, Schmidt FH (1929) Z Elektrochem 35: 605
160. Elod E, Banholzer W (1926) Z Elektrochem 32: 555
161. Dixon JK (1931) J Am Chem Soc 53: 1763
162. Rusov MI, Pevsner Ts V (1954) Zh Fiz Khim 28: 1765
163. Brill R, Tauster S (1962) J Chem Phys 36: 2100
164. Emmett PH, Kummer JT (1943) Ind Eng Chem 35: 677
165. Bokhoven C, van Raayen W (1954) J Phys Chem 58: 471
166. Aika K, Kumasaka M, Oma T, Kato 0, Matsuda H, Watanabe N, Yamazaki K, Ozaki A,
. Onishi T (1986) Appl Catal 28: 57
167. Holzman PR, Shiflett WK, Dumesic JA (1980) J Catal 62: 167
168. Morikawa Y, Ozaki A (1971) J Catal 23: 97
169. Baris H, Glinski M, Kijenski J, Wokaun A, Baiker A (1986) Appl Catal28: 295
170. Shiflett WK, Dumesic JA (1981) Ind Eng Chem Fundam 20: 246
171. Temkin MI, Morozov NM, Shapatina EN (1963) Kinet Ketal 4: 260
172. Tamaru K (1965) Proc 3rd Intern Congr Catal, Sachtler WHM et al (ed) Amsterdam, North
Holland Pub p 664
173. Tamaru K (1988) Acc Chem Res 21: 88
174. Shindo H, Egawa C, Onishi T, Tamaru K (1980) J Chem Soc Faraday Trans 1 76: 280
175. Egawa C, Nishida T, Naito S, Tamaru K (1984) J Chem Soc Faraday Trans 1 80: 1595
176. Loeffler DG, Schmidt LD (1976) J Catal44: 244
177. Loeffler DG, Schmidt LD (1976) J Catal41: 440
178. Lotz CR, Sebba F (1957) Trans Faraday Soc 53: 1246
179. Segal N, Sebba F (1967) J Catal 8: 105
180. Kummer JT, Emmett PH (1951) J Chem Phys 19: 289
181. Morikawa Y, Ozaki A (1968) J Catal 12: 145
182. Schulz G, Schaefer H (1969) Z Phys Chem (N.F.), 64: 333
183. Schulz-Eklolf G (1971) Ber Bunsenges Phys Chem 75: 110
184. Guyer WRF, Joris GG, Taylor HS (1941) J Chem Phys 9: 287
185. Gorbunov AI, Boreskov GK (1960) In: Problemy Kinet Katal Akad Nauk (USSR), No. 10:
pp 192
186. Joris GG, Taylor HS (1939) J Chem Phys 7: 893
187. Gasser RPH, Lowrence CP, Newman DG (1965) Trans Faraday Soc 61: 1771
188. Gasser RPH, Hale A, Marsay CJ (1967) Trans Faraday Soc 62: 1789
 3 Ammonia Synthesis over Non-Iron Catalysts and Related Phenomena 147

189. Boreskov GK, Kolchanova VM, Rachkovskii EE, Filimonova SN, Khasin AV (1975) Kinet
Katal 16: 1218
190. Moore GE, Unterwald FC (1968) J Chern Phys 48: 5393
191. Aika K, Ozaki A (1969) J Catal 14: 311
192. Rambeau G, Amariglio H (1981) J Catal 72: 1
193. Rambeau G, Amariglio H (1981) Appl Catal 1: 291
194. Rambeau G, Jorti A, Amariglio H (1982) Appl Catal 3: 273
195. Rambeau G, Jorti A, Amariglio H (1982) J Catal 74: 110
196. Danielson LR, Dresser MJ, Donaldson EE, Dickinson JT (1978) Surf Sci 71: 599
197. Danielson LR, Dresser MJ, Donaldson EE, Sandstrom DR (1978) Surf Sci 71: 615
198. Feulner P, Menzel D (1982) Phys Rev B25: 4295
199. Menzel D, Pfnur H, Feulner P (1983) Surf Sci 126: 374
200. Rigby LJ (1965) Can J Phys 43: 532
201. Yates JT Jr, Madey TE (1965) J Chern Phys 43: 1055
202. Joyner RW, Rickman J, Roberts MW (1974) J Chern Soc Faraday Trans 1 70: 1825
203. Delchar TA, Ehrlich G (1965) J Chern Phys 42: 2686
204. Goymour CG, King DA (1973) J Chern Soc Faraday Trans 1 69:749
205. Yates JT Jr, Klein R, Madey TE (1976) Surf Sci 58: 469
206. Fuggle JC, Menzel D (1978) Yak Tech 27: 130
207. Wilf M, Folman M (1975) Surf Sci 52: 10
208. Adams DL, Germer LH (1971) Surf Sci 26: 109
209. King DA, Wells MG (1974) Proc Roy Soc (London) Ser A 339: 245
210. Singh-Boparai SP, Bowker M, King DA (1975) Surf Sci 53: 55
211. Tamm PW, Schmidt LD (1971) Surf Sci 26: 286
212. Adams DL, Germer LH (1971) Surf Sci 27: 21
213. Clavenna LR, Schmidt LD (1970) Surf Sci 22: 365
214. Han HR, Schmidt LD (1971) J Phys Chern 75: 227
215. Tompkins FC, (1978) Chemisorption of Gases on Metals, Academic Press, London, p 26.
216. Egawa C, Naito S, Tamaru K (19!i3) Surf Sci 131: 49
217. Pasternak RA, Endow N, Bergsnov-Hansen B (1966) J Phys Chern 70: 1304
218. Mahnig M, Schmidt LD (1972) Z Phys Chern (N. F.) 80: 71
219. Kunimori K, Kawai T, Kondow T, Onishi T, Tamaru K (1976) Surf Sci 54: 525
220. Egawa C, Naito S, Tamaru K (1983) Surf Sci 125: 605
221. Haase G, Asscher M (1987) Surf Sci 191: 75
222. Grunze M, Golze M, Fuhler J, Neumann M, Schwarz E, Proc 8th Intern Congr Catal 1984,
DECHEMA (ed) Verlag Chemie, Weinheim Vol 4 p 133
223. Haase G, Asscher M (1987) Chern Phys Lett 142: 241
224. Khrizman lA, Korneyiuchuk G (1943) Acta Physicochim (URSS) 18: 420
225. Kazusaka A (1971) J Res Inst Catal Hokkaido Univ 19: 42
226. Kokes RJ, Emmett PH (1958) J Am Chern Soc 80: 2082
227. Honda F, Hirokawa K (1977) J Electron Spectrosc Relat Phenom 10: 125
228. Grunze M, Driscoll RK, Burland GN, Cornish JCL, Pritchard J (1979) Surf Sci 89: 381
229. Wilf M, Dawson PT (1976) Surf Sci 60: 561
230. Schwaha K, Bechtold E (1977) Surf Sci 66: 383
231. Kiss J, Berko A, Solymosi F (1981) Magy Kern Foly 87: 566
232. Hendrickx HACM, Hoek A, Nieuwenhuys BE (1983) Surf Sci 135: 81
233. Gorodetskii VV, Sobyanin VA (1980) Proc 7th Intern Congr on Cat ai, Seiyama T, Tanabe K,
(eds) Elsevier, Amsterdam, 1981, p. 566.
234. Kharitonov AS, Boreskov GK, Panov GI, Pankrotiev Yu D (1983) React Kinet Catal Lett 22:
309
235. Panov GI, Boreskov GK, Kharitonov AS, Moroz EM, Sobolev VI (1984) Kinet Katal 25: 123
236. Magomedkov EP, Kasatkina LA (1978) Trudy MKhTl im Meneleeva DI No 99: 60
237. Panov GI, Boreskov GK, Kharitonov AS (1980) Dokl Akad Nawk, SSSR 252: 646
238. Schreifels JA, Deffeyes JE, Neff LD, White JM (1982) J Electron Spectrosc Relat Phenom 25:
191
239. Varnakova RG, Morozova LV, Khamidova Kh Kh (1982) Zh Fiz Khim 56: 1533
240. Eischens RP, Jacknow J (1965) Proc 3rd Intern Congr Cat ai, Sachtler WHM et al (ed)
Amsterdam, North Holland Pub pp 627
148 K. Aika and K. Tarnaru

241. Hardeveld RV, Montfoort AV (1966) Surf Sci 4: 396

242. Hardeveld R V, Montfoort AV (1969) Surf Sci 17: 90
243. Borodko YuG, Lyutov VS (1971) Kinet Katal 12: 238
244. Egerton TA, Sheppard N (1974) 1 Chern Soc Faraday Trans 1 70: 1357
245. Eischens RP (1972) Acc Chern Res 5: 74
246. Chang CC, Kokes Rl (1973) 1 Phys Chern 77: 2640
247. Cohen de Lara E, Delaval Y (1978) 1 Chern Soc Faraday Trans 2 74: 790
248. Sakata Y, Kinoshita N, Dornen K, Onishi T (1987) 1 Chern Soc Farad Trans 1 83: 2765
249. Sakata Y, Abe H, Kondo 1, Maruya K, Dornen K, Onishi T (1989) Chern Lett 711
250. Okawa T, Onishi T, Tarnaru K (1977) Z Phys Chern (N. F.) 107: 239
251. Brill R, liru P, Schulz G (1969) Z Phys Chern (N. F.) 64: 215
252. Kinoshita K, Kido K, Dornen K, Aika K, Onishi T (1986) 1 Chern Soc Farad Trans 1 82: 2269
253. Wang HP, Yates JT lr (1984) 1 Phys Chern 88: 852
254. Delong KP, Bongenaar-Schlenter BE, Meirna GR, Verkerk RC, Lammers MJJ, Geus lW
(1983) 1 Catal 81: 67
255. Oh-kita M, Aika K, Urabe K, Ozaki A (1976) 1 Catal 44, 460; (1975) 1 Chern Soc Chern
Cornrnun 147
256. Aika K, Midorikawa H, Ozaki A (1982) 1 Phys Chern 86: 3263
257. Nakata T, Matsushita S (1965) 1 Catal 4: 631
258. Bradshaw AM, Pritchard 1 (1970) Surf Sci 19: 198
259. Lyutov VS, Borodko YuG (1971) Kinet Katal 12: 1566
260. Kubota 1, Aika K (1991) 1 Chern Soc Chern Cornrnun 1544; (1994) J Phys Chern 99:
261. Wey JP, Worley CG, Neely WC, Worley SD (1992) 1 Phys Chern 96: 7088
262. Ravi A, King DA, Sheppard N (1968) Trans Faraday Soc 64: 3359
263. Kubsov SA, Borovkov VYu, Kazansky VB, Gagarlin SG (1984) Chern Phys Lett 107: 337
264. Rissman EF, Parry 1M (1975) 1 Phys Chern 79: 1975
265. Zverev SM, Srnirnov KS, Tsyganenko AA (1988) Kinet Katal 29: 1439
266. Bozso F, Ertl G, Weiss M (1977) J Catal 49: 18
267. Madey T, Yates JT Jr , Erickson NE (1974) Surf Sci 43: 526
268. Brundle CR (1976) J Vac Sci Tech 13: 301
269. Kishi K, Roberts MW (1977) Surf Sci 62: 252
270. Ertl G, Thiele N (1979) Appl Surf Sci 3: 99
271. Rao CNR, Ranga Rao G (1991) Surf Sci Report 13: 221
272. Walker AP, Rayment T, Lambert RM (1989) 1 Catal 117: 102
Chapter 4
Kinetics of Ammonia Synthesis and Decomposition
on Heterogeneous Catalysts

John B0gild Hansen

Haldor Topsoe AjS Copenhagen, Denmark

Contents

4.1 Early Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

4.2 The Temkin-Pyzhev Rate Equation . . . . . . . . . . . . . . . . . . . 152

4.3 The Rate Equations of Brunauer, Love, and Keenan . . . . . . . . . 157

4.4 Rate Equations Derived from the Langmuir Isotherm . . . . . . . . 165

4.5 Kinetics Based on Surface Science Techniques . . . . . . . . . . . . . 177

4.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182

4.7 Transport Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

4.7.1 Mathematical Model . . . . . . . . . . . . . . . . . . . . . . . . 184

4.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

Nielsen (Ed.), Ammonia

© Springer.Verlag Berlin Heidelberg 1995
150 J. B. Hansen

4.1 Early Developments

Despite the vast practical importance of ammonia synthesis on heterogeneous

catalysts, more than 25 years passed from when these catalysts were introduced
until the first workable rate equation was proposed in the literature.
The ground work had been laid by the publication of the rates of the
ammonia synthesis as a function of operating conditions. Larson and Tour [1]
published in 1922 data of a study on doubly promoted (i.e. Al 20 3 and K 20)
catalysts between 400 and 475°C and 1-100 atm. This was followed in 1926 by
the Almquist and Crittenden paper [2].
Benton [3] attempted to develop a rate equation from these data. He noted
the retarding effect of ammonia on the synthesis rate and explained this by the
accumulation of ammonia on the surface of the catalyst.
Winter [4] determined the rate of NH3 decomposition on an iron foil at
atmospheric pressure in the temperature range 500 to 700°C. The experiments
were carried out in the presence of excess H2 in order to avoid the formation of
iron nitride. Many earlier studies had used pure NH 3, thereby transforming the
catalyst into a nitride phase and making it drastically different from the catalyst
under synthesis conditions.
Winter fitted the data to the following rate equation:

(1)

Here P NH3 and P H2 are the partial pressures of NH3 and H 2, t is the contact
time.
To rationalize his findings, Winter assumed that the steps leading to adsor-
bed nitrogen in the atomic state on the catalyst surface were so fast that the
equilibrium
NH3(gas) <=± N(ads) + 1.5 H 2(gas) (2)

is established. This equilibrium can be obtained by combining the two equilibria

NH3(gas) <=± N(gas) + 1.5 H2(gas) (3)

and
N(gas) <=± N (ads) (4)
From equilibrium (3) one gets
p1.5 p
K=~ (5)
PNH3

Here PN, the pressure of nitrogen atoms in the gas, is a virtual pressure and
serves as an auxiliary quantity only.
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 151

Assuming low coverage and Henry's law to be valid one obtains for the
coverage eN of nitrogen far from equilibrium
eN = aPN (6)
Inserting P N from Eq. (5) one arrives at
P NH3
eN = a K p1.5 (7)
H2

Winter further assumed that nitrogen is desorbed as atoms, which recom-

bine in the gas phase according to the scheme
N(ads) <=:± N(gas) (8)
2 N(gas) <=:± N z (gas) (9)
These assumptions lead to the rate of decomposition being proportional to eN,
i.e.
(10)

which is very close to rate equation (1).

The idea that adsorption of N z is the rate determining step in the synthesis
reaction was beginning to take hold, as discussed by Taylor [5], Frankenburger
[6], and Emmett et al. [7]. Indirect evidence supporting this theory began to
accumulate.
Emmett and Love [8] found that hydrogenation of iron nitride takes place
at temperatures below that of synthesis, which suggests that the hydrogenation
of adsorbed nitrogen is fast.
Emmett and Brunauer [9] determined the rate of N z adsorption. It was
found to be comparable to the rate of NH3 formation or sufficiently fast to be
the first step in the synthesis mechanism.
Taylor and Jungers [10] demonstrated that isotopic exchange between NH3
and D2 takes place even at room temperature. NH 2D, NHD z and ND3 were
detected on the doubly promoted catalyst used in the study. They concluded
that
N 2(gas) --+ 2 N(ads)
must be the rate determining step in the ammonia synthesis.
Kozhenova and Kagan [11] demonstrated that the hydrogenation of adsor-
bed nitrogen proceeds much faster than N z adsorption.
Data on the rate of synthesis and decomposition continued to be published.
Finkelstein studied NH3 decomposition on promoted and unpromoted iron and
the synthesis on iron [12]. Chrisman [13] did experiments with doubly pro-
moted catalyst at atmospheric pressure. Some empirical rate equations were
proposed. Finkelstein and Rubanik [14] proposed that the reaction is second
order and the rate determining step is
2Kx + N2 --+ 2KxN
152 1. B. Hansen

with the following hydrogenation steps being fast. On this basis they determined
the activation energy from earlier data to be between 18.6 and 22 kcal/mol.
Roiter [15] proposed empirical rate equations for the synthesis rate, being
different for subatmospheric pressures and pressures between 50-200 atm. He
did not discuss the reaction mechanism.

4.2 The Temkin-Pyzhev Rate Equation

Temkin and Pyzhev used the data of Winter as the starting point for their
development of the first successful rate equation for ammonia synthesis named
after them [16-18]. They demonstrated, that using Winter's assumptions leads
to an apparent activation energy for decomposition Ed larger than
97000 cal/mol, clearly disagreeing with the experimentally found energy of
51 000 cal/mol. Furthermore, they pointed out that the desorption of nitrogen in
the atomic form is extremely unlikely due to the large amount of energy needed.
If, however, the predominant view that the rate of ammonia decomposition
is determined by the rate of
2 N(ads) +=± N 2 (gas) (11)
is accepted, then the rate should be proportional to e~ again in conflict with the
experimental findings.
In order to overcome this difficulty Temkin and Pyzhev proposed to aban-
don the Langmuir adsorption concept, which basically assumes the surface of
the catalyst to be energetically uniform. Instead they chose to use for the
adsorption equilibrium the isotherm proposed by Frumkin and Slygin [19],
1
e = f lnaoP (12)

where e is the coverage, P the equilibrium pressure, and ao and f are constants,
and for the rate of adsorption,
(13)
an equation developed by Zeldowitsch and Roginsky [20]. Here P is the gas
pressure and Ka and g are constants. For the rate of desorption they used the
equation proposed by Langmuir [21]
(14)
where kd and h are constants.
Temkin and Pyzhev pointed out that choosing the three equations above is
equivalent to assuming a linear dependence of the heat of adsorption with the
coverage. In accordance with Polanyi [22] the change in energy of activation
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 153

should be a certain fraction rx of the change in the heat of adsorption. These

assumptions involve the concept of a heterogenous surface. Temkin and Pyzhev
ascribed this heterogeneity to considerable repulsive forces between the adsor-
bed molecules. They noted that these assumptions describe the intermediate
degrees of coverage and are valid for adsorption of both molecules and atoms.
The constants f, g and h are related as follows
g
rx=- (15)
f

{J=~ (16)
f
rx+{J=l (17)
Another assumption for the Temkin-Pyzhev equation is that the nitrogen
adsorption is not influenced by hydrogen and ammonia, an assumption which
seems to be corroborated by the experimental data of Emmett and Brunauer
[7].
Following Winter, Temkin and Pyzhev assumed that the amount of adsor-
bed nitrogen is determined by the equilibrium with hydrogen and ammonia in
the gas phase, i.e., the nitrogen equilibrium pressure is given by

(P) ' P~H3

N2 eq = k - 3 - (18)
P H2
Inserting into Eq. (12) one finds

e -- -11nao k' P~H3-3- (19)

f P H2
Using Eq. (14) the rate of desorption becomes

w = kd(aok')P (~~cy = k2 (;£3)P (20)

In order to get Eq. (20) to agree with Winter's experimental findings, Temkin
and Pyzhev had to choose
rx = {J = 0.5

Using data from Emmett and Brunauer [7], Temkin and Pyzhev were able
to demonstrate that the activation energy for decomposition on the average
should be 46500 cal/mol, in good agreement with Winter's results.
To arrive at an expression for the synthesis rate equation, Eq. (19) is inserted
in Eq. (13) to get the net rate as the difference between the forward and reverse
processes,

dP NH3
dt
= k P
1 N2
(p~2)a
p2
_k 2
(P~H3)P
p3
(21)
NH3 H2
154 J. B. Hansen

Assuming IX = f3 = 0.5, Eq. (21) reduces to

(22)

The retarding effect of ammonia on the synthesis rate observed by Benton

[3] is also borne out by Eq. (21), but the underlying reason is not the accumula-
tion of NH3 on the surface, but rather an increase in the amount of nitrogen on
the surface due to the equilibrium

(23)

Equation (22) can be rewritten by considering that at equilibrium

dPNH3 = 0
dt
and therefore

(24)

or, using the usual equilibrium constant Kp, one has

kl = K2 (25)
k2 P

If (P NH3 )eq denotes the partial pressure of ammonia in equilibrium with

hydrogen and nitrogen, given by

(26)

then Eq. (22) reduces to

(27)

At low pressures, or low P NH3 , P NH3 is a good approximation for the

(P NH3 )eq
efficiency 1], defined as the ratio of the actual ammonia yield to the yield under
equilibrium conditions at a given temperature and pressure. With this approxi-
mation, Eq. (27) can now be integrated to give

(28)
Here V is the space velocity, the ratio of the volume of gas at standard
conditions to the volume of catalyst per hour, and}' characterizes the deviation
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 155

from stoichiometric gas composition

P H2
(29)
Y = 0.75P
and k is given by

k = 273¢kl (30)
0.751.ST
where ¢ is the catalyst bed void volume ratio.
Temkin and Pyzhev tested the rate equation by applying Eq. (28) to a series
of experiments at atmospheric pressure with a doubly promoted catalyst, first by
varying the H1iNl ratio at 400°C using a space velocity of 30000h- 1 . As
predicted by Eq. (22), the H1iNz ratio giving the highest yield at low conversion
was 1.5.
A second set of experiments was carried out at 400 and 450°C. They showed
a satisfactory constancy of k at each temperature as expected if Eq. (28) were
correct. The apparent activation energy for the decomposition rate constant kl
was calculated to be 40000 cal/mol.
The decomposition data of Chrisman [13] were evaluated with Eq. (20), and
an activation energy of 39700 cal/mol was found.
Temkin and Pyzhev derived the following integrated form of their rate
equation without making any simplifying assumptions, except assuming H 2iNl
to be 3.

(31)

where z is the mole fraction of NH3 in the gas mixture, given by

(32)

and V0 the space velocity at the entry to the reactor. L is defined by

(33)

Temkin and Pyzhev made an error in deriving Eq. (31), as pointed out first by
Emmett and Kummer [23 J. Emmett and Kummer introduced a factor of
(1 + Z)3 instead.
The correct first denominator term, (1 + Z)l was first worked out by
Kodama et al. [24]. Temkin also arrived at (1 + Z)2 in [25]. The error introduc-
ed by a wrong denominator is normally small as Z is small.
Temkin and Pyzhev used Eq. (31) to test their rate equation with the data of
Larson and Tour [2] at 10, 31.6 and 200 atm and at 420 and 450°C. They
concluded that the data agreed well with Eq. (21), however, the rate constants
156 J. B. Hansen

calculated for the 200 atm data were lower than the ones calculated for the lower
pressures. Temkin and Pyzhev attributed this difference to the fact that the
measurements at high pressures were carried out with different equipment than
those at low pressures. Temkin and Pyzhev later pointed out that at pressures
above 300 atm fugacities should be used instead of pressures and the constants
k2 and kl should be assumed to depend on the pressure.
By differentiating Eq. (21) with respect to temperature, the optimum syn-
thesis conditions (maximum rate) is found at a temperature at which the
equilibrium constant is equal to

P NH3
p O. 5 p1.5 =
~d
If (34)
Nz H2 S

where Ed and Es are the apparent activation energies for decomposition and
synthesis, respectively.
Love and Emmett [26] studied the decomposition of ammonian over both
singly (10.2% A1 20 3) and doubly promoted (10.2% Al 20 3 1.59% K 20) catalysts
at atmospheric pressure.
The doubly promoted catalyst showed the rate equation
dP NH3 _ k P~it (35)
-~- p O. 85
H2

which is very close to Temkin and Pyzhev's Eq. (20) if f3 is set at 0.29. The
apparent activation energy for decomposition was found to be
45600 ± 2000 cal/mol.
The data obtained on singly promoted catalysts, however, could not to be
fitted to the Temkin-Pyzhev equation. Their performance depended in a highly
complex manner on the temperature and the partial pressures and the activation
energy depended on the temperature and even displayed hysteresis effects. These
deviations from Temkin-Pyzhev kinetics could be removed by impregnating the
catalyst with a KOH solution although the activity dropped considerably.
The decomposition rate found was approximately independent of the am-
monia and hydrogen partial pressures at 390 and 450°C. At 429 DC, it was
proportional to P~·22 /P~J: i.e., showing a negative f3 value in the Temkin-
Pyzhevequation.
Love and Emmett suggested that this abnormal behaviour could be due to
the presence of substantial amounts of.NH and NH2 groups on the surface on
singly promoted catalysts, whereas on doubly promoted catalysts, such groups
would be "driven off" by the alkali.
This interpretation was supported in some chemisorption experiments with
H2 at 100°C on singly and doubly promoted catalysts previously exposed to N2
at 450°C (Brunauer and Emmett [27]).
Frankenburger later explained the phenomena by NH, NH2 and NH3 being
adsorbed on the acidic Al 20 3 groups in singly promoted catalysts. These acidic
functions could be neutralized by K 20 [28].
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 157

4.3 The Rate Equations of Brunauer, Love, and Keenan

Brunauer et al. [29J extended the range of the Eqs. (12), (13), (14), which are the
basis for the Temkin-Pyzhev equation. In order to develop an adsorption
isotherm, Brunauer et al. subdivided the surface into elements each following
(the original) Langmuir isotherm, so that
ad1 - 8)p = b 1 8e- q / RT (36)
where q is the heat of adsorption, and ai and b i are constants. Assuming q to be
linear function of the surface area covered,
q = qo - As (37)
where qo is the heat of adsorption at 8 = 0, one finds
a e-As/RTp ds
8-
-
J_o
1

0
_ _~o=-
1 + aoe-BS/RTp
(38)

Integration gives

8 = RT In 1 + aoP (39)
A 1 + aoe- a/RT
where
a i qo RT
ao = _e / (40)
bi
ifao»l»aoe-A/RT, then Eq. (39) gives the Frumkin and Slygin isotherm (Eq.
(12)) used by Temkin and Pyzhev, if
1 RT
(41)
f A
The Temkin-Pyzhev equation is thus only valid at medium surface coverage as
already pointed out by the authors.
Brunauer et al. also developed equations for the rates of desorption and
adsorption using as starting points the equations proposed by Taylor [5J,
w = b8e - Ed/RT (42)
z = aP(1 - 8)e- E ./ RT (43)

where wand z are rates of desorption and adsorption, respectively, and band
a are constants.
They assumed the activation energies to vary linearly with coverage:
Ed = E~ - Bs (44)
Ea=E~-Js (45)
and, furthermore, divided the surface into two parts, an almost completely
158 J. B. Hansen

covered part and an almost completely bare part, which is equivalent to

assuming Band J to be sufficiently large. Integrating only over the relevant part
of the surface gives
w = be-E~/RT(RT/B) (e B8 /RT - 1) (46)
(47)

e
Equation (46) gives w = 0 for = 0 and the maximum value of w for = 1. e
e e
Equation (47) gives z = 0 for = 1 and a maximum value of z for = O. For
e,
large values of Eq. (46) reduces to the Langmuir desorption rate equation (14)
e,
and for small values of Eq. (47) gives the Zeldotwitsch equation with
h = B/RT (48)

g = J/RT (49)
Brunauer et al. also deduced rates for adsorption and desorption if the surface is
homogeneous, but strong attractive and repulsive forces exist between the
adsorbed molecules or atoms. They assumed the activation energy to be linear
with coverage
w = be-E~/RT e B8 /RT (50)
z = aPe-E~/RT(l - e)e- J8 /RT (51)

and found the adsorption isotherm formula by equating Eqs. (50) and (51)

P = [:0 (1 - e)}AO/RT (52)

Here A, B, J are again related by the equations

B +J =A (53)
(54)

At intermediate surface coverage Brunauer et al. used Eqs. (46) and (47)
(which then reduces to the Temkin Eqs. (13) and (14)) to calculate the adsorption
rate of nitrogen

dV = k PV e-JV/VmRT _ k V eBV/VmRT (55)

dt a m d m

where Vm is adsorbate volume in a monolayer. Equation (55) contains four

constants kaVm' kdVm,J/Vm, and B/Vm which can be derived from the experi-
mental nitrogen adsorption rate curve.
Using Emmnett and Brunauer's data [7] on doubly promoted catalyst,
Brunauer et al. calculated the four constants and obtained a close fit to the
experimental rate curve. At equilibrium
kd B+J
InP = I n - + - - (56)
ka VmRT
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 159

By using the constants, determined in the adsorption rate experiment,

Brunauer et al. achieved an excellent fit to the adsorption isotherm determined
experimentally with the same catalyst.
Substituting the f3 term in the Temkin-Pyzhev equations
B 800
f3 = B + J = 2900 = 0.276 (57)

into Eq. (20) gives, using the f3 determined from N 2 adsorption,

dP NH3 _ k P NH3
O.55

-~- 2 p O.S 3 (58)

Hz

in good agreement with what Love and Emmett had with the same catalyst
referred to before. An activation energy of 49000 cal/mol could also be esti-
mated in good agreement with the experimental value.
In 1943, Emmett and Kummer [23] presented the results of high pressure
experiments on a doubly promoted catalyst (3.02% Al 2 0 3 , 0.94% K 2 0) at 33.3,
66.6 and 100 atm, H2/N2 ratios: 3/1,1/1 and 1/3 and space velocities from 25 000
to 125000 h -1 at 370, 400 and 450°C were used. The data were analysed with
the Temkin-Pyzhev equation using (J. = f3 = 0.5. The rate constant k was con-
stant with variations in space velocity, except at 370°C where it decreased with
an increase in space velocity of 5. Apparent activation energy for decomposition
was found to be from 45000 to 53000 cal/mol.
Although the Temkin-Pyzhev equation seemed to fit the data reasonably
well with respect to variations in space velocity and temperature, all data sets
exhibited a clear decrease in the rate constant with increasing pressure.
Using a f3 value of approx. 0.3 as found by Love and Emmett [26] an
attempt was made to calculate the data by the Temkin-Pyzhev equation with
(J. = 0.67, but this gave a poorer agreement. With respect to changes in gas

composition the agreement with (J. = 0.67 was fair at 370 and 400°C, but the rate
constant varied by a factor 2 at 450°C between the 3: 1 and 1: 3 H2/N2 mixtures.
Emmett [30] presented a review on the ammonia decomposition reaction in
1946.
In 1947, Temkin and Kiperman [31] gave a general discussion of the
Temkin-Pyzhev equation and its usefulness for both the synthesis reaction and
the decomposition. They pointed out that it would not be valid at low ammonia
partial pressures because the equilibrium
(59)
would be displaced to the right leading to low surface coverages, where the
Temkin-Pyzhev equation is no longer valid. The rate r should be independent of
coverage and given by
(60)
Sidorov and Livshits [32] carried out experiments at 300 atm in the temperature
160 1. B. Hansen

range 450-500°C. They obtained a good fit of the data by using the Tem-
kin-Pyzhev equation with CI. = 0.5 and an activation energy of 176 kJ/mol. They
still used (1 + Z)3 in the denominator in the integration of the Temkin-Pyzhev
equation introduced by Emmett and Kummer [23].
Some data on the synthesis rate were given by Kobayashi and Kubota in
[33].
The less-than-satisfactory agreement with experimental data at higher pres-
sures ofthe Temkin-Pyzhev rate equation has been mentioned. In 1950, Temkin
modified the original equation by introducing fugacities instead of partial
pressures [34]. Furthermore, the original equation, Eq. (21), was multiplied by
a term

dP NH3 = [k P (P~2)~ _k (P~H3)1-~J e (Cl.Vs - Va)P (61)

dt 1 N2 p2 2 p 3 RT
NH3 H2
to take high pressures into account. Here Va is the partial molar volume of the
activated complex of the nitrogen adsorbed. Vs the partial molar volume of
adsorbed N 2, and P the total pressure. Temkin suggested using the molar
volume of solid nitrogen of 27 cm 3 / mol both Vs and Va'
In a series of experiments with triply promoted catalyst at 1 atm, 400°C and
SV = 30000 at different H2/N2 ratios Nielsen [35J found that maximum con-
version was obtained at a H2/N2 ratio of 1.5 as predicted by the Temkin-Pyzhev
equation (Eq. (22)) when the backward reaction can be ignored. At 330 atm,
450°C, and SV = 15000 the decomposition rate constant (k in Eq. (31)) was
however found to increase with approximately a factor of two when the H2/N2
was increased from 1/1 to 6/1.
Further discussions by Temkin et al. of the use of the Temkin-Pyzhev
equation can be found in [36].
Brill [37J carried out synthesis experiments on pure iron, singly promoted
(3% A1 20 3 ) and doubly promoted catalyst (3% A1 20 3 , 2% K 20) at atmo-
spheric pressure. The catalyst was reduced within 20-80 h at 450°C with a space
velocity of 4000 h - 1. The experiments were carried out at 314-320 DC, because
abnormal reaction rates were observed at 450°C, attributed to diffusion effects.
When the reaction rate data were fitted to the Temkin-Pyzhev equation taking
CI. = 0.5 a dependency on space velocity became apparent, k being smaller at low
space velocities than at high ones. This trend could be removed by using
different CI. values. The best fit was obtained with CI. values 0.7 for pure Fe, 0.7 or
somewhat larger for Fe-AI and 0.65 for doubly promoted catalyst.
Annable was the first to use the Temkin-Pyzhev equation to analyze operat-
ing data from industrial ammonia units [38J: an adiabatic reactor operating at
245 atm and a multi bed quench reactor operating at 300 atm.
For a given inlet gas composition, pressure, and temperature, numerical
integration was carried out throughout the reactor. The rate constant k2 in the
Temkin-Pyzhev equation was adjusted so that the calculated temperature
profiles matched the measured ones. Radial gradients and axial heat conduction
were ignored. The catalyst particles were assumed to have the same temperature
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 161

as the gas. The variance between the actual and calculated ammonia outlet
concentrations was between 0 and -7%. The accuracy of k2 was estimated
to be 10% for the adiabatic reactor and 16% for the quench reactor. The tem-
perature covered the range 374-555°C and the efficiencies the range 0.09 to
0.96. The results in the form of k2 as a function of temperature were plotted as
Arrhenius plot. Although the experimental scatter was considerable
(around 30%), the Temkin-Pyzhev equation simulated the plant performance
well.
Activation energies for decomposition reaction of 45800 cal/mol and
47900 cal/mol were found at 245 atm and 300 atm, respectively, in good agree-
ment with Emmett and Kummer's results [23]. A trend in k2 with pressure could
be observed. Whereas Emmett and Kummer found that k2 was proportional to
p-O.6\ Annabel found k2 to be proportional to p-o.s.
The Temkin-Pyzhev equation was also used to analyze the performance of
a catalyst charge with a feed gas containing larger amount of poisons than the
experiments above. The activity was only one third of the activity of a catalyst
operating with a relatively pure gas. A bend in the Arrhenius plot could be
observed around 450°C. The Temkin-Pyzhev equation with the as-found con-
stants was eventually used to find the optimum temperature curve in an ideal as
well as a practical reactor.
Kiperman and Granovskaya [39] studied the kinetics at low ammonia
partial pressures and efficiencies below 1%. They found the rate to be propor-
tional to the nitrogen partial pressure in agreement with the Temkin equation
(Eq. (60)).
Sidorov and Livshits used the special high-pressure version of the Temkin
equation (Eq. (61)) to study the kinetics on doubly promoted catalyst at 500°C
[40]. They found the rate constant to be independent of pressure between 10 and
400 atm with 0( = 0.5. The total pressure range studied was 10-500 atm.
Adams and Comings made an extensive study on the ammonia synthesis
rate on an industrial ammonia catalyst (2.84% Ah03, 1.04% K 20, 0.26% Si0 2)
[41]. They used an isothermal reactor (within 1 to 3°C) holding 1 ml of catalyst,
in a tubular catalyst chamber measuring 0.95 cm (inside diameter) by 1.25 cm
(length).
A total of 300 runs were made at 350,400,450,475 and 500°C and 100, 200
and 300 atm. The H2/N2 ratios studied were 3/1, 1/1 and 1/3 and the space
velocities ranged from 10000 to 125000h- 1 .
They used the original Temkin-Pyzhev equation with 0( = 0.5 to correlate
the data, but agreement between theory and experiment was even less satisfac-
tory than that obtained by Emmett and Kummer [23].
Partial pressures were replaced with fugacities calculated according to Lewis
and Randall rules, and the data sets were recalculated. The rate constant k' still
showed considerable variations with operating conditions. Below 400 °C it
decreased with space velocity and above 400 °C it increased. It decreased with
both pressure and H2/N2 ratio. In view of this, Adam and Comings resorted to
a graphical approach to reactor design.
162 J. B. Hansen

Bokhoven et al. later pointed out that the experimental data might have
been influenced by back diffusion in the reactor due to the shallow bed used
[42].
Frankenburger in a comprehensive review article [28] on the ammonia
synthesis extensively discussed the physical meaning of ex and f3 in the Tem-
kin-Pyzhevequation.
Furthermore, he discussed the unresolved problems in interpreting the
kinetics of ammonia synthesis. In principle, it should be possible by theoretical
means to calculate, for instance, the rate constant k t . Emmett and Brunauer [7]
had already pointed out that only approx. lout of 106 N2 molecules striking the
catalyst surface and possessing the necessary activation energy are actually
chemisorbed on the surface. Frankenburger pointed out that such calculations
ignore the entropy changes associated with the chemisorption step. The entropy
effects are linked to the probability that a molecule will actually be chemisorbed.
The entropy of the chemisorbed state is mostly lower than that of the gaseous
state.
Frankenburger suggested a two-step mechanism for N2 adsorption involv-
ing a short lived "physical" adsorption on the surface followed by nitrogen
dissociation. Another reason for the heterogeneous behaviour of the ammonia
catalyst surface, in addition to the two already mentioned, was suggested by
Frankenburger: "previously adsorbed particles influence the entire catalyst and
particularly its surface in such a way that it exerts forces of attraction or
repulsion toward N 2 molecules from the gas phase different from those exerted
by a completely bare surface."
Bokhoven et al. in a review article [42] critically discussed the application of
lhe Temkin-Pyzhev equation to fit the data published before 1955. They quote
their own unpublished results of synthesis experiments at 1 atm on a doubly
promoted catalyst at 350°C with H2/N2 = 3.0. The best fit is obtained with
ex = 0.6.
Concerning high pressure experiments, Bokhoven does not discuss the data
of Larson and Tour [1], because these were influenced by diffusion effects. The
data of Adams and Comings [41] were not used either because of the possible
influence of back diffusion in these experiments.
Bokhoven et al. used data from Emmett and Kummer [23], Sidorov and
Livshits [32], [42] and Nielsen [35]. They pointed out that the data of Emmett
and Kummer are restricted to a narrow efficiency range. Regarding the data of
Nielsen, they only compare those at an intermediate space velocity, because
non-isothermal conditions could be more or less pronounced with higher or
lower space velocity.
The calculated rate constants from the data of Nielsen and Sidorov and
Livshits are influenced by the H2/N2 ratio and are decreasing with pressure; in
the latter case there is only a small decrease, however, k2 is independent of
efficiency.
Bokhoven et al. cited some of their own, unpublished results obtained with
a differential reactor at 350°C whereby the exponents of the partial pressures
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 163

were directly determined. The rate equation was found to be

dP NH3 = k pl. 1 p~~o (62)
dt N2 pl.Z
NH3

which is close to the Temkin-Pyzhev equation if Q( = 2/3. The pressures were

P N2 = 2.7 - 10 atm, P H2 = 9.4 - 22 atm and P NH3 = 0.2 - 1.1 atm. A higher
value of Q( removed some of the inconsistencies in the literature data surveyed,
but on the other hand it made the pressure dependencies of k z worse.
Bookhoven attributed the different pressure dependencies of k z in the
various studies to differences in the catalysts used. The results on doubly
promoted catalyst are best fitted by the Temkin-Pyzhev equation. It was noted
that KzO addition to singly promoted catalysts lowers its activity at 1 atm, but
increases it at 200 atm.
Bokhoven also discusses the decomposition kinetics, especially the results of
Love and Emmett [26]) on singly promoted catalysts. He repeated their experi-
ments with a gas containing 6%NH3 and 94% Hz and found that the singly
promoted catalysts obeyed the Temkin-Pyzhev equation with an activation
energy of 60 kcaljmol up to 411°C, whereafter the activation energy became
much lower. Hysteresis effects were also observed, although less pronounced
than in Love and Emmett's work. When the catalyst was impregnated with
KOH, the abnormal behaviour disappeared and the activation energy in the
temperature range 400-450°C was constant (59 kcaljmol).
Bokhoven interpreted the results obtained on singly promoted catalysts of
Love and Emmett to be due to nitrogen dissolution in iron. The necessary high
nitrogen fugacities are due to the relatively high ammonia concentrations used.
This could also explain the observed hysteresis effect as being due to slow
adjustment of the adsorption equilibrium. The activity of the catalyst is expected
to depend on the amount of N z dissolved in the iron, which could explain the
increase in decomposition rate with increasing hydrogen pressure.
Peters and Krabetz [43] reported data on both singly and doubly promoted
catalysts for the synthesis reaction at 1 atm. At 324°C they found Q( = 0.5 and
for 372-378°C they found Q( = 0.5-0.6 for singly promoted catalysts.
Scwab and Krabetz [44] studied the decomposition of NH3 on pure iron in
the temperature range 380-500 °C and the pressure range 0.013-0.106 atm. The
decomposition rate was described by

dP P O.9-1.0
-~=k NH3 (63)
dt p 1 .4-1.5
H2

corresponding to a f3 value in the Temkin-Pyzhev equation of 0.45-0.50.

Kawamura et al. [45] measured the rate of synthesis for doubly promoted
and triply promoted (Al z0 3, KzO, CaO) catalysts. The operating parameters
were: temperature = 400-550 DC, pressure 100 or 200 atm, and space velocities
between 5000 and 400000 h - 1. They tried without success to fit the data to the
published rate equations.
164 J. B. Hansen

Shishkova et al. [46] confirmed the validity of the Temkin-Pyzhev equation

at pressures up to 300 atm using a commercial catalyst in a recycle reactor.
Logan et al. [47] studied the decomposition of NH3 at low pressures
(P = 0.01-0.044 atm) from 335 to 470°C and found the rate law
dPNH3 _ k P~J3 (64)
-~- po.?
H2

i.e., f3 ~ 0.25 in the Temkin-Pyzhev equation. The activation energy was found
to be 38.8 kcal/mol.
Mills and Bennett made an extensive kinetic study at 100-1000 atm and 400
and 450°C on doubly promoted catalysts [48]. H2/N2 ratios of 3/1, 1/1 and 2/1
were used and the space velocities ranged from 2000 to 230000 h - 1. The reactor
used held 1.625 g or 0.602 cm 3 of catalyst and had the dimensions 4.76 cm in
diameter and 5 cm long. The maximum temperature difference recorded was
3°C and in most cases it was less than 1 °C. Approximate calculations demon-
strated that the experimental data should be free of axial or radial dispersion of
heat or mass. The size of the catalyst used was 2 mm diameter, and intraparticle
diffusion effects were thought to be absent. The lowest effectiveness factor
estimated was 0.88. Reduction of the catalyst was done according to the
procedures of Nielsen [35].
The experimental data were presented in the form of curves showing the
space time yield (cm 3 NH3 produced per cm 3 per hour) as a function of space
velocity. All data were evaluated with the Temkin-Pyzhev equation modified for
high pressure (Eq. (61)) according to Temkin [34]. The ideal solution method
was selected for calculating the fugacities.
The Temkin-Pyzhev equation in form of Eq. (61) was found to describe the
variations with space velocity not too close to equilibrium fairly well, but k2 was
clearly decreasing with pressure. To fit the data, the partial molar volumes of
nitrogen would have to be in the order of 200 or 500 cm 3/mol instead of the
more probable 20-50 cm 3 /mol. The activation energies for decomposition were
calculated at the different pressure levels.

Pressure Energy of Activation

atm cal/mol

100 36000
400 44000
1000 53000
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 165

4.4 Rate Equations Derived from the Langmuir Isotherm

If one assumes the catalyst surface to be homogeneous energetically, then the

classical Langmuir isotherm can be applied. This is equivalent to assuming q to
be constant in Eq. (36) in the treatment by Brunauer et al. [28].
Assuming the rate-determining step to be dissociative adsorption of nitro-
gen, the rate of synthesis becomes
r = kP N2 (1 - 8)2 (65)
If the most abundant surface species is assumed to be nitrogen, the coverage
8 can be determined from the equilibrium
N(ads) + 1.5 H2 ~ NH3 (66)
with the adsorption equilibrium constant K. According to Langmuir, 8 is
determined by

K(~~~;)
8= H2 (67)

1+ K(~~7)
and the rate then becomes

(68)

If instead of adsorbed nitrogen, NH is assumed to be the most abundant

surface species, while N2 adsorption is still rate-determining, one arrives at
k'P N2
r = {
1
(P )}2
+ K~ ;H: 3
(69)

because the coverage is determined by the equilibrium

(NH)ads + H2 ~ NH3
and K~ is the equilibrium adsorption constant for this equilibrium.
If nitrogen is assumed adsorbed in the molecular form and this is the most
abundant surface species, then one obtains

r = 1 + K (P~H3) (70)
p1.5
H2

Kubota and Shindo [49] carried out the analysis above and reevaluated the
data given in Refs. [23, 32, 33, 41,50,51]. They used fugacities instead of partial
166 1. B. Hansen

pressures and included terms for the reverse reaction. The data were well fitted.
The results obtained on doubly promoted catalysts were best fitted by Eq. (68),
whereas Eq. (69) worked best for singly promoted catalysts. The following
activation energies were found:

95 kJ/mol Emmett and Kummer [27]

77-95 kJ/mol Kobayashi and Kubota [31]
76 kJ/mol Uchida and Kuraishi [50]

Kubota and Shindo pointed out that the data could also be correlated by
Eq. (70); thus, kinetics alone cannot be proof of a mechanism. Boudart [52] as
early as 1956, had noted that it should be possible to describe the ammonia
synthesis kinetics by the Langmuir isotherm. This was also discussed by Stelling
and von Krusenstierna [53].
Both Temkin [54] and Kwan [55] showed that assuming a simple Freun-
dlich isotherm can also lead to the Temkin-Pyzhev equation.
Ozaki et al. [56] carried out experiments at and below atmospheric pressure
with two types of doubly and triply promoted catalysts (A = 0.85% Al z0 3 ,
0.27% KzO and B = 2.6% Al z0 3 , 1.6 KzO and 1.6% SiOz). Amounts of ap-
prox. 0.8 g were used. Reduction was carried out in pure H2 for 100 hours at
400°C. The experiments were carried out at 218, 251, 278, and 302°C and the
pressure was kept at 1/3, 1/2 and 1 atm. Stoichiometric gas was obtained by
cracking NH 3 and ND 3'
The experimental data were first fitted to the Temkin-Pyzhev equation. It
was noted that the value of IX decreased from 0.8 to 0.4 with increasing efficiency.
This can be attributed to low coverage by nitrogen, thus making the assump-
tions for the Temkin-Pyzhev equations invalid as noted by Brunauer [27] and
also Temkin [17], [31]. At a fixed temperature and at relative high efficiencies,
k still depended on pressure. Ozaki et al. correlated the data with Eq. (68) and
found k independent of pressure but K dependent on pressure. A successful fit,
however, was obtained using Eq. (69) with both k' and K~ independent of
pressure.
The isotope effect KH/Ko also was different from the one calculated from
partition functions if Eq. (68) was used. Good agreement between the theoretical
and experimental isotope effect was obtained with Eq. (69). The rate constants
k are identical for H2 and D2 strongly indicating that nitrogen adsorption is the
rate determining step.
Ozaki et al. also extended the Temkin-Pyzhev equation by using the iso-
therm developed by Brunauer et al. [27].

(71)
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 167

. P
and noting that if 1 »e-o.sfK o P~~;
H2

then

e = ~ In {I + Ko (~t:; )} (72)

Using the Zeldowitsch equation (Eq. (13)), one arrives at the rate equation

r = {I + Ko r
(~t::) a
(73)

.h g
WIt 0: = f.
The different possible rate equations with the underlying assumptions are
summarized for the main surface species as follows:

(74) (75)

(76) (77)

(78) (79)

For Eqs. (74) and (75) a uniform surface is assumed, whereas for Eqs. (76) and
(77) a heterogeneous surface according to Brunauer is considered. Equations
(78) and (79) are associated with a heterogeneous surface with the Temkin-
Pyzhev approximation. Ozaki et al. pointed out that, at sufficiently high efficien-
cies, Eqs. (76) and (77) reduce to Eqs. (78) and (79). If 0: = 1, Eqs. (76) and (77) are
identical with Eqs. (74) and (75). It is noted that Eqs. (78) and (79) can be
regarded as approximations to Eqs. (74) and (75), in which case IY. will be without
any physical meaning.
As to the reason for the presence of NH instead of N on surface, Ozaki et al.
pointed out that the catalysts may not have been completely reduced. Since N. ds
is more thermodynamically stable than NH. ds the reason for the dominance of
NH. ds must be a kinetic one.
Brill and Tauster [57] carried out experiments on doubly promoted cata-
lysts. No experimental details about catalysts or reduction procedures were
given. The constant 0: in the Temkin-Pyzhev equation was increased from 0.11
to 0.7 for doubly promoted catalyst between 182 and 369°C and from 0.47 to 0.7
for singly promoted catalyst between 242 and 286 0c. There was no change in
168 1. B. Hansen

ex with efficiency. Brill pointed out that the change in (J. observed by Ozaki et al.
[56J could be due to temperature changes rather than efficiency changes. He
also noted that it is impossible to calculate the activation energy from the
Temkin-Pyzhev equation, when (J. is changing with temperature, because k is
a function of temperature according to
k=kt- a (k 2K p )a e -E/RT (80)
Using a Langmuir equation Eq. (74), i.e. Nads being the most abundant
reaction intermediate, Brill obtained a good fit to the data and calculated the
activation energy on the doubly promoted catalyst to be 17.4 kcal/mol.
Krabetz and Peters [58] used both the Temkin-Pyzhev equation and the
Langmuir formulation to test data from synthesis experiments on singly and
doubly promoted catalysts at atmospheric pressure and temperatures between
280 and 370 DC. For the doubly promoted catalyst, the Temkin-Pyzhev equation
fitted the data well if (J. = 0.8 at 333 DC and (J. = 0.7 at 372 DC. The Langmuir
equation (Eq. (80)) with N as most abundant reaction intermediate, also gave
a satisfactory fit yielding an activation energy of 84-105 kJ/mol. For the singly
promoted catalyst the best fit was obtained with the Ozaki-Taylor-Boudart
equation Eq. (75), i.e. NH as most abundant reaction intermediate. The activa-
tion energy was 59 kJ/mol.
Nielsen et al. [59J carried out an extensive kinetic study on a commercial
triply promoted KMIR (K 20, CaO, A1 20 3 ) catalyst. It had been prereduced in
the size range 3-6 mm but was tested in the size range 0.3-0.7 mm. The sample
was reduced again at 150 atm up to 400 DC after which the pressure was
increased to 300 atm and the temperature to 480 DC. The reactor had an internal
. diameter of 5 mm and was equipped with three thermocouples of 1 mm in outer
diameter in the catalyst bed. Total catalyst volume was 2.5 cm 3 .
The operating conditions studied were as follows:
Pressure: 149-309 atm abs
Temperature: 330-495 DC
Space velocity: 13200-105600 vOl/h
H2/N2 ratio: 6.23-1.15
The inlet gas did not contain any inerts or ammonia. A total of 35 runs at
different conditions were carried out. Throughout the experiment the stability of
the catalyst activity was checked by measurements at standard conditions. The
influence of different departures from ideal conditions on the reliability of the
measurements is discussed in the article. In most of the experiments, the
temperature was held within 5 DC, but in runs at high pressures and space
velocities, temperature differences up to 17 DC were observed. A weighed mean
temperature was used in the analysis of the experiments. Computer calculations
using a stepwise kinetic integration, simulating the actual temperature vari-
ations, showed that using such a weighed average temperature as the assumed
isothermal operating temperature removed any error from the runs where the
temperatures differences were below 5 DC and the difference between calculated
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 169

and measured ammonia outlet percentages were small even with the 17 °c
temperature difference.
The effect of diffusion was also investigated by means of computer calcu-
lations. A run at rather extreme conditions, 450°C, 317 atm, and space velo-
city = 32100 vol/h, had an effectiveness factor of 0.5 and 0.8 at 10 and 20% into
the catalyst bed respectively. In the main part of the bed, the effectiveness factor
was very close to 1.0 as in the majority of the runs. Flow conditions in the
experiment were in the viscous region or in the transition region between
viscous and turbulent flow. The non-ideal conditions due to flow and wall effects
were believed to have minor effects. The fact that the catalyst was reduced in the
3-6 mm size range affect the intrinsic activity due to the self-poisoning by water
diffusing out of the pores during the reduction process.
Nielsen et al. analyzed their data using the rate Eqs. ( 74), (75), (76) and (77)
developed by Ozaki et al. [56]. They combined the four equations into
k 1 PNz
(81)

Introducing the equilibrium constant K a2 ;

K2 = aN
2
H3,eq (82)
a aHz,eq
3 a Nz,eq
and taking the reciprocal reaction (ammonia decomposition) into account and
substituting activities for partial pressures, they arrived at:

r= k~{aNZK; - (~)}
{1 + K3 (a:;'3 ) f"
(83)

Expressing
k~ = k20 e - Ez/RT (84)
K3 = K30e-E3/RT (85)
it can be seen that Eq. (83) contains six unknown coefficients k20' E 2, K30' E 3,
wand a. The best values for the constants were found by computer calculations
to be
K 30 = 3.07 X 10- 2 E3 = - 19361 cal/mol
lk20 = 1.06 X 1013 E2 = 17429 cal/mol
w = 1.564 a = 0.640

1 A printing error in [59] giving k 20 as 2.12 x 10 13 or twice the correct value was later corrected by
Nielsen in [60].
170 J. B. Hansen

Equation (83) fitted the data at 370, 410, 450 and 490°C quite well when
these values are used, whereas it predicted too high reaction rates at 330 °C
compared to the measured ones. In this connection it should be remembered
that even minute amounts of oxygen-containing compounds will severely retard
the reaction rate, especially at low temperatures, as will be discussed later.
The authors pointed out that the second term in the denominator bracket far
exceeds 1 under most circumstances. In fact, only when the ammonia content is
down to around 0.1 %, is it comparable to 1. This means that Eq. (83) is reduced
to the original Temkin-Pyzhev equation. For the Temkin-Pyzhev equation,
Nielsen et al. derived an apparent activation energy of 42300 kcaljkmol with
IX = 0.64. The rate constant was found to be independent of pressure, which was
ascribed to the use of fugacities instead of partial pressures.
Temkin et al. [61] studied the rate of ammonia synthesis far from and near
the equilibrium on a doubly promoted catalyst at and below atmospheric
pressure. The catalyst, of 0.1 to 0.2 mm in size, was reduced in the synthesis gas
at a space velocity of 100000 h - 1 and a final temperature of 550°C. Batches of
0.02 to 0.05 g catalyst were used in the experiments. In one series of experiments
the partial pressure of nitrogen was fixed at 100 mm Hg and the hydrogen
pressure varied from 100 to 600 mm Hg. In another series the opposite was
done. Temperatures were 450°C and 350 °C in both series and the space
velocities used ranged from 138000 h - 1 to 2640000 h - 1, which ensured an
ammonia percentage less than 0.01 mol %, i.e. far from equilibrium.
The data could be reproduced with a kinetic equation of the form
r = k' p Hz
O. 5 p O. 5
Nz (86)
Another series of experiments were carried out at 350 and 450°C at atmospheric
pressure, but varying the H2/N2 ratio from 8.5 to 0.15. The data from this
experiment could also be fitted to Eq. (86). The apparent activation energy for
k was found to be 11.5 kcaljmol.
The authors interpreted the applicability of Eq. (92) as an indication that, at
low ammonia concentrations, the equilibrium between adsorbed nitrogen and.
ammonia plus hydrogen in the gas phase is no longer established.
Another set of experiments was carried out with stoichiometric gas near or at
an intermediate distance from equilibrium. The pressure was fixed at 1 atm and
the temperature varied in steps of 25°C from 325 to 550°C. The space velocity
ranged from 61000 h -1 to 542000 h -1.
The original Temkin-Pyzhev equation (Eq. (21)) with IX = 0.5 reproduced the
data above 400°C reasonably well. The activation energy for k2 was found to be
constant at 37.0 kcaljmol above 400°C. This corresponds to an activation
energy for k1 in Eq. (21) of 12.0 kcaljmol which is close to the 11.5 kcaljmol
found for k' in Eq. (86). At the temperatures of 325 and 350°C, k2 in Eq. (22)
declined at lower ammonia concentrations indicating that Eq. (22) is no longer
valid because the conditions are far from equilibrium.
In a later communication [62], Temkin et al. developed a rate equation
covering the complete range of operating conditions including the transition
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 171

region between Eq. (21) and Eq. (86). They assumed an associative mechanism
for the synthesis as follows:
"'1

Step 1 N2 + * <=± N2* (87)

(0- 1

"'2_
Step 2a N 2* + H2 <=± N 2H2* (88)
"'-1

Step 2b N 2H2* + 2H2 <=± 2NH3 + * (89)

Only step 2b is assumed to be sufficiently rapid so that equilibrium is
established. Another assumption is that adsorbed N 2 covers the surface to
a moderate degree, whereas N 2H 2, N2H4 etc. only cover a small fraction of the
surface. Furthermore, it is assumed that the adsorption energy for N 2 varies
linearly with surface coverage as discussed earlier. For N2H2 the adsorption
energy is assumed to be independent of coverage. Using these assumptions the
following rate equation for the synthesis is obtained:

0p 0 X-2. P~H3
op
1 II X2. H2 - X-I KPl
XI N2
W =____ 2b H2 (90)
f sin exII (op X-2. p~H3)a(,0 0 p )l-a
Xl N2+K -p2 X-l+X2. H2
2b H2
where
f=~
RT
and X?, X?. I and xg., X?. 2. are the rate constants at zero surface coverage for the
reactions described by step 1 and 2a. These are related to the equilibria
constants
o
XI
K oI _- -o- (91)
X-I
o
K O _ X2. (92)
2. - 0
X-2.
and
K?Kg.K 2b = K (93)
Introducing the substitutions

(94)

and
X?. I
1 =-0- (95)
X2.
172 1. B. Hansen

together with the equilibrium relations, one obtains

k P 1- a (1 _
± Nz
~ P~H3
K P p3
)
N2 Hz (96)
ill = (_1 + ~ P~H3 )a(_1 + 1)1 a
P Hz K P Nz P~z P H2
It can be seen that

(97)

If the probability of hydrogenation of adsorbed nitrogen is much higher than

the probability of nitrogen desorption, it follows that

(98)

which obviously requires that the hydrogen partial pressure is not negligible.
Close to equilibrium, the term

will be close to 1, and Eq. (96) approaches the normal Temkin-Pyzhev equation
(Eq. (21)). Ifthe system is very far from equilibrium, then
1 P~H 1
---~-«­ (99)
K PNzP Hz P Hz

Equation (96) then converts into Eq. (86) deduced in Ref. [61]. Calculations
showed, however, that under industrial ammonia synthesis conditions the orig-
inal Temkin-Pyzhev equation is invariably applicable. However, it was neces-
sary to apply Eq. (96) in order to get constant values of Kad at 350°C and 325 °C
for the data reported in [61]. Additional data are given at 450°C in [62]. When
the total pressure was varied from 0.5-1.0 atm, the hydrogen/nitrogen ratio
from 0.5-5, and the space velocity from 200000 to 526000 h - \ the relative
yields Z/Zeq were in the range 0.13 to 0.35. The calculated L values were
satisfactorily constant, although a tendency towards lower k+ values with
higher yields could be observed.
Bridger and Snowdon [63] cited tests from both differential and integral
reactors without giving any experimental details. They found the extended
Temkin-Pyzhev equation, Eq. (96), to describe the experimental data best with
a = 0.465 and E2 = 40.4 kcaljmol.
Aika and Ozaki [97] did experiments on 10.4 g of unpromoted catalyst at
305°C at and below atmospheric pressure. The catalyst had been reduced at
400°C for 200 h and stabilized for approx. two weeks. They tested three kinetic
equations for the best fit to the experimental data.
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 173

1) The Temkin-Pyzhev equation Eq. (21)

2) A Langmuir-type equation with NH as the main adsorbed species.
3) A Langmuir-type equation with N as the main adsorbed species.
The least sum of squared standard deviations between measured and cal-
culated rate constants

(100)

was found with equation type 3.

Aika and Ozaki thus found N the most abundant surface species at 305°C, in
contrast to Ozaki et al. who found it to be NH. Aika and Ozaki attributed this
discrepancy to the absence of promoters in the catalyst they used. The fact that
Nielsen et al. [59] also found N to be the most abundant species for a doubly
promoted catalyst, and the Temkin-Pyzhev equation best, they ascribed to the
higher temperatures used in the study of Nielsen et al. compared to the low
temperature of 305°C used by Aika and Ozaki.
Logan and Philip [65] analyzed the data of Ozaki et al. again. They used the
extended isotherm and rate ofN 2 adsorption developed by Brunauer et al. [27]
as did Ozaki et al. The quality of the fit was assessed by the variance as defined
by

(101)

where Ya and y denote the measured and calculated NH3 mole fractions
respectively. Ozaki selected the best linear log-log plot ofYa vs l/v, where v is the
flow rate.
However, Logan and Philip did not as Ozaki et al. limit the value of g/f = IX
to 1.0 and indeed found a best fit with IX around 0.8. Furthermore, they found
a more normal isotope effect of Kh/Kd = 3.8 at room temperature consistent
with N being the most abundant species on the catalyst surface and adsorbed
H participating in the slow step of nitrogen dissociation.
Based on certain circumstantial evidence, Carra and Ugo [66] proposed
a mechanism where the rate-determining step is the hydrogenation of a half-
hydrogenated surface nitrogen, dehydroimide (N2H). This leads to the following
rate equation using the Langmuir adsorption isotherm
KP N2 (b h p H ,)O.5
r = 1 + (b P )05 (102)
H H2 · + b NH3 + PNH3

which can be approximated by

kKb o 5 p1.5 P p1.5 P
r = _ _H_ H2 N2 ~ keff H2 N2 (103)
bNH3 P NH3 P NH3
which is identical with the original Temkin-Pyzhev equation with IX = 0.5.
174 1. B. Hansen

Takezawa [67, 68] determined the rate of ammonia decomposition on

a triply promoted catalyst (4.72 A1 2 0 3 , 0.31 % KzO and 0.05% silica). The
pressure was kept constant at 1 atm with helium and the ammonia partial
pressure was varied between 0.25 and 0.88 atm, the nitrogen partial pressure
between 0 and 0.49 atm. They found the Temkin-Pyzhev equation

(104)

to be valid at 420°C with f3 = 0.48.

Above 479°C they found however

V = k2 ( p[; )0.75
p
(105)

to be valid. At slight inhibitive effect of nitrogen could be observed with

a reaction order around 0.1. They interpreted the findings by a change in the
rate-determining step from nitrogen desorption to dehydrogenation of adsorbed
amino radical NH z (a).
In order to determine the effect of potassium they eluted part of the
potassium oxide by immersing the catalyst in water, which gave a catalyst
containing 0.25 wt% KzO [69]. The rate of ammonia decomposition was
measured again at 378 DC, and the following rate dependency was found:

(106)

which indicates that dehydrogenation of an adsorbed amino group may be

rate-determining. Takezawa concluded that the change in rate determining step
at low temperatures was due to the lower potassium content of the catalyst.
Brill [70] applied the following rate equation proposed by Ozaki et al. [56]

(107)

with

(108)

(i.e. assuming a uniform surface with N adsorbed as the most abundant reaction
intermediate) to experimental data at low temperatures (approx. 265-340oq,
presumably atmospheric pressure, and a stoichiometric gas mixture using pure
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 175

iron as catalyst. Brill also applied the rate equation

dP NH3
-- = kP N (1 - 8) - k 2 8 (109)
dt 2

with

cK (Pp1.5
-1 NH3 )

8 = (110)
1+CK- (P
Hz

1 NH3 )
p1.S
H3

where K is the equilibrium constant for ammonia synthesis. This is equivalent to

assuming a uniform surface and hydrogenation of molecularly adsorbed nitro-
gen to be rate determining.
Both Eq. (107) and (109) in their integrated form represented data at 340°C
equally well. Equation (109) could, however, be adapted better to an Arrhenius
plot than Eq. (107).
Brill noted that Eq. (109) gives a better approximation to the integrated rate
equation,

kl = (~) P NH3 ,eq (111)

t P N2

than Eq. (107) at yields below 1% (whether x is the equilibrium yield

ZNH,/ZNH3,eQ) .
Kazarnovskaya et al. [71] made experiments in a recycle reactor at
400-450°C, and 200, 250, 300 atm with different particle sizes at four space
velocities. For the smallest particle size (0.25-0.5 mm), the rate constant in the
Temkin-Pyzhev equation with IX = 0.5 was calculated. The calculated rate
constants increased systematically with increasing space velocity, i.e. decreasing
ammonia concentration.
Dyson and Simon [72] used the data of Nielsen et al. to estimate the
parameters in the original Temkin-Pyzhev equation,

They calculated the fugacity coefficients in a manner slightly different from

Nielsen et al. and estimated k for IX = 0.5 and IX = 0.75.
Both values of IX gave reasonably good fits, and Dyson and Simon chose
IX = 0.5 as this gave the simplest rate equation. With IX = 0.5 they found:

2k = 1.7698 x 10lse-4o:T65

Dyson and Simon found, as Nielsen et al. did, the rate constant k to be
independent of pressure.
176 J. B. Hansen

Ferraris and Donati [73J reevaluated the data of Nielsen et al. using both the
Temkin equation Eq. (112) modified by Dyson and Simon and Eq. (83) used by
Nielsen et al. They used the same thermodynamic data as Dyson and Simon.
A non-linear parameter estimation program was used to calculate the constants
in the two equations minimizing the sum of squares defined by:
N
SSQ = L: (Zeale - Zmeas)2 (113)
i= 1

where Zeale and Zmeas are the calculated and measured ammonia mole fractions.
For the Dyson and Simon equation, Ferraris and Donati found
SSQ = 9.30 x lO-3 with oc = 0.5. The best fit was obtained with oc = 0.73 and
2k = 1.30 x 10 15 e -4~:30. This gave SSQ = 6.33 x 10- 3.
Using the values of Nielsen et al. Ferraris and Donati found
SSQ = 6.87 x lO-3. Reestimation of the parameters gave:
K30 = 2.1496 E3 = - 18458 caljmol
K 20 = 1.349 X 1015 E2 = 16480 caljmol
W = 1.574 oc = 0.6923
The SSQ became 6.26 x lO - 3.
In a later article [74J, Ferraris et al. again used the data by Nielsen et al. to
test 23 different rate equations for ammonia synthesis. Some of the rate equa-
tions considered both associative and dissociative mechanisms. Others were
purely empirical. They concluded that a large number of the models represented
the data equally well, in fact better than the Temkin-Pyzhev equation. This
again proves that it is extremely difficult, and in fact usually impossible, to draw
definite conclusions about the mechanism from kinetic data alone.
Cappeli and Collina [75J investigated the kinetics of five different commer-
cial ammonia synthesis catalysts. The following rate equations were tested:
1) The Temkin-Pyzhev equation modified by Dyson and Simon (112)
2) the equation proposed by Nielsen et al. (83)
and 3) a new equation:

(114)

The new equation was claimed to give the best description of the experi-
ments. The k4aNH3 term was found to be insignificant compared to the other
terms in the denominator. Industrial data from a third bed (adiabatic) with
intermediate cooling were also well described by Eq. (114).
Guacci et al. [76J tested eight different commercial ammonia catalysts under
industrial conditions. They used the Temkin-Pyzhev equation modified by
Dyson and Simon to analyze their results. The oc values found for the different
catalysts ranged from 0.426 to 0.687
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 177

Altenburg et al. [77] carried out kinetic studies on a commercial, promoted

catalyst (AI 20 3 2.90, K 20 1.30, Zr02 0.45 and Si0 2 0.45 wt%). The potassium
content was varied from 0.1 to 3.8 wt% by elution and reimpregnation of
prereduced catalyst with KOH. Three temperature levels of 350, 400 and 450°C
were used. The pressure was varied from 5 to 150 atm. Altenburg et al. used the
modified Temkin-Pyzhev equation

(115)

to analyze the results. kl' ill, and Q( were estimated by minimizing

SSQ = 2:(1 _ kl
kl
meas)2
calc
(116)

at each temperature level. The Q( values varied from 0.48 to 0.79 in a nonsyste-
matic manner, not correlated with the potassium content. The reaction order of
hydrogen through the value of ill did, however, increase systematically from 0.71
to 1.55 with increasing potassium content.
Altenburg et al. speculated that this trend could be explained by assuming
a fast reaction to NH, and other partly hydrogenated species between adsorbed
Nand H from polarized hydroxylic groups on the oxidic part of the catalyst,
followed by a reaction between hydrogen and the 0 groups. This pathway is
blocked by potassium either by reaction with the acidic OH groups or by
screening of adsorbed N from these groups. They noted that the promoting
effect of potassium was most pronounced at higher pressures.
tU Zhenming [78] tested two different commercial ammonia catalysts using
the Temkin-Pyzhev equation for analysing the results. The pressure was fixed at
around 148 atm and the temperature was varied from 338-550°C. The catalyst
size was 0.2-0.3 mm, and a stoichiometric feed gas was used. The optimum
Q( value was 0.5, and activation energies of 42.1 and 44.8 kcaljmol were found.

Using slightly different mechanistic assumptions than Temkin et aI., Huan

Kahui [79, 80] derived a kinetic equation similar to the one proposed by
Temkin et al. Liu De-ming et al. [81] tested Huan Kahui's equation on data
obtained with an industrial catalyst and compared the fits with other rate
equations proposed by Capelli and Collina [75], Dyson and Simon [72], Buzzi
Ferraris et al. [75], and Nielsen et al. [60]. The modified Temkin equation was
found to give the best fit.

4.5 Kinetics Based on Surface Science Techniques

Bowker et al. [82] used computer calculations to extrapolate the results ob-
tained by Ertl et al. [83] at low pressures and coverages to industrially relevant
conditions around 450°C and above 100 atm. They used the reaction mecha-
178 J. B. Hansen

nism proposed by Ertl et al.

Nz + * -+ Nz-*
N 2 -* + * -+ 2N-*
Hz + * -+ Hz-*
Hz-* + * --+ 2H-*
N-* + H-* --+ NH-* + *
NH-* + H-* -+ NHz-* + *
NH z-* + H-* --+ NH3* + *
NH 3-* -+ NH3 + * (117)

and assumed iron surface area for a commercial ammonia synthesis catalyst of
1.5 m Z/g and an energetic homogeneous surface where single-site Langmuir
isotherms describe the adsorption processes. To each of the elementary reac-
tions a preexponential factor and an activation energy were assigned. The
preexponential factors were estimated from transition state theory. All adsorp-
tion and desorption reactions of molecular Hz, N z and NH3 were assigned
a preexponential factor of 10 13 , all other reactions a factor of about lO z1 . The
experimentally available activation energies at low surface coverages were taken
from the work of Ertl et al. Knowing the overall heat of reaction (50 kJ mol- 1 )
and assigning an activation energy of 21 kJ mol- 1 to all the dehydrogenation
steps, the complete potential energy diagram could be constructed.
Using these parameters the performance of a back mix reactor was cal-
culated at conditions similar to the experimental data of Nielsen et al. [59]. The
model predicted ammonia percentages which were 1.5 x 10 5 lower than those
experimentally found.
Lowering the preexponential factors for both the forward and reverse of step
2 by a factor of 10 7 only improved the fit slightly. However, the model predicted
that the surface was totally hydrated instead of being totally nitrided as in the
case of "normal" preexponential factors of lO z1 . Using the energies determined
at high surface coverages had virtually no effect on the model predictions. The
authors attributed the very poor agreement with the experimental findings to
an energy well in the model too deep for adsorbed nitrogen atoms in the
model.
Reasonable fits could be obtained (within 10% of experimental values) if
activated dissociation of nitrogen was invoked as, found by Scholten et al. [85]
on singly promoted catalysts. The authors point out, however, that there is no
surface science experimental evidence for activated nitrogen dissociation on
potassium promoted iron.
Stoltze and N0rskov [85-88] used almost the same reaction scheme as
Bowker et al. i.e.:
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 179

Step 1 N 2 (g) + * -> N r *

Step 2 N 2 -* + * -> 2N-*
Step 3 N-* + H-* -> NH-* + *
Step 4 NH-* + H-* -> NH r * + *
Step 5 NH r * + H-* -> NH r * + *
Step 6 NH 3-* -> NH3(g) + *
Step 7 H2(g) + 2* -> 2H-* (118)

the only difference being that a molecularly held precursor state of H2 is not
considered for the dissociation of hydrogen. The following assumptions were
made: the gas is considered to be ideal and all surface sites are assumed to be
equal (homogeneous surface). Adsorbate-adsorbate interaction is thus con-
sidered to be absent until the coverage reaches one adsorbate atom per two iron
atoms. Above this coverage adsorbate-adsorbate repulsion excludes further
adsorption. Hydrogen and nitrogen as well as reaction intermediates are as-
sumed to chemisorb competitively. The active area of the catalyst is taken to be
that of the CO chemisorption area. Finally, all the reaction steps are assumed to
be in equilibrium except step 2 i.e., this step is assumed to be the sole rate-
determining step.
By generalizing Fowler and Guggenheim's [89] statistical mechanical de-
scription of adsorption that covers competitive gas adsorption, the equilibrium
constants were calculated. For instance, for step 1:

K 1 -_ ZN2'
0 (119)
ZN2

where ZN2' is the partition function for N2(adsorbed) and Z~2 the partition
function for N2 (gas) at the thermodynamic reference pressure Po taken to be
101.325 kPa. The partition functions were calculated from

(120)

Ztran" Zvib, Zrot are the partition functions for translational, vibrational, and
rotational degrees of freedom, whereas Ex is the ground-state energy for the
species. The equilibria can be specified as follows:

Po
N2
Kl ( P ) 8* = 8N2* (121)

K3 8N*8 H* = 8NH*8* (122)

K4 8NH*8 H* = 8NH2 *8* (123)

Ks8NH2*8H* = 8NH3*8* (124)

180 J. B. Hansen

P NH3
K 6 0NH3 * = To 0* (125)

K7 (::2 }O*)2 = (ON*)2 (126)

where Ox is the surface coverage by species X, and 0* is the fraction of the surface
not covered by any surface intermediate. For step 2 the rate can be calculated
from
(127)
where
(128)
Noting that
(129)

and solving for the different surface coverages, it is found that

(130)

The rate of ammonia synthesis is then determined by:

(131)

where Kg is the gas phase equilibrium constant for the synthesis.

In order to determine the input parameters for the model, parallel basic
thermodynamic data were taken from [90], frustrated vibration was assumed to
have a barrier for surface diffusion of 0.05 eV for N r * and 0.5 eV for other
intermediates, and vibrational frequencies for adsorbed species were taken from
electron energy loss spectra. TPD spectra for H-* and NH3* and N-* together
with the initial sticking coefficient for N2 into 2N-* on K covered iron were used
to determine the ground-state energies and A2 and AH}. Values for the ground
state energy ofNH-* and NH 2-* were assigned to be between the values ofN-*
and NH 3*. The active area determined by CO chemisorption was used in the
calculations.
The validity ofthis model was assessed by comparison with the experimental
data of Nielsen [60]. The maximum deviation between observed and calculated
rate constants for the ammonia synthesis is 1.4, despite the wide span of
pressures from 1 to 300 atm suggests good representation of the kinetics and
mechanism by the model.
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 181

The surface coverages by reaction intermediates could also be calculated by

the model. It was found that N-* is by far the most abundant surface intermedi-
ate under typical synthesis conditions, but the sum of the coverages by other
intermediates is also larger than the coverage by free sites. This fact is seen by the
authors to be the cause for the complicated kinetics deduced by Ozaki et al.
[56], rather than adsorbate-adsorbate interaction or surface heterogeneity.
The data of Temkin et al. [61] for low conversions were also analysed by this
model in ref. [87] to adequately describe the observations. The failure of the
normal Temkin-Pyzhev equation was found to be due to high coverage by H-*,
rather than to a change in the rate-determining step. Stoltze and N0rskov also
demonstrated how to deduce the apparent reaction orders and activation
energies from the model.
Bowker et al. refined their model proposed in [91] by modifying the
preexponential factors for the molecular nitrogen adsorption and desorption in
step 2 to 10 11 . 2 and 10 10 . 2 , respectively, and the factor for the dissociation of the
molecularly held nitrogen to 10 15 . 2 , so that these agreed with the experimental
findings of Ertl et al. [92, 91]. The reactor model was furthermore changed to
a plug flow type as used in the experiment by Nielsen et al. [59]. The ammonia
percentage was calculated to be 3.0%, while the experimental value was 13.2%.
Using the heat of dissociative nitrogen adsorption and the lower preex-
ponential factor proposed by Stoltze and N0rskov, Bowker et al.'s model gave
an ammonia exit concentration of 5.4%, which corresponds to a catalyst activity
40 times lower than that experimentally found. Bowker et al. also calculated an
erroneous TPD peak position and peak width using Stoltze and N0rskov's data.
The poor fit is again ascribed to too high a surface coverage of nitrogen, causing
"self-poisoning" to occur.
Bowker et al. again pointed out that a net activation barrier to nitrogen
dissociation would make the model predict yields closer to the experimentally
observed ones. The authors believe that measurement of the nitrogen desorption
energetics at high nitrogen coverage are necessary as nitrogen dissolution into
the bulk phase might playa role at high nitrogen coverage.
Stoltze and N0rskov [94] in a comment to the article by Bowker et al. stated
that their parameters were chosen so that the TPD spectrum for nitrogen
desorption is reproduced exactly. They ascribed the indequacy of the model of
Bowker et al. to unrealistic choices for preexponential factors. They pointed out
that in case of hydrogen chemisorption these choices lead to an entropy of
chemisorption of zero, which clearly contradicts the experimental findings.
Bowker again in Ref. [105] discussed the different kinetics proposed based
on surface science studies. Bowker referred to the work of Rettner and Stein
[106] in which a well defined molecular beam of nitrogen was used to measure
the sticking probability on Fe(111) surfaces. Their measurement showed an
increase in the sticking probability from 10 - 6 to 10 - 2 with an increase in
average beam energy from 0.1 to 1.0 eV. This is consistent with an activation
energy for nitrogen activation of around 35 kllmol in good agreement with the
result of Scholten for zero coverage [84]. Bowker pointed out that Ertl et. al.
182 1. B. Hansen

table 4.1. Parameters of kinetic models

Kinetic Step Activation Preexpontial Ammonia Surface

model energy factor yield coverage
kJ/mol S-1 % %

Ertl 1 0 10" H(a) 12.1

[92,93] ~1 46 1010 3.0 N2(a) 1.5
2 31 10 7 N(a) 84
~2 199 10 13 NH(a) 2
Modified 1 0 10" H(a) 0.6
Ertl ~1 46 10 10 5.4 N 2(a) 0.1
[85,86,87J 2 31 10 10 N(a) 98.3
~2 144 10 10 NH(a) 1
Activated N 2 1 0 10 13 H(a) 22
dissociation ~ 1 46 10 13 14.2 N2(a) 3
[105J 2 96 10 13 N(a) 54
~2 134 10 13 NH(a) 8

[92, 93J in their work did not vary the gas temperature, but only the temper-
ature of the solid. The three different kinetics proposed are summarised in table
4.1, where the different activation energies and preexponential factors for the
critical steps 1 and 2 in the reaction scheme (118), i.e.
Step 1 N 2 (g) + * -+ N 2 -*
Step 2 N 2 -* + * -+ 2N-*
used in the kinetics, are given together with the yield calculated from the models
(experimentally found 13.2%) and the surface coverages by the predominant
speCIes.
Bowker concluded that it is necessary to invoke an activated nitrogen
dissociation step in order to reproduce the experimental findings on industrial
catalysts as well as the measured desorption spectra widths. It is not clear from
the article, whether the surface coverages given in the table are average or
typical for the exit of the reactor. Bowker discussed the differences between the
different crystal planes of iron, both promoted and non promoted, but he still
assumed in his model the surface to be homogeneous.

4.6 Concluding Remarks

This review has illustrated, that the large amount of work done on the kinetics of
ammonia synthesis and decomposition has stimulated new concepts of reaction
kinetics catalyzed by heterogeneous catalysts. The concept of a non-uniform
surface vis-a-vis a homogeneous surface is just one example, which is still
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 183

debated today. Ammonia synthesis also is one of the first reactions, where micro
kinetics based on modern surface science techniques has been used with a fair
degree of success. However, there is no doubt that the finer details of the
ammonia synthesis reaction mechanism and kinetics still will be a controversial
topic for years to come.
In summary, it is noteworthy that the first workable rate equation for
ammonia synthesis was proposed before the second world war by people with
very limited experimental resources. The Temkin-Pyzhev equation still is the
rate equation of choice for engineering purposes. Further progress in better
understanding of the fundamental rate-determining process(es) is still, however,
needed in order to develop better catalysts.

4.7 Transport Phenomena

The kinetics discussed in the preceding section has been (or should have been)
derived under conditions where transport processes do not restrict the measured
reaction rate. In ammonia synthesis as with many other catalytic reactions it
may be necessary to consider also the mass and heat transfer to and inside the
catalyst particle, depending on operating conditions and catalyst particle size.
If mass and/or heat transfer is slow compared to the reaction rate, temper-
ature and concentration gradients between bulk gas and particle center will
result, thus affecting the overall reaction rate of the particle. In industrial
reactors the diffusion inside the catalyst pore system is the most important
transfer process. In laboratory reactors with low linear velocity and/or large
particles, there may also be significant concentration and temperature gradients
between particle surface and bulk gas [95, 96]. Due to the high heat conductivity
of the iron catalyst, a catalyst pellet will be almost isothermal [96].
The intrinsic activity of an ammonia synthesis catalyst is also known to
depend on the particle size during reduction [60,97,43]. This effect is due to an
irreversible poisoning of the outer reduced part of the particle by water from the
inner, still reducing part. The water causes the iron particles to recrystallize,
resulting in a lower surface area and hence lower activity. This effect can be
circumvented by a very careful reduction at high space velocities; this is possible
in laboratory reactors, but normally not feasible in industrial converters.
Thus the lower activity of large particles does not necessarily mean that the
reaction is diffusion limited at the operating conditions; diffusion retardation
can be recognized, however, by a lower activation energy for the larger particles.
The activity of a catalyst particle is often expressed by the effectiveness factor
11 defined as the ratio of the actual reaction rate R to the reaction rate Rkin at
conditions on the outer surface of the particle.
(132)
184 J. B. Hansen

0.8

Qld..
1::.
b) e 425°C
e" • 450°C
o 4750C
0.6
A t:. 5000C

° 525°C
t= 0.4
Fig 4.1. Dependence of the effectiveness
factor, on the ammonia concentration
0.2 z at different temperatures and two differ-
ent sizes of the catalyst pellets, a) 4.5 mm
b) 10 mm. SA-1 catalyst, P = 300 atm
0 (from [98])
0 0.1 0.2 0 0.1 0.2
z z

The effect of the operating conditions and particle size on 1'/ is illustrated in
Fig. 4.1 [98]. It is mainly the concentration of ammonia that varies throughout
the pellet, whereas the relative changes in the N z and Hz concentrations are
much smaller. Therefore the effectiveness factor depends mainly on the diffusion
rate of ammonia.
If the diffusion coefficient in a straight cylindrical pore is denoted by D, the
effective diffusion coefficient Deff in the catalyst is given by

Deff =-D
e (133)
T

where e is the porosity of the catalyst particle and T the so-called tortuosity
factor depending on the geometry of the pores.
D can be calculated by the methods outlined by Satterfield [99]. Under
industrial conditions, i.e. at high pressures, diffusion occurs mainly in the bulk
and D is thus the bulk diffusion coefficient of the components in the mixture. At
low pressures Knudsen diffusion will also playa role, and the pore volume
distribution must be known in order to calculate D.

4.7.1 Mathematical Model

For a spherical catalyst grain, the concentration C i at radial distance r of the

catalyst is given by

r z -d [Di
- dCiJ
- +v· R -- 0 (134)
dr rZ dr 1

with the boundary conditions

de
_I = 0 at r = 0 (135)
dr
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 185

and·
(136)
where D j and Vj are the effective diffusion coefficient and stoichiometric coeffic-
ient of component i. R is the reaction rate for a component with Vj = 1, and rp is
the particle radius. If the particle is not spherical, the proper particle dimension
is
(137)
Vg/Ag is the ratio of surface to volume of the catalyst grain. Values of the
tortuosity factor reported in different investigations are shown in Table 4.2.
In solving (134) D j is usually considered constant, although D j depends on
the gas composition. This approximation seems reasonable considering the
uncertainty in r. For simple reaction kinetics, the solution to (134) is given in
terms of the Thiele modulus [102]. The more complicated kinetics for ammonia
synthesis does not give a simple analytical solution. Generally, a numerical
integration has to be carried out.
Bokhoven and Rayen [97J measured reaction rates on 0.5-0.7 mm and
2.4-2.8 mm particles at 1, and 30 atm and 325-550°C. The effective diffusion
coefficient of NH3 was calculated from the results of O 2 diffusion measurements
on a catalyst with the same surface area and porosity as the catalyst used in the
activity measurements. The authors approximated the reaction rate by a pseudo
first-order reaction suggested by Wagner [103J to calculate the effectiveness
factor. Good agreement between measured and calculated data was obtained.
However, the approximation above is only good, if the reaction is near equilib-
rium.
Peters and Krabetz [43J measured reaction rates at 1 atm on Al 2 0 3 pro-
moted iron catalysts with varying particle size (0.5-5.2 mm) and mean pore
radius (110-310 A). The effective diffusion coefficient ofNH 3 was also measured
in a Wicke-Kallenbach experiment. They also found good agreement between
the measured and calculated data using a pseudo first-order approach.
Dyson and Simon [72J calculated 1] for a 6-10 mm size industrial catalyst
(using rate data reported by Nielsen [59J) by numerical integration of the

Table 4.2. Tortuosity factors for different catalysts and the method by which they
were determined

Catalyst e r Reference Method

Fe + MgO, SiO 0.52 3.8 Bokhoven and van Diffusion

Rayen [97] Measurement
KMI 0.45 2.2 Nielsen [100] Electron microscope
study of pore system
GK-l 0.22 16 Rusov [101] Rate measurements
GK-2 0.30 6.5
SA-l 0.12 3.0
186 J. B. Hansen

differential equation. Their model also includes bulk flow due to volume change
during the reaction. An expression for the effectiveness factor as a function of
conversion and temperature was given at 150, 225, 300 atm.
Nielsen [100] also solved the simultaneous differential equations numer-
ically and reported the results for a hypothetical reactor operating at 450°C,
214 atm, and with a 3: 1 H2/N2 feed gas containing 3% ammonia and 12%
inerts. The concentration profiles calculated at the bed inlet and outlet are
shown in Figs. 4.2 and 4.3. Nielsen used r = 2.2, based on a detailed, elec-
tron-microscopic study of the pore system. Figs 4.4 and 4.5 show the calculated
ammonia concentration in the bulk at half pellet radius and the effectiveness
factor for particle diameters of 5.7 mm and 1.5 mm as a function of axial
distance. For the 1.5 mm particles, the effectiveness factor very quickly ap-
proaches 1. The higher reaction rates of the 1.5 mm particles are utilized in
modern radial flow converters.
Rusov et al. [101] investigated a complicated model that also included molar
flux due to the volume change during the reaction and different diffusibilities of
the various components. They concluded, that using a simple model with
constant D j = DNH3 for all components was almost as accurate as the complic-
ated model. The reaction rates for five different catalysts were measured in an
ideal back-mix reactor operating at 400-525°C and P = 300, 450 atm. The
kinetic parameters in the Temkin-Pyzhev equation were estimated on crushed
particles. DNH3 and r could be calculated from rate measurements on large
particles. The tortuosity of the investigated catalysts varied between 3 and 15. It
should be noted that the porosity of these investigated catalysts was rather low.
In the above-mentioned investigation (J. changed from 0.35 to 0.9 upon
. increasing the temperature. Beskov et al. [98] showed that the reaction rates on

~ 454~ ~
450'---- - - - - - - - - - - '

24 72

16 48 N
Z0
0-
•
M
J:
Z N
o J:
0- 0
0-

8 24
Fig 4.2. Radial concentration profiles in 5.7 mm
nitrogen
ammonia catalyst particles located at the bed inlet.
Pressure 214atm, temperature 450°C, inerts 12%
(from [100])
O~----------------~O
Center Surface bulk
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 187

~ 454~ j
450~t:-============::::1-

--
24 72

ammonia

16 - hydrogen ............." 48 -£
;/!
+
N
l:
;e
8 r- - 24
Fig 4.3. Radial concentration profiles in 5.7 mm
nitrogen ammonia catalyst particles located at the bed out-
let, Pressure 214 atm, temperature 450°C, inerts
13.7% (from [100])
o o
Center Surface bulk

16 (2) 0.8 Fig 4.4. Axial profiles of NH3 in

the bulk (1) and at half pellet radius
z'"
l: (2) as well as effectiveness factor
III E (3) for particle diameters of
.,.e 0.4
0 5.7 mm. Pressure 214 atm, temper-
ature 450°C. inerts 12%, NH3 3%,
N2 21.25%, H2 63.75%, feed rate
6700 kmoljh (from [100])
00 2 4 6 8 100
Axial distance from inlet.M.

0.8

£z Fig 4.5. Axial profiles of NH3 in

III the bulk (1) and at half pellet radius
;e 0.4 (2) as well as effectiveness factor
E(3) for particle diameters of
1.5 mm. Other conditions as in Fig.
4.4 (from [100])

Axial distance from inlet. M.

large particles could also be fitted to the Temkin-Pyzhev equation, but with
lower values of IX. In the temperature range 450-525 °e, IX changed only slightly
(IX = 0.4-0.5) for the particle sizes 4-5 mm and 10 mm.
Kazarnovskaya et al. [71] reported a lot of data from an experiment in
a recycle reactor with four different particle sizes. The reactor operated at
188 J. B. Hansen

400-550 DC, P = 200, 250, 300 atm and four different space velocities. For the
largest particle size (7-10 mm) they found an apparent activation energy of
20 kcaljmol. This was half the activation energy for the small particle size
(0.25-0.5 mm).
Sokolinskii et al. [104] found the following approximate expression for 1'/ in
reasonable agreement with the experimental data reported in [71] for high
values of 1'/:

1+ ( -C)2
1] - 1 _ ~ r~Robs Ceq
(138)
- 15 DNH3C
1 _ (~)2
Ceq

where Robs is the observed reaction rate, and c and Ceq are the concentration and
equilibrium concentration of ammonia, respectively. This expression can be
used to determine whether the measured rate data are diffusion-limited.

4.8 References

l. Larson AT, Tour RS (1922) Chern Met Eng 26

2. Almquist JA, Crittenden ED (1926) Ind Eng Chern 18: 1307
3. Benton AF (1927) Ind Eng Chern 19: 494
4. Winter E (1931) Z Physik Chern B 13: 401
5. Taylor HS (1931) J Am Chern Soc 53: 578
6. Frankenburger W (1933) Z Elektrochem 39: 45,97,269
7. Emmett PH, Brunauer S (1934) J Am Chern Soc 56: 35
8. Emmett PH, Love KS (1933) J Am Chern Soc 55: 4043
9. Emmett PH, Brunauer S (1933) J Am Chern Soc 55: 1738
10. Taylor HS, Jungers JC (1935) J Am Chern Soc 57: 660
1l. Kozhenova KT, Kagan M Ya (1940) J Phys Chern USSR 14: 1250
12. Finkelstein WS (1934) Acta Physicochim USSR 1: 521
13. Chrisman IA (1936) Acta Physicochim USSR 4: 899
14. Finkelstein WS, Rubanik MJ (1935) J Phys Chern USSR 6: 1051
15. Roiter VA (1940) J Phys Chern USSR 14: 1229
16. Temkin MI, Pyzhev V (1939) J Phys Chern USSR 13: 851
17. Temkin MI, Pyzhev V (1935) Acta Physicochim USSR 12: 327
18. Temkin MI (1940) J Phys Chern USSR 14: 1241
19. Slygin A, Frumkin A (1935) Acta Physicochim USSR 3: 791
20. Zeldowitsch Ya, Roginsky SZ (1934) Acta Physicochim USSR 1: 449
2l. Langmuir I (1932) J Am Chern Soc 54: 2798
22. Evans MG, Polanyi M (1936) Trans Farad Soc 32: 1333
23. Emmett PH, Kummer JT (1943) Ind Eng Chern 35: 677
24. Kodama S, Fukui K, Mazume A (1951) J Che Soc Japan, Ind Chern Sect 54: 157
25. Temkin MI (1957) Khim Nauka Promst 2: 219
26. Love KS, Emmett PH (1941) J Am Chern Soc 63: 3297
27. Brunauer S, Emmett PH (1940) J Am Chern Soc 62: 1732
28. Frankenburger W (1955) Catalysis Vol. III, Reinhold Publ, NY
29. Brunauer S, Love KS, Keenan RG (1942) J Am Chern Soc 64: 751
30. Emmett PH (1946) Coll Chern 6: 214
 Kinetics of Ammonia Synthesis and Decomposition on Heterogeneous Catalysts 189

31. Temkin MI, Kiperman S (1947) Zhur Fiz Khim 21: 927
32. Sidorov IP, Livshits VD (1947) Zhur Fiz Khim 21: 117
33. Kobayashi H, Kubota H (1949) Rep Faculty Eng, Hokkaido Univ 3: 136
34. Temkin MI (1950) Zhur Fiz Khim 24: 1312
35. Nielsen A (1950) An Investigation on Promoted Iron Catalysis for the Synthesis of Ammonia,
Ed I, Gjellerup, Copenhagen
36. Temkin MI, Kiperman SL, Lukjanova 11 (1951) Doklady Akad Nauk (USSR) 74: 763
37. Brill R (1951) J Chern Phys 19: 1047
38. Annable D (1952) Chern Eng Sci Vol 1,4: 145
39. Kipermann S, Granovskaya VS (1952) Zhur Fiz Khim 26: 1615
40. Sidorov IP, Livshits VD (1952) Zhur Fiz Khim 26: 538
41. Adams RM, Comings EW (1953) Chern Eng Prog 49: 359
42. Bokhoven C, van Heerden C, Westrik R, Zwietering P (1955) Catalysis vol III, Reinhold, NY
43. Peters C, Krabetz R (1956) Z Elektrochem 60: 859
44. Scwab GM, Krabetz R (1956) Z Elektrochem 60: 855
45. Kawamura M, Irie T, Yokota H (1957) 60: 1239
46. Shiskova VV, Sid oro v IP, Temkin MI (1957) Tr GIAP 7: 62
47. Logan SR, Moss RL, Kemball RL (1958) Trans Faraday Soc 54: 922
48. Mills AK, Bennett CO (1959) Jour AIChE 5: 539
49. Kubota H, Shindo M (1959) Kagaku Kogaku 23: 242
50. Emmett PH (( [8J) Fixed Nitrogen, Reinhold, NY
51. Uchida H, Kuraishi M (1955) Bull Chem Soc Japan 28: 106
52. Boudart M (1956) Jour AIChE 2: 62
53. Stelling PO, Krusenstierna 0 (1958) Acta Chern Scand 12: 1095
54. Temkin MI (1949) Probl Kinet Katal 6: 54
55. Kwan T (1956) J Phys Chem 60: 1033
56. Ozaki A, Taylor HS, Boudart M (1960) Proc Roy Soc (London) A 258: 47
57. Brill R, Tauster S (1962) J Chern Phys 36: 2100
58. Krabetz R, Peters C (1963) Ber Bunsen-Ges Phys Chern 67: 381
59. Nielsen A, Kjrer J, Hansen B (1964) J. Catal 3: 68
60. Nielsen (1968) An Investigation on Promoted Iron Catalysts for the Synthesis of Ammonia, Ed
3, Gjellerup, Copenhagen
61. Temkin MI, Morozov NM, Shapatina EN (1963) Kinetics Katal 4: 260
62. Temkin MI, Morozov NM, Shapatina EN (1963) Kinetics Katal. 4: 565
63. Bridger GW, Snowdon (1970) Catalyst Handbook, Chapter 7, Wolfe Scientific, London
64. Aika K, Ozaki A (1969) J Catal 13: 232
65. Logan SR, Philip J (1968) J Catal 11: 1
66. Carra S, Ugo R (1969) J Catal 15: 435
67. Takezawa N, Toyoshima I (1966) J Phys Chern 70: 594
68. Takezawa N, Toyoshima I (1966) J Res Inst Catalysis, Hokkaido Univ 14: 41
69. Takezawa N, Toyoshima I (1966) J Catal 6: 145
70. Brill R (1970) J Catal 16: 16
71. Kazarnovskaya DB, Atamonovskaya RM, Bomshtein EI (1973) Zh Fiz Khim 47: 1445
72. Dyson DG, Simon JM (1968) Ind Eng Chem Fundam 7:605
73. Buzzi Ferraris G, Donati G (1970) Ing Chim Italiano 6: 11
74. Buzzi Ferraris G, Donati G, Rejna F, Carra S (1974) Chern Eng Sci 29: 1621
75. Cappelli A, Collina A (1972) I Chern E Symp Series 35, 5: 10
76. Guacci U, Traina F, Buzzi Ferraris G, Barisone R (1977) Ind Eng Chern 16, No 2: 166
77. Altenburger K, Bosch H, van Ommen JG, Gellings PJ (1980) J Catal 66: 326
78. Zhenming 1. (1981) J Chern Ind Eng (China) 1: 39
79. Huang KH (1981) Stud Surf Sci 7: 554
80. Huang KH (1980) Scientia Sinica 24, NO.6: 800
81. De-ming L, Shi-liang Z, Zi-xing H, Xi-jun W (1979) J Chern Ind Eng (China) 2: 133
82. Bowker M, Parker IB, Waugh KC (1982) Appl Catal 14: 101
83. Ert! G (1983) Catalysis, Science and Technology, Vol 4, Anderson JR, Boudart M (Ed)
Springer Berlin Heidelberg New York, p 273
84. Scholten JJF, Zwietering P, Konvalinka JA, de Boer JH (1959) Trans Faraday Soc 55: 2166
85. Stolze P, Norskov JK (1985) Phys Rev Lett 55, No 22: 2501
86. Stolze P, Norskov JK (1987) J Vac Sci Technol A 5(4), 581
87. Stolze P (1987) Phys Scripta 36: 824
190 J. B. Hansen

88. Stolze P, Norskov JK (1988) J Catal110: 1

89. Fowler R, Guggenheim EA (1960) Statistical Thermodynamics, Cambridge Univ Press,
Cambridge
90. JANAF Thermochemical Tables, US National Bureau of Standards, US GPO, Washing-
ton DC (1971)
91. Bowker M, Parker I, Waugh KC (1988) Surf Sci 197; L223
92. Ert! G, Lee SB, Weiss M (1982) Surf Sci 114: 515
93. Ert! G, Huber M, Lee SB, Weiss M (1982) Surf Sci 114: 527
94. Stolze P, Norskov JK (1988) Surf Sci 197: L230
95. Nielsen A (1970) Catal Rev 4: 1
96. Nielsen A (1977) Ammonia Part III, Fertilizer Science and Technology Series Vol 2,
Slack AV, Russell James G (Ed) Marcel Dekker, NY
97. Bokhoven C, van Rayen W (1954) J Phys Chern 58: 471
98. Beskov VS, Malinovskaya OA, Slinko MG, Streltsov OA, Rusov MT (1971) Teor Osn
Khim Tekhnol 5: 537
99. Satterfield CN (1970) Mass Transfer in Heterogeneous Catalysis, MIT Press, Cambridge,
Massachusett
100. Nielsen A (1969) Chim Ind (Milan) 51: 1052
101. Rusov MT, Slinko MG, Streltsov OA, Malinovskaya OA, Beskov VS, Loza AN (1971)
Teor Osn Khim Tekhnol 5: 382
102. Thiele EW (1939) Ind Eng Chern 31: 916
103. Wagner C (1943) Z Phys Chern 193: 1
104. Sokolinskii YA, Kazarnovskaya DB, Atamonovskaya RM (1973) Kinet Katal 14: 1351
195. Bowker M (1992) Catal Today 12: 153
106. Rettner C, Stein H (1987) Phys Rev Letts 58, 2768
Chapter 5
Poisoning of Ammonia Synthesis Catalysts

P. E. H0jlund Nielsen
Haldor Topsoe A/S Copenhagen, Denmark

Contents

5.1 Introduction. . . . . . . . . 192

5.2 Permanent Poisons. . . 192

5.3 Other Permanent Poisons. ...... 193

5.4 Temporary Poisons 193

5.5 Early Studies . . 194

5.6 Later Studies . 195

5.7 References. . . 197

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
192 P. E. Hojlund Nielsen

5.1 Introduction

An ammonia catalyst is poisoned when its catalytic activity is reduced due to the
presence of certain elements. These elements may be present in certain gaseous
compounds contained in the synthesis gas or solid and as such introduced to the
catalyst during the manufacturing process. With respect to the latter, reference is
made to Chap 2 by P. Stoltze.
Concerning gaseous catalyst poisons, a distinction should be made between
permanent poisons causing an irreversible loss of catalytic activity and tempor-
ary poisons which lower the activity while present in the synthesis gas. For a
review of poisoning of synthesis catalysts, see [1,2]. Permanent poisons accu-
mulate on the surface and may be detected by chemical analysis of the poisoned
catalysts, whereas temporary poisons cause a partial coverage of the catalyst
surface. Since oxygen is the most common temporary poison, it is difficult to
detect the amount on the spent catalyst by analysis since the promoter phases
are difficulty reducible metal oxides like alumina, magnesia, silica, and pot-
assium oxide.

5.2 Permanent Poisons

A permanent poisoning is caused by a compound whose presence in the

synthesis gas causes an activity decline, and the activity is not regained when the
compound is removed from the synthesis gas. The poisoning action may be due
to the formation of surface and/or bulk compounds with iron or one of the
promoters, in particular potassium, or the deposition of solid materials in the
pore system. For a discussion of permanent poisons, see [1,2].
Among these permanent poisons, sulfur is the most common. In some plants
based upon gasification of coal or oil, traces of COS and H 2 S may reach the
synthesis converter. In natural-gas-based plants, the Cu-based low temperature
shift catalyst and the Ni-based cleanup methanation catalyst act as efficient
guards. In most cases the sulfur poisoning is caused by lubricating oil. Thus, it
has been considerably reduced by the use of low sulfur oil and centrifugal
compressors instead of piston compressors [1].
Sulfur is chemisorbed very strongly by the iron surface, as discussed in
Stoltze's chapter. Studies of sulfur poisoning of the synthesis catalyst have been
carried out [5-8J, whereas in [3,4J, the interaction between single crystal
surfaces of iron and sulfur was discussed.
It was found that H 2 S adsorption led to an increase in the work function of
iron as was also the case for CO, CO 2 and H 2 0 [5]. But contrary to the latter,
 Poisoning of Ammonia Synthesis Catalysts 193

heating in H2 did not lower the work function to its unpoisoned level, indicating
an irreversible poisoning. Experiments on iron and single (AI 20 3 ) promoted
catalysts were carried out [6-8]. They found that about 0.2 mg S/m2 catalyst
surface completely deactivated the catalyst. The maximum uptake of S was
found to be 0.4-0.5 mg/m 2 catalyst surface. Chemisorbed sulfur could not be
removed by heating the synthesis gas up to 900 K. Catalysts to which sulfur was
added during activation were just as poisoned as the ones where sulfur was
added along with the synthesis gas. A very high level of sulfur was measured, and
bulk FeS was found on the spent catalyst. An amount of 6300 ppm of sulfur
completely deactivated the catalyst, whereas amounts as low as 200 ppm gave a
partial deactivation to the order of 15% [8]. Analysis of spent industrial catalyst
from the inlet layer reveals from 1000 to 5000 ppm of sulfur on catalysts KMI
and KMII [1]. Considerable poisoning by exposure to 20-30 ppm(vol) H 2S for
24 h at a space velocity of 10 000 h -1 was observed [1].

5.3 Other Permanent Poisons

In general, any compound that can lower the surface tension of iron may be
considered as a poison [9]. Phosphorus and arsenic compounds are also known
as permanent poisons [1]. However, in the natural gas-based plants they are
rare and only found in exceptional cases. A catalyst prepared with 1-3 wt%
PiOs was examined and found to be severely poisoned [10]. Chlorine [1] is a
serious poison that reacts with the potassium promoter and forms potassium
chloride which is slightly volatile and consequently removed from the catalyst
surface. It is expected that other halogens behave similarly [1]. Calculations
show that ppb amounts in the synthesis gas should result in a marked
deactivation [2].

5.4 Temporary Poisons

A temporary poison lowers the activity of an ammonia synthesis catalyst while it

is present upon its surface.
If there is no poison in the synthesis gas, the poison will be removed from the
surface and the catalyst will regain its activity. An oxygenic compound reacts in
that manner with the catalyst surface [1]. Unsaturated hydrocarbons may also
react as a reversible poison [1], but usually oxygenic compounds are the most
important, temporary poisons for ammonia synthesis catalysts.
194 P. E. Hojlund Nielsen

The most common oxygenic compounds encountered in ammonia synthesis

are H 20, CO, CO 2 and 02. The equivalence of these has been known since the
early studies on ammonia synthesis catalysts. This is due to the methanation
activity of the ammonia synthesis catalysts,

COz will undergo a similar reaction and 0z will become hydrogenated.

Although the effect of an oxygenic compound is temporary in the sense that
the catalyst regains its activity when the poison is removed from the gas stream,
a severe poisoning for an extended period of time may result in a permanent loss
of activity, probably due to recrystallization, see [1] and [11].

5.5 Early Studies

Studies of ammonia catalyst poisoning by oxygenic compounds date back to

1922 when Larson carried out experiments with 400-1000 ppm of H 20 at
pressures up to 100 atm [12]. Experiments were also carried out with CO, and
the equivalence of the various oxygenic compounds was recognized. Experi-
ments with up to 1600 ppm 0z in the synthesis gas at ambient pressure were
.~arried out, see [13]. The amount of 0z taken up by the catalyst was deter-
mined. The promoted catalysts were found to take up a larger amount of
oxygen. The thermodynamic stability of the surface oxide was evaluated by [14]
and found to be more stable by 12000 cal per gram atom of iron than a gram
atom of iron in magnetite. Poisoning experiments at high pressure were carried
out. The results were used to suggest that the amount of oxygen take up by a
catalyst varied with [pH zO/pH2J1/ 2. The validity of this result is questioned,
however, as the experiments were carried using a "pure" gas with about 100 ppm
of oxygenic impurities, see [2]. An interesting set of experiments were carried
using predetermined volumes of CO, CO 2, 0z and S02' see [15]. These
volumes of CO, COz, O 2 and S02 were injected, which resulted in a release of
ammonia in amounts exceeding that of thermodynamic equilibrium; the experi-
ment was conducted at 1 atm. The ammonia release was followed by an activity
loss, which in the case of SOz was irreversible. Injection of 80 cm 3 of CO
resulted in an almost complete poisoning of 10 cm 3 of catalyst. About 10 cm 3 of
NH3 was released during the poisoning. It was suggested that the release ofNH 3
was due to a replacement of chemisorbed N with oxygen. In this case it would
mean that N occupies around 1/8 of the available surface sites.
The iron catalyst was examined using chemisorption. Through this method
it was found that the active surface area was a much smaller fraction of the total
surface area that was determined by nitrogen condensation, see [16]. However,
 Poisoning of Ammonia Synthesis Catalysts 195

the conclusions may be severely affected by the limited ultrahigh vacuum

technology available in 1940.

5.6 Later Studies

High pressure studies carried out using O 2 , CO, CO 2 as poisoning agents [1]
demonstrated that 50 ppm Oz is just as poisonous as 100 ppm CO, and that a
6-day exposure to 100 ppm of CO at 450°C did not harm the catalyst.
Experiments were conducted using radioactive tracer methods with CO and
COz as poisons [17]. A slight amount of COz was found to be irreversibly
absorbed; the potassium promoter was thought to be responsible for that.
Studies were carried out with a number of promoted catalysts at temperatures
between 300 and 450°C, a space velocity of 5600, and a pressure of 100 kg/m 2
(9.8 MPa) [18]. At lower temperatures, some of the CO was not converted. At
higher temperatures, the poisoning was found to be reversible. The experiments
at lower temperatures were not continued long enough to determine if the
poisoning was reversible.
Changes in work function originating from the chemisorption of oxygenic
compounds and the HzS upon the catalyst were studied [5]. HzS was found to
cause a permanent increase in the work function, whereas the oxygenic com-
pounds caused a temporary increase. Experiments were conducted at ambient
pressure using a singly promoted catalyst with water as poisoning agent [19].
Between 1.2-32.3 mg of oxygen was found chemisorbed onto the catalyst which
has a total surface area of 250 m Z• However, the very low space velocities
employed and the duration of the experiment were not sufficient to reach steady
state. Experiments were conducted at ambient pressure [20-21] which sugges-
ted a kinetic expression, something which had been done earlier based on
theoretical consideration [22]. In [21] the rate expression was applied to data
obtained at elevated pressure. The expression being

k P k P~H3
+ N2 - -~
w = H2
[ P~H3 + C' P Ja
H20
P~2 P H2

where k+ is the rate constant for the synthesis reaction; k is the rate constant for
the decomposition reaction and (X is a constant, see also the chapter by B0gild
Hansen.
The expression assumes a displacement reaction
196 P. E. H0jlund Nielsen

The values of C depend on temperature [see 21]:

Cj Fe + Al 2 0 3 + K2 0 0.74 0.46 0.28 0.20 0.14

Cz Fe + Al z0 3 0.63 0.39 0.24 0.17 0.12

C 1 = 1.30 X 10- 6 exp 8930/T

C 2 = 1.15 x 1O- 6 exp8930/T
A satisfactory description of earlier results including [12] can be found in
[21]. It should be recalled, however, that experiments using a pure synthesis gas
could not be carried out at that time. Examination of poisoning reported in [23]
using H 20 vapour amounts between 1.8-95.8 ppm at 1 kg/cm 2 (98.1 kPa),
demonstrated that the activation energy for ammonia synthesis does not depend
on the amount of poison. He could, however, neither describe his results by
means of Smirnov's theory, nor by means of Ozak~ Taylor and Boudart's theory
[24]. It was found that oxygen reduces the rate of nitrogen adsorption [25]. By
doing studies of isotope exchange, the amount of N2 chemisorbed was reduced
only 1.8 times, whereas the rate of homo-exchange was reduced 17 times,
contrary to the theory adopted in Ref. [21].
Recent work has been carried out by P. E. H0jlund Nielsen [26], describing
the poisoning by multiplying the rate expression with a correction factor, 1-8,
-where 8 is given by

8= a + bT + cTln[H~~J
By varying the H2/N2 ratio, it was found that the expression
8= a + bT + c 1TlnXH20
where XH20 is the molar fraction of H 20, gave a better overall description [27].
This apparent independence of the hydrogen pressure was explained in Refs.
[28] and [29]. By using surface science data, kinetics valid at high pressure were
established. From these data it was established that during synthesis the surface
was covered by N-species; see also the old result in Ref. [15]. By assuming that a
chemisorbed 0 also influenced chemisorption upon neighbouring sites, an
explanation of the results of Ref. [15] could be established. Accordingly, Stoltze
and N0rskov [28, 29] simply added an equilibrium reaction to their set of
equations, see the chapter by B0gild Hansen upon kinetics, the equation being
H 20 (g) + 3*f:+2H - * + 0 - *
where * denotes a surface site.
By doing so, a satisfactory description of the results reported in Refs. [26, 27]
could be obtained.
 Poisoning of Ammonia Synthesis Catalysts 197

----::..::--..:::..,--
----,
1.0
'"'''. ,
"'-. ", SObar-400°C
O.B '\/-
.'( \SObar-4S00C
\. \\
Z.
:2: 0.6 \. \
\
tI
Cl \. \\
<II
:g> \ \
q; 0.4
a:
\. \\
\. \
\
\ \
0.2 \ \
\ \
\, \
0 ''''''-.. ' .... -
ru 100
ppm "0"

Fig. 5.1. Poisoning of ammonia synthesis catalyst

Recently reported data [30] confirm primarily the equivalence of H 2 0, CO,

O 2 and CO 2 , The poisoning was studied at 350-400 and 450°C, see Fig. 5.1. It
W,lS found [30] that the results could be described by adding an adsorption
factor for oxygen to the rate equation, a procedure in accordance with Refs [28]
and [20].
The results of the latest studies are a sort of complimentary situation, since
poisoning can be described either by multiplying the rate equation by a
correction term, [26, 27], or modifying the rate equation by including an
adsorption term for oxygen. At low temperatures the latter treatment would be
in disagreement with the results given in Ref. [23] since the adsorption of
oxygen is, in practice, irreversible under these conditions. However, for all
practical purposes, the two approaches should give an equally valid description
of reversible oxygen poisoning.

5.7 References

1. Nielsen A (1968) An Investigation on Promoted Catalysts for the Synthesis of Ammonia, Jui.
Gjellerup, Copenhagen, pp 126
2. H0jlund Nielsen PE, (1991) Catalytic Ammonia Synthesis, ed. by J. R. Jennings, Plenum
New York
3. Benard J, Oudar J, Barbouth N, Margot E, Berthier Y (1979) Surf Sci 88: L35
198 P. E. Hojlund Nielsen

4. Grabke HJ, Paulitschke W, Tauber G, Vielhaus H (1977) Surf Sci 63: 377
5. Enikeev E, Krylova AV (1962) Kinet Katal 3: 139
6. Brill R, Tauster S (1963) Ber Bunsen-Ges phys Chern 67: 390
7. Tauster S, PhD Thesis, Polytechnic Institute of Brooklyn, June 1964, available university
microfilms, 64-10, 728
8. Brill R, Schaefer H, Zimmerman G (1968) Ber Bunsen-Ges Phys Chern 72: 1218
9. Rostrup-Nielsen JR, Hojlund Nielsen PE (1985) In: Oudar J, Wise H (eds) Deactivation and
Poisoning of Catalyst Dekker, New York, p 259
10. Marakhovets et al. (1972) Izv Vyssh Ucheb Zaved Khim Tekhnol p 735
11. Catalyst Handbook, ed. M. V. Twigg, Wolfe London 1989, p 407
12. Larson AT, Tour RS (1922) Chern Met Eng 26: 647
13. Almquist JA, Black CA (1926) J Am Chern Soc 48: 2814
13. Almquist JA (1926) J Am Chern Soc 48: 2820
14. Emmett PH and Brunauer S (1930) J Am Chern Soc 52: 2682
15. Ussatschew PW, Tarakanowa WJ, Komarov WA (1934) Z Elektrochem 40: 647
16. Brunauer S, Emmett PH (1940) J Am Chern Soc 62: 1732
17. Bokhoven C (1954) Proceedings of the Second Radiosotope Conference, Oxford Butterworths,
London, pp 53
18. Uchida H, Todo N (1951) Report of the Tokyo Industrial Research Institute Laboratory 46: 213
19. Royen P, Langhans GH (1962) Z Anorg Allg Chern 315: 1
20. Smirnov lA, Morozov NM, Temkin MI (1965) Kinet Katal 6: 351
21. Smirnov IA (1966) Kinet Katal 7: 107
22. Kiperman SL (1954) Zhur Fiz Khim 28: 389
23. Brill R. Hensel J, Schaefer H (1969) Ber Bunsen-Ges Phys Chern 73: 1003
24. Ozaki A, Taylor H, Boudart M (1960) Proc Roy Soc (London) Ser A258, 47
25. Boreskova EG, Kuchaev VL, Temkin MI (1984) Kinet Katal 25: 112
26. Hojlund Nielsen PE; The Poisoning of Ammonia Synthesis Catalysts by Oxygenic Compounds,
lecture given at the ACS Meeting, Washington, D.C. Sept. 1983.
27. Andersen S, Haldor Topsoe AjS, Private Communication
28. Stoltze P (1987) Phys Scr 36: 824
29. Stoltze P, Norskov JK (1987) J Vac Sci Technol A5, 581
30. Kirkerod T, Skaugset P (1991) Abstracts of IV Nordic Symposium on Catalysis, Trondheim
Chapter 6
Ammonia Production Processes

Ib Dybkjaer
Haldor TopS0e A/S Lyngby, Denmark

Contents

6.1 Introduction. .202

6.2 Historical.. · 202

6.3 Synthesis Gas Production. .204

6.3.1 Steam Reforming .. 205
6.3.2 Partial Oxidation and Gasification . 209
6.3.2.1 Partial Oxidation . . . 209
6.3.2.2 Gasification....... 210
6.3.3 Carbon Monoxide Conversion. · 212
6.3.3.1 Shift Conversion. . . 212
6.3.3.2 Selective Oxydation . · 213
6.3.4 Acid Gas Removal. . . . . . . · 213
6.3.5 Final Purification . . . . . . . · 215
6.3.5.1 Copper Liquor Wash. 215
6.3.5.2 Cryogenic Purification · 216
6.3.5.3 Pressure Swing Absorption. · 217
6.3.5.4 Methanation......... · 217
6.3.5.5 Methanolation........ · 218
6.3.6 Synthesis Gas from Unconventional Sources · 218
6.3.7 Synthesis Gas Compression . . . . . . . . . . · 219

6.4 Synthesis of Ammonia. . . . . . . . . . . . . . . . .. 221

6.4.1 The Ammonia Synthesis Loop. . . . . . . .. 221
6.4.1.1 Types of Ammonia Synthesis Loops 222
6.4.1.2 Mathematical Models . . . . . . 225
6.4.1.3 Effect of Individual Parameters 226
6.4.2 Design of Ammonia Synthesis Reactors 229
6.4.2.1 General.............. 229

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
200 Ib Dybkjaer

6.4.2.2 Reactor Calculations . . . . . . . . . .230

6.4.2.3 Principal Reactor Types . . . . . . . · 232
6.4.3 Commercial Ammonia Synthesis Converters. · 235
6.4.3.1 Converters with Internal Cooling in Catalyst Beds. · 235
6.4.3.1.1 Countercurrent Flow in Cooling Tubes . · 235
6.4.3.1.2 Co current Flow in Cooling Tubes · 237
6.4.3.1.3 Cross Flow . . . . . . . . . . . . . . . . . · 237
6.4.3.2 Converters with Quench Cooling . . . . . . . . . . · 238
6.4.3.2.1 Axial Flow, Quench Cooling Between
Separate Catalyst Beds . . . . . . . ... · 238
6.4.3.2.2 Axial Flow, Quench Cooling by
Injection of Gas into the Catalyst Beds .240
6.4.3.2.3 Radial Flow or Axial/Radial Flow . · 240
6.4.3.3 Converters with Indirect Cooling . . . . . . . . . . · 243
6.4.3.3.1 Axial Flow . . . . . . . . . . . . . . . . · 243
6.4.3.3.2 Radial Flow or Axial/Radial Flow . .247
6.4.3.4 Revamp of Ammonia Synthesis Converters . . .247
6.4.4 Waste Heat Recovery in Ammonia Synthesis Loops .. · 251
6.4.5 Construction Materials in Ammonia Synthesis Loops. · 252
6.4.6 Product Recovery from Ammonia Synthesis Loops · 253
6.4.7 Purge Gas Recovery. . . . . . . . . . . . . · 254
6.4.7.1 Cryogenic Separation . . . . . . . . . . . . . · 255
6.4.7.2 Separation Through Membranes . . . . . . · 256
6.4.7.3 Separation by Pressure Swing Adsorption. · 256
6.4.7.4 Use of Metal Hydrides ... 257

6.5 Complete Ammonia Production Processes. 257

6.5.1 General................ 257
6.5.2 Energy Balances . . . . . . . . . . . . 258
6.5.2.1 Thermodynamic Analysis. . 258
6.5.2.2 Comparison of Energy Consumption for
Different Cases . . . . . . . . . . . . . . . . 261
6.5.2.2.1 Theoretical Minimum Energy
Consumption. . . . . . . . . . . 261
6.5.2.2.2 Influence of External Conditions 262
6.5.2.3 Steam and Power System. . . . . . . . 264
6.5.3 Processes Based on Steam Reforming of Light
Hydrocarbons . . . . . . . . . . . . . . . . . . . 269
6.5.3.1 Theoretical Considerations. . . . . . . 269
6.5.3.1.1 Stoichiometric Production of
Ammonia from Methane. . . 269
6.5.3.1.2 Energy Analysis of a Low Energy
Ammonia Process . . 269
6.5.3.2 Process Schemes. . . . . . . . . 271
6.5.3.2.1 Early Developments. 271
 Ammonia Production Processes 201

6.5.3.2.2 Recent Developments. . . . . 276

6.5.3.2.3 Commercial Processes. . . . . 280
6.5.3.2.4 Revamp of Existing Plants. . 295
6.5.4 Processes Based on Partial Oxidation of Hydrocarbons
or Gasification of Solid Feedstocks. . . . . . . . . . . .. . 298
6.5.5 Integrated Production of Ammonia and Other Products . 300
6.5.5.1 Ammonia and Urea. . . . . . . . 300
6.5.5.2 Ammonia and Other Products. . 302

6.6 Storage of Ammonia . . . . . · 303

6.6.1 Pressurized Storage . · 305
6.6.2 Atmospheric Storage. · 306

6.7 References. . . . . . . . . . . · 308

202 Ib Dybkjaer

6.1 Introduction

The present chapter describes process technology for the production of am-
monia. The first part gives a short review of early developments in ammonia
technology. The second part contains a brief description of processes used in the
production of ammonia synthesis gas, i.e. a mixture of hydrogen and nitrogen
with or without minor amounts of impurities such as methane, argon, etc. A
third part describes the conversion of synthesis gas to ammonia. This part
includes discussion of the design of ammonia synthesis converters. A fourth part
describes integrated processes for production of ammonia from primary raw
materials such as hydrocarbons and coal. A final part deals with storage of
ammonia.
Several publications are available which give an integrated treatment of
important aspects of ammonia production. Some examples are [1-9, 876, 949].
Relevant patents are reviewed in [10, 11]. Historical aspects are treated
especially in [12-17, 877]. Reviews of the technological status at various times
are given in [1-9J and in [18-42,949]. Catalytic reactions and catalysts relevant
for ammonia production are treated in [43-46].

6.2 Historical

The main purpose of the present chapter is to describe modern technology for
ammonia production. For the sake of completeness a brief description of the
early development of ammonia synthesis technology will be given in the
following. More details can be found in the relevant literature.
The history of modern ammonia production started in Germany just after
1900, when Fritz Haber and his assistants developed the process concept which
is the basis for all ammonia production even today. Haber patented his work in
two famous patents, the "circulation patent" [47J and the "high pressure patent"
[48]. The claims of the patents read:
"Circulation patent": "Verfahren zur synthetischen Darstellung von Am-
moniak aus den Elementen, wobei ein geeignetes Gemenge von Stickstoff und
Wasserstoff kontinuerlich der Ammoniakbildung mittels erhitzter Katalysa-
toren und nachofolgender Ammoniakentziehung unterworfen wird, dadurch
gekennzeichnet, daB hierbei unter dauerndem Druck gearbeitet und dafiir
gesorgt wird, daB die Warme der ammoniakhaltigen Reaktionsgase auf das von
neuem der Reaktion zu unterwerfende ammoniakfreie Gasgemisch iibertragen
wird".
"High Pressure patent": "Verfahren zur Darstellung von Ammoniak aus den
Elementen durch Katalyse unter Druck bei erh6hter Temperatur, dadurch
 Ammonia Production Processes 203

gekennzeichnet, daB die Vereinigung unter sehr hohen Drucken von etwa 100
Atmospharen, zweckmassig aber von 150 bis 250 Atm. und mehr vorgenommen
wird."

The German company BASF had been following Haber's work through
their two employees, Carl Bosch and Alvin Mittasch and eventually bought
Haber's patents. Development work was continued on catalyst, process, and
equipment design, and in 1913 the first commercial plant with a capacity of 30
tons/day was commissioned at Oppau/Ludwigshafen near Mannheim in Ger-
many.
By 1916 the capacity had been increased to 250 tons/day. In 1917 a second
plant was started at Leuna near Leipzig. At the end of the First World War this
plant produced 240000 tons/year [14]. In 1937, the annual world capacity was
755000 tons/year, of which 72% was still concentrated in Oppau and Leuna
[16].
Soon after the First World War, others started development work, partly on
the basis of the German pioneering effort. A plant was started at Terni, Italy in
1920 based on technology developed by Luigi Casale. In France, a plant was
constructed in 1921 at Grand Paroisse near Montereau, based on developments
by M. G. Claude. Both the Casale and the Claude technologies operated at
extremely high pressures. In further early developments in Europe, technology
developed by Giacomo Fauser was adopted by the Italian company Monteca-
tini, and Friederich Uhde GmbH constructed a plant to operate on coke oven
gas from the Mont Cenis company. This plant used a special process later
known as the Mont Cenis process. Its features were a very low operating
temperature and pressure and a special catalyst based on iron cyanide.
Meanwhile, an independent development took place in the USA. As early as
1918 a plant called "US Nitrate Plant No.1" was constructed in Muscle Shoals,
Alabama. This plant was, however, not very successful, but the American
development continued, and during the 1920s several successful plants were
constructed, some based on American, some on European technology. In 1928,
the Nitrogen Engineering Corporation (NEC) was commissioned to construct a
plant in Europe [49], thereby entering the worldwide competition.
During the 1930s and during the Second World War production capacity
was increasing rapidly using now well-established technologies. In 1945 about
125 plants existed with a total capacity of 4.5 million tons [49]. The most
important processes were: Haber-Bosch, Casale, Claude, Fauser, NEC, and
Mont Cenis. For descriptions of the technical features of these processes, see
[50].
Just before the second world war it was a general belief [18] that ammonia
synthesis technology was mature, and that no significant further developments
could be expected. This was to some extent true, since most of the features which
characterize modern synthesis technology - including catalyst type and most
major converter designs - were already well proven in industrial operation.
However, since that time, there has been a tremendous development in the scale
204 Ib Dybkjaer

of operation, and at the same time, the main type of feedstock has shifted from
coke to natural gas, naphtha or fuel oil. New technology for the production of
synthesis gas has been developed, and integrated, efficient processes for large
scale production of ammonia from the different feedstocks have emerged.

6.3 Synthesis Gas Prodnction

The ultimate raw material for ammonia production is synthesis gas, a mixture of
hydrogen and nitrogen with or without minor quantities of impurities such as
methane, argon, etc. Normally, the concept "Production of Ammonia" is,
however, understood to cover the complete transformation of primary raw
materials into ammonia [8]. The nitrogen part of the synthesis gas is always in
some way derived from air, either in a separate process step, most often by
cryogenic air separation, or in integrated processes, where air is used as a
reactant in such a way that oxygen is consumed in the gas preparation process,
while nitrogen remains as constituent in the synthesis gas (see below). When
nitrogen is produced in a separate process by air separation, the oxygen part of
the air is most often used in the process for production of the hydrogen part of
the synthesis gas.
The most important primary raw materials for the hydrogen part of the
synthesis gas are natural gas and other light hydrocarbons. Other raw materials
include heavy hydrocarbons, solid materials such as lignite and coal, and
hydrogen recovered from off-gases from other processes. Production of syn-
thesis gas can be subdivided into the following process steps (see Fig. 6.1).
-Gas Preparation
From natural gas and light hydrocarbons by steam reforming, catalytic
autothermal reforming, or partial oxidation.
From heavy oil or coal by partial oxidation or gasification.
From water by electrolysis.
From hydrogen-rich off-gases by separation of hydrogen.
-Gas Purification
In most cases, the gas purification train contains the following process steps:
- Carbon monoxide conversion
- Carbon dioxide removal
- Final purification
Final purification can be by:
- Methanation or Methanolation
- Cryogenic purification ("Nitrogen wash")
- Copper liquor wash
 Ammonia Production Processes 205

As an alternative to these process concepts, the gas purification including

carbon dioxide removal can be done by Pressure Swing Absorption (PSA).
In the following paragraphs, each of the individual process steps is briefly
discussed. Integrated process schemes are discussed in Sect 6.5.

6.3.1 Steam Reforming

Production of ammonia synthesis gas from light hydrocarbons is normally

initiated by a two-step process consisting of so-called primary and secondary
reforming. The feedstock, which may range from natural gas to heavy naphtha,
is mixed with steam in an amount (normally expressed by the steam to carbon
ratio) determined by feedstock properties and operating conditions. A typical
range is 2.5-4 moles of steam per mole of carbon in the feed. The steam-
hydrocarbon mixture is first passed through the primary (tubular) reformer,
which is heated by burning fuel. The feed is converted to a mixture of carbon
oxides, hydrogen and unconverted methane. This mixture is further reacted with
air in the secondary reformer, which is an adiabatic reactor. Typical outlet
temperatures from the two steps are 750-820 DC from the primary reformer and
950-1025 DC from the secondary reformer. The pressure in modern plants
typically is about 25-35 bar. Before reforming, the hydrocarbon feed must be
purified to remove compounds such as sulfur, which are poisonous to the
reforming catalyst and downstream catalysts. The desulfurization takes place by
adsorption of sulfur-containing compounds on active carbon or on molecular
sieves or by catalytic hydrogenation of the organic sulfur compounds and
subsequent adsorption of hydrogen sulfide on zinc oxide. References on de-
sulfurization of reformer feedstock are [51-61]. Bulk removal of hydrogen
sulfide by absorption in amine solutions and other solutions is discussed in [62,
63]. The effect of a possible mercury content in ammonia plant feedstocks is
reviewed in [878]. Removal of mercury and arsenic is discussed in [879].
In the primary reforming reaction, hydrocarbon feed is converted with steam
according to the following reactions:
CnHm + nH 20 -+ nCO + (m/2 + n) H2 ( - fl Hg 98 < 0) (1)
CH 4 + H20~CO + 3 H2 ( - fl Hg 98 = - 206 kJ/mol) (2)
CO + H20~C02 + H2 ( - fl Hg 98 = 4 kJ/mol) (3)
Higher hydrocarbons are decomposed irreversibly and completely with
steam by reaction (1). The final gas composition is governed by the two
reversible reactions, the methanation reaction (2) and the carbon monoxide
conversion or shift reaction (3).
The conversion of methane is favoured by high temperature, low pressure,
and a high steam content. The overall conversion is strongly endothermic, and
heat must be added to the process. Therefore, the reaction is carried out in
206 Ib Dybkjaer

Natural Gas
LPG, Naphtha

Enriched
Air Air

Autothermal
Reforming

,--- ---,
I Cryogenic I
I Purifier, I

L~r_J

Fig. 6.1. Possible process steps in ammonia synthesis gas production

 Ammonia Production Processes 207

Natural Gas Hydrc

LPG, Naphtha Enriched e.g. fr,
Fuel Oil, Coal Air Electr

Fig. 6.1 (Continued)

reactor tubes which are heated from the outside by burning fuel. The reactor
tubes are made of high alloy steel and filled with a nickel-based catalyst. A
comprehensive treatise on steam reforming reactions and steam reforming
catalysts is given in [64, 65]. General literature on reforming: [66-72, 880, 881];
208 Ib Dybkjaer

steam reforming of naphtha (73-81). Catalyst formulations are discussed in [43,

44, 64, 65J and [82-85, 882]. Special problems in primary reformers are related
to the risk of carbon formation [65, 86, 87J and catalyst poisoning, especially by
sulfur [88, 883, 961]. Design of reformer furnaces and selection of tube material
are discussed in [89-99, 884, 885, 962]. Creep rupture of reformer tubes is
discussed in [100-104, 886]. Higher hydrocarbons (naphtha) may be converted
in an adiabatic pre-reforming step [105-112, 887, 888]. Poisoning problems in
pre-reformers are treated in [113]. Actual experience from installation of a pre-
reformer in an ammonia plant is reported in [963].
Special reformer designs have been suggested in which the heat for the
reforming reaction is supplied by heat exchange with a hot gas [114-117, 889,
890]. Application of such reformers in the production of ammonia synthesis gas
is suggested in [118-124, 891]. A design where the primary reformer is split into
a number of adiabatic vessels with reheat between the vessels is proposed in
[125, 126].
In secondary reforming, the exit gas from the primary tubular reformer
reacts adiabatically with air over a nickel-based catalyst. In this way, residual
methane in the process gas is reduced to a low level and at the same time the
nitrogen part of the ammonia synthesis gas is added. In most cases, the amount
of air is adjusted in such a way that the stoichiometric ratio of hydrogen to
nitrogen equal to 3.0 is obtained in the synthesis gas after carbon monoxide
conversion and gas purification. However, there are also process schemes in
which non-stoichiometric amounts of air are used; this is discussed in Sect. 6.5.
Secondary reforming is carried out in an adiabatic, refractory-lined vessel.
Gas from the primary reformer typically at about 800°C and containing
8-12 vol% (dry) residual methane is mixed in a burner with preheated air. The
partially reacted mixture passes through a bed of catalyst, where the reactions
are completed. The gas leaves the reactor at about 1000°C with a residual
methane content of typically 0.3-1.0 vol% (dry).
The design of secondary reformers is discussed in [90, 92, 127]. Special
designs where primary and secondary reforming or primary reforming and non-
catalytic partial oxidation are combined in one unit are suggested in [128-130,
964].
As an alternative to the combination of primary and secondary reforming,
processes have been developed where all the conversion takes place in equip-
ment similar to secondary reformer, but where the oxidizing agent is not air, but
oxygen or oxygen-enriched air. Such so-called "catalytic partial oxidation", or
"autothermal reforming" processes are described in [131-135]. A process where
process air is preheated in a regenerative system to such high temperatures that
stoichiometric ammonia synthesis gas can be produced from natural gas and air
is described in [892]. An experimental study of catalytic partial oxidation is
reported in [136]. Also non-catalytic partial oxidation oflight hydrocarbons has
been used for production of ammonia synthesis gas [137, 138].
Steam reforming was first used for the production of ammonia synthesis gas in
the early 1930s [71, 72]. During the Second World War the process became
 Ammonia Production Processes 209

dominant in the US ammonia industry, when natural gas became the preferred
feedstock. In early plants using steam reforming, the operating pressure was low
[71, 139-141]. As developments in metallurgy allowed, the pressure was
gradually increased up to the present levels of about 35 bar [142-145].

6.3.2 Partial Oxidation and Gasification

Non-catalytic partial oxidation or gasification of hydrocarbons or solid feed-

stocks are alternatives to steam reforming and auto thermal reforming for
production of synthesis gas.
In contrast to the reforming processes, which accept only light hydrocarbons
(up to and including naphtha) as feedstock, the partial oxidation and gasific-
ation processes are very flexible with respect to feedstock. These processes have
in fact been used on all types of raw materials including natural gas, liquid
hydrocarbons including naphtha and heavy oil, and coal and lignite. It is normal
to refer to processes treating gaseous or liquid hydrocarbons as "partial
oxidation processes", whereas processing of solid materials is called "gasification
processes".

6.3.2.1 Partial Oxidation

In partial oxidation processes, a hydrocarbon feedstock is reacted with oxygen
or an oxygen-containing gas at high temperature and normally high pressure to
produce a gas containing hydrogen, carbon oxides, some steam, and traces of
methane and of impurities such as hydrogen sulfide. The process is not sensitive
to sulfur as is the reforming process, thus it is not normal practice to desulfurize
the feedstock.
Reviews ofthe technology and of available processes are given in [146-150].
The reactions taking place may be represented by the following simplified
equations:
CnHm + nH 20 --+ nCO + (m/2 + n)H2 ( - AH~98 < 0) (1)
CH 4+ H20~CO + 3 H2 ( - AH~98 = 206 kJ/mol) (2)
CO + H20~C02 + H2 ( - AH~98 = 41 kJ/mol) (3)
CnHm + n/2 O 2 --+ nCO + m/2 H2 ( - AH~98 > 0) (4)
CnHm --+ nC + m/2H 2( - AH~98) (5)
The hydrocarbon feedstock reacts with oxygen and steam according to reac-
tions (1) and (4), and the methanation reaction (2) and the shift reaction (3)
proceed to near equilibrium. It is important to avoid excessive soot formation
shown by equation (5).
210 Ib Dybkjaer

In general, it has been found that soot formation can be reasonably

suppressed - but never completely eliminated even with light feedstocks such as
natural gas - if the oxygen/carbon atomic ratio is kept above 1, and if the
reaction temperature is above about 1300°C [146]. In early plants the reaction
pressure was low, but in more recent designs, pressures up to 80 bar have been
used.
Two partial oxidation processes dominate the market completely, the
Texaco process and the Shell process. Reaction conditions are very similar in the
two cases, and the main difference is in the method of gas cooling and in the
technology used for soot removal. In both cases, preheated feedstock, oxygen,
and a small amount of steam are reacted in a refractory lined vessel at
temperatures up to 1500 °C and at a pressure which in the Shell process typically
is 60 bar and in the Texaco process up to 80 bar. The raw synthesis gas -
essentially a 1: 1 mixture of hydrogen and carbon monoxide (when the feedstock
is heavy oil) with small amounts of carbon dioxide and steam and traces of
methane, hydrogen sulfide, and other impurities - is cooled and treated for soot
removal. The Texaco process uses quench cooling or combined cooling in a
waste heat boiler and by quench, and the soot removal system consists of a
Venturi scrubber followed by washing with water. The soot is recovered as a
slurry in water, which is recycled to the reactor. In the Shell process the gas is
cooled in a waste heat boiler, and the soot is removed by washing with light oil.
The soot is recovered in the form of carbon pellets. The treated gas from the
Texaco process is at about 200°C and saturated with water, whereas the treated
gas from the Shell process is cold and essentially water free. This difference is
important for the layout of downstream processes. The two processes are further
described in (151-158), (893). The economics of steam reforming and partial
oxidation are compared in (159-161). Experience from construction and opera-
tion of plants based on partial oxidation of heavy fuel oil is reported in
(162-169), (893).

6.3.2.2 Gasification
In certain situations coal and lignite are attractive as feedstocks for the
production of ammonia, and plants using these raw materials have been
constructed in many parts of the world. It should also be noted that coke was
the most important feed- stock in the early days of the ammonia industry in
plants using the water gas process (see Sect. 6.3.6). Natural gas or heavier
hydrocarbons replaced coke during and after the Second World War except in
special circumstances. During the oil crises of the 1970s, coal appeared to be the
future feedstock for ammonia production, and many studies and much develop-
ment work were carried out on processes for the production of ammonia from
coal. However, later developments in oil prices made coal less competitive as a
feedstock, and the expected large scale change in the industry from natural gas
and heavier hydrocarbons to coal never did materialize.
 Ammonia Production Processes 211

Technology for the gasification of coal or lignite is reviewed in (170-183).

The classical processes are two low pressure processes, the Koppers-Totzek
process and the Winkler process, and two high pressure processes, the Lurgi
process and the Texaco process. All these processes are used in industrial plants.
A significant number of other processes have been developed to the stage where
they have been proven in large demonstration units or in plants for other
purposes than the production of ammonia synthesis gas.
The Koppers-Torzek process [184-188] is used in several ammonia plants
today. The process operates at low pressure, and the reaction between pulver-
ized coal and oxygen mixed with steam is carried out in an entrained bed with a
residence time of less than 1 second and at a temperature of 1500-1600°C.
Ammonia plants based on this technology are described in [189-193]. A high
pressure version of the process has been developed, initially in cooperation
between Shell and Koppers [194-196]. Later, this process was further de-
veloped and demonstrated separately by Shell [197, 198,894] and in a coopera-
tion between Krupp and Koppers as the Prenflo process [199-201]. The
classical Winkler process [202] is a fluid bed process operating at low pressure
and at somewhat lower temperature than the Koppers-Totzek process, around
1000°C. A high pressure Winkler process for gasification of lignite has been
developed and commercialized by Rheinbraun [203, 204]. The Lurgi gasifier
[205, 206] is a moving bed process. It operates at about 30 bar. Coal in the form
of 5-40 mm particles is fed through a lock hopper to the top of the gasifier.
Oxygen and steam is fed to the bottom, and the gasification takes place in the
lower part of the gasifier. The hot gases pass countercurrently through the coal
which moves slowly down through the gasifier. The hot gases de gasify the coal
and are at the same time cooled to an outlet temperature of about 600°C. The
gas contains impurities including about 10% methane and significant amounts
of naphtha, tar etc. [207] This high content of impurities is a major dis-
advantage for the Lurgi process in the production of ammonia synthesis gas.
The classical Lurgi process operates as a dry ash gasifier. A further development,
the British Gas-Lurgi process operates as a slagging gasifier (208).
All the above-mentioned processes use a dry coal feed. In contrast, the feed
to the Texaco process [209-212] is a slurry of coal in water. The process is a
further development of the Texaco process for partial oxidation of hydrocar-
bons. The process is an entrained bed process operating at up to 40 bar and at a
temperature of about 1400°C. The raw synthesis gas may be cooled either by
direct quench or by steam production in a waste heat boiler followed by quench
cooling. A large, modern ammonia plant based on gasification of coal by the
Texaco process is described in [213, 214]. A widely-published commercial scale
demonstration project on the retrofit of an existing natural gas-based ammonia
plant to coal feed also used the Texaco process [215-219]. The Dow process is
also based on slurry feed. This process has, however, until now been used only
for gasification in Integrated Gasification Combined Cycle (lGCC) power
plants. A review of gasification processes especially for such plants is given in
[220].
212 Ib Dybkjaer

6.3.3 Carbon Monoxide Conversion

The product gas from steam reforming, catalytic partial oxidation, partial
oxidation, or gasification processes contains in all cases significant amounts of
carbon monoxide and carbon dioxide. Since the feed gas for the ammonia
synthesis loop must be completely free of these compounds, they must be
removed in the gas preparation part of the plant. Carbon monoxide is removed
in a conversion step by the so-called "water-gas shift" reaction or just the "shift"
reaction, in some cases followed by further conversion by selective oxidation.

6.3.3.1 Shift Conversion

In the shift conversion process, carbon monoxide is converted to carbon dioxide
with the formation of additional hydrogen by the following reaction:
CO + H 20 -+ CO 2 + H2 ( - AH~98 = 41 kJmol) (3)
Several catalysts and processes are available for this reaction. In the classical
process, the catalyst is magnetite, Fe 3 0 4 , promoted with chromia and in some
cases with potassium or other promoters [221-223]. This catalyst requires
temperatures above about 350 DC, and the exit carbon monoxide concentration
is normally about 3 vol% in the dry gas. Conversion was in some cases
improved by installing a two stage unit with carbon dioxide removal between
two high temperature shift reactors [224]. However, this system was rather
expensive and did not gain wide acceptance.
In the early 1960s, copper-based catalysts became available for a second
conversion step at temperatures about 200 DC, whereby the carbon monoxide
concentration can be reduced to 0.1-0.3 vol% in the dry product gas [224-227].
This represented a major improvement in process economics, and today virtua-
lly all ammonia plants use the two step carbon monoxide conversion, or shift
conversion, consisting of a high temperature and a low temperature step.
Comprehensive discussions of shift conversion may be found in [44, 228]. Shift
catalysts are described in [229-234, 896-898]. The formation of methanol as a
by-product in low temperature shift reactors is discussed in [899]. The low
temperature shift catalysts are sensitive to poisoning, especially by chlorine and
sulfur. Poisoning problems are discussed in [235-240]. One way of protecting
the catalyst from poisoning is to install a guard vessel upstream the main reactor
[241]. The guard vessel absorbs the poisons, and it is designed for easy catalyst
change. In this way the life of the catalyst in the main reactor is significantly
extended.
In some cases gas with a high sulfur content coming from partial oxidation
or gasification units is passed directly to shift conversion. The high sulfur
content can be tolerated by magnetic type catalysts, although the activity is
reduced [221]. But more efficient sulfur tolerant catalysts based on cobalt-
molybdenum sulfide have become available [169, 242-245]. With such catalysts
 Ammonia Production Processes 213

low temperature conversion can also be obtained in sulfur containing gases

[244,245].
In modern ammonia plants based on steam reforming, the tendency is to
reduce the steam/carbon ratio in the reformer and thereby the steam content in
the feed gas to the shift converter. At low steam contents, problems may occur in
units using the classical magnetite catalyst because this catalyst is partly
converted into iron carbide which promotes by-product formation [246-248,
896]. The by-products are hydrocarbons, mainly methane and lower olefins, and
oxygen-containing organic compounds such as alcohols and acids. They do not
only represent a loss of production, but also have a detrimental effect on the low
temperature shift catalyst [249]. It has been found that adding a small amount
of copper to the iron-based high temperature shift catalyst overcomes this
problem, at least in some situations [250, 251]. At more extreme conditions,
copper-based medium or high temperature catalysts without carbide-forming
elements must be used [246, 251].

6.3.3.2 Selective Oxidation

In order to further reduce the carbon monoxide concentration in the synthesis
gas, a selective oxidation unit has in some cases been installed after the low
temperature shift unit. After the gas has been cooled and condensate removed, a
small amount of air is introduced, and the gas is passed to a reactor containing a
precious metal catalyst. Carbon monoxide reacts with oxygen, and excess
oxygen is removed by reaction with hydrogen. The process is known as the
"Selectoxo" process [252-254]. Use of the process in a revamp project is
described in [255].

6.3.4 Acid Gas Removal

In plants based on steam reforming of light hydrocarbons, the product gas from
the shift conversion contains about 18 vol% (dry basis) carbon dioxide. Gas
originating from partial oxidation of heavy hydrocarbons or from gasification of
solid feedstocks contains even higher concentrations of carbon dioxide, and in
addition hydrogen sulfide may also be present in significant quantities depend-
ing on the sulfur content in the feed and on the type of shift conversion
technology employed. In plants using standard shift systems, sulfur is most often
removed from the raw gas upstream of the shift section in selective sulfur
removal units, whereas the sulfur is left in the gas in plants that use special sulfur
tolerant shift catalysts. In every case all sulfur compounds and all carbon oxides
must be removed from the gas before it can be used for ammonia synthesis.
Essentially all sulfur compounds (if present) and the bulk of the carbon dioxide
are removed in so-called acid gas removal units or carbon dioxide removal
units, which in all cases feature absorption of the acid gases in a suitable solvent.
214 Ib Dybkjaer

The absorption, which may be physical or chemical in nature, takes place at

synthesis gas pressure in an absorption vessel with trays or packing, and the rich
solution is regenerated in another vessel at reduced pressure by stripping and/or
heating (reboiling). Depending on the properties of the solvent and the operating
conditions, part of the dissolved carbon dioxide may be released from the
solvent by pressure release (flashing).
The desired carbon dioxide concentration in the treated gas depends on the
downstream process used for removal of the remaining part (see Sect 6.3.6). If
cryogenic gas purification is used, very low concentrations are required (most
often the last traces are removed by adsorption on molecular sieves). If
methanation is used for gas purification, levels of 500-1000 vol ppm carbon
dioxide are typical. In old-fashioned plants using copper liquor wash for carbon
monoxide removal, carbon dioxide concentration out of the absorption vessel
could typically be up to 0.5 vol%.
A large number of acid gas removal processes have been developed; some are
exclusively, or at least predominantly, used for natural gas sweeting, i.e. removal
of hydrogen sulfide and possibly carbon dioxide from natural gas. Surveys of
available acid gas removal processes may be found in [256-264, 953]. A
theoretical review is given [265], and the choice between various processes is
discussed in [266-272].
The solvents most used in carbon dioxide removal from ammonia synthesis
gas can be characterized according to the nature of the absorption process.
Chemical absorption, i.e. processes where the carbon dioxide reacts with the
solvent by a chemical reaction which is reversed in the solvent regeneration
stage, is most often based on the use of alkanolamines, mainly MEA (mono-
ethanolamine) [273], or hot solutions of potassium carbonate [274] as solvents.
A series of processes based on the use of MEA with an added corrosion
inhibitor called Amine Guard were developed by Union Carbide Corp. and
have been extensively used under the name Amine Guard II, III, and IV. The
heat requirement for solvent regeneration has been gradually decreased in this
series of processes through optimization of process layout and operating
conditions [275-279]. The effect of the corrosion inhibitor is to allow an
increase of the MEA concentration in the solvent, thereby increasing the
capacity for carbon dioxide and reducing the recirculation rate and thus the
energy consumption in the process.
Several processes are based on the use of a hot solution of potassium
carbonate as a chemical solvent. The processes differ mainly in the type of
activator used to increase the reaction rate between carbon dioxide and the
solvent. The Carbosolvant process [280, 281] uses Sulfosolvan B; the Giammar-
co-Vetrocoke process [282-284, 950] used initially arsenic oxide, in a later
development glycine as activator, and latest a dual activator system. The
Catacarb process [285] uses an amine borate activator, whereas the Benfield
process [286-289] and the Carsol process [290, 291] use enthanolamines as
activators. All of these processes have been improved through the years to
minimize energy consumption and to reduce the risk of corrosion. Some case
 Ammonia Production Processes 215

stories describing operational problems in hot potassium carbonate systems are

reviewed in [901]. New activators have also been developed which may improve
the performance of existing installations [902, 903, 950].
One of the first - if not the first - physical solvent to be used for carbon
dioxide removal was water [292]. Other physical solvents are sulfolane (tetrahy-
drothiophene 1.1 dioxide) + diisopropanol in the Sulfinol process [293]; meth-
anol in the Rectisol process [294-297]; polypropylene carbonate in the Fluor
process [298-301]; methylpyrrolidone in the Purisol process [295, 302, 303];
polyethyleneglycol dimethylether in the Selexol process [304-310, 904]; and
polyethyleneglycol methylisopropyl ether in the Sepasolv MPE process [311].
In the physical solvent processes no heat is required for solvent regeneration,
which is done by pressure release and stripping with air or inert gas. Because of
the physical nature of the solvents, a low temperature is advantageous. The
Rectisol process is operated as low as - 40°C, often in combination with a
liquid nitrogen wash for final purification in plants based on partial oxidation or
gasification [296].
Some processes combine the use of chemical and physical solvents. An
example is a dual solvent process based on the use of monoethanolamine and
triethanolamine (MEA/TEA) [312-314]. In other cases the process is based on
the use of a solvent which exhibits at the same time some of the properties
attributed to both physical and chemical solvents. This is notably true for
activated methyldiethanolamine, which is used in the MDEA process developed
by BASF [311, 314-320, 905, 906]. This solvent, operating at an intermediate
temperature, releases a significant part of the dissolved carbon dioxide by
flashing, thus reducing the energy required for reboiling and thereby the overall
energy consumption.

6.3.5 Final Purification

After bulk removal of carbon monoxide in the shift conversion section and of
carbon dioxide in the carbon dioxide removal section, the synthesis gas still
contains typically 0.2-0.5 vol % carbon monoxide and 0.01-0.2 vol % carbon
dioxide. These compounds must, together with any water present, be removed
quantitatively, i.e. to low ppm levels, before the gas can be admitted to the
synthesis converter, because all oxygen containing compounds are poisons to
the ammonia synthesis catalysts [238]. The most important technologies for this
final purification are discussed in the following.

6.3.5.1 Copper Liquor Wash

The oldest process for final purification of ammonia synthesis gas is the copper
liquor wash, which was used in the world's first ammonia plant in Oppau,
Germany in 1913 [321]. Descriptions of the process are given in [321-323]. The
216 Ib Dybkjaer

process removes carbon monoxide and carbon dioxide to the required low levels
by absorption in a solution containing cuprammonium salts of acetic, formic, or
carbonic acid. The chemistry involved is rather complex; both the Cu + jCu + + -
ratio, the pH, the ammonia concentration, the total concentration in water, the
temperature and pressure etc. must be controlled to obtain proper absorption
and to avoid precipitation of copper salts. Reference is made to [321, 323]. The
loaded solution is regenerated by pressure release and heating, and the composi-
tion is adjusted by necessary addition of ammonia, adjustment of the
Cu + jCu + + -ratio (by addition of air if required) etc., before the solution is
recycled to the absorber.
The copper liquor washing process has the advantage that the carbon oxides
are removed by absorption rather than by a hydrogen-consuming reaction as in
the more commonly used methanation reaction. It is, however, difficult to
operate; maintenance coats are high; it has a high energy consumption; it tends
to pollute the environment; and it therefore has largely been replaced by other
purification processes. It is not used in any modern ammonia plants.

6.3.5.2 Cryogenic Purification

Cryogenic purification of synthesis gas for ammonia production has mainly
been applied in plants based on partial oxidation of heavy oil or gasification of
coal (see Sect 6.3.2). A special process unit, the so-called "Purifier", is used for
final purification in the C. F. Braun ammonia process [324, 325, 954]. The
purifier is installed downstream of carbon dioxide removal and methanation. In
the unit, the synthesis gas is partly condensed to remove excess nitrogen which
has been introduced in the secondary reformer. All water and carbon dioxide
must be removed in an absorption unit upstream of the purifier to prevent
freezing, and, in addition, part of the inerts (methane, argon) and removed
together with the excess nitrogen in the purifier. The resulting synthesis gas is
completely dry and has a low inerts content, so that energy consumption in the
synthesis loop is reduced [326-331]. Operating experience from an industrial
plant is reported in [332, 333].
In plants based on synthesis gas production by partial oxidation or gasific-
ation using oxygen, the final purification is often done in a "nitrogen wash" by
scrubbing the gas with liquid nitrogen. In many cases the nitrogen wash is
integrated with a Rectisol carbon dioxide removal unit [296] or with other
carbon dioxide removal processes operating at low temperature. As in the
purifier process mentioned above, it is necessary in a nitrogen was to remove all
traces of carbon dioxide, water, and other impurities which could possibly cause
fouling of heat exchanger surfaces by freezing. Unscheduled defrosting of a
nitrogen wash requires a long down time and is therefore very undersirable.
Another concern is that the gas may, at the same time, contain olefins and traces
of nitrogen oxides. If these deposit together in the heat exchangers, explosions
may occur [332, 333]. The necessary pretreatment of the gas is normally done by
absorption of the impurities on molecular sieves and activated carbon. Invest-
 Ammonia Production Processes 217

igatic;ms on the risk imposed by the simultaneous presence of nitric oxides and
olefins in cryogenic units are reported in [334].
After pretreatment the gas is cooled by heat exchange with the cold product
gas. Additional refrigeration is obtained by expansion of the gas and the
nitrogen, or by addition ofliquid nitrogen. The gas is cooled to about - 190°C
~t which temperature partial condensation of carbon monoxide and methane
.::urs. The gas is contacted countercurrently in a washing column with liquid
ltrogen. The gas leaving the top of the washing column is hydrogen containing
.tbout 10 vol% nitrogen and only traces (about 100 ppm) of argon and methane.
After addition of nitrogen to adjust the hydrogen to nitrogen ratio to 3.0 this
is an excellent feed gas to a so-called "inert free" synthesis loop which may often
be operated without withdrawal of a purge gas stream. The liquid nitrogen
leaving the bottom of the washing column contains essentially all the carbon
monoxide and methane present in the original feed gas. These compounds can
be used as fuel or returned to the feed gas preparation.
The nitrogen required for the nitrogen wash is normally produced in the air
separation plant which produces oxygen for the gasification or partial oxidation
process. High purity nitrogen is required with less than 10 ppm oxygen.
Descriptions of the nitrogen wash technology may be found in [335-340].

6.3.5.3 Pressure Swing Absorption

The Pressure Swing Absorption (PSA) technology is most often used for
hydrogen purification in plants based on steam reforming or for hydrogen
recovery from off gases. It has, however, also been proposed to use the PSA
technology for final purification of synthesis gas in certain process concepts for
the production of ammonia. In most of these cases it replaces both carbon
dioxide removal unit and final purification. The schemes may be based on
production of pure hydrogen which is then mixed with nitrogen produced in a
separate unit [341-344], or a special version of a PSA unit may be used to add
nitrogen in the unit itself, thereby improving the hydrogen recovery [345-349].
In the ICI LCA process [350-357] a special PSA unit is used for carbon dioxide
removal and partial purification including removal of excess nitrogen upstream
of a methanation unit. The PSA technology as such is described in [358-361].

6.3.5.4 Methanation
By far the most important process for removal of the last traces of carbon
monoxide and carbon dioxide from ammonia synthesis gas in methanation. In
this process, carbon oxides are reacted to extinction (less than 10 ppm) at
250-350°C over a nickel-containing catalyst according to the following ex-
othermic reactions:
co + 3H 2 -+ CH 4 + H 20( - LlHg 98 = 206 kJ/mol) (6)
CO 2 + 4H2 -+ CH 4 + H 20 ( - LlHg 98 = 165 kJ/mol) (7)
218 Ib Dybkjaer

The process is simple, reliable and inexpensive both in investment and in

operating cost. It causes a certain loss of hydrogen due to the reactions (6) and
(7) and due to the increased content of methane in the synthesis gas, which causes
an increased loss with purge gas from the synthesis loop. The effect on overall
energy consumption is, however, at normal concentrations of carbon oxides
(0.2-0.5 vol% carbon monoxide and 0.01-0.2 vol% carbon dioxide) rather
limited, and the process is used in virtually all modern ammonia plants based on
steam reforming of natural gas or other light hydrocarbons. For references on
methanation, see [362-364].
The gas from a methanation unit will contain water, possibly from the
carbon dioxide removal section and at least as formed by reaction (6) and (7).
This water must be removed before the gas reaches the ammonia synthesis
catalyst. The dehydration can be done either by adsorption on molecular sieves
[365], whereafter the gas can be added to the synthesis loop just upstream of the
synthesis converter, or by co-condensation and washing with ammonia. This
last method is used in many ammonia plants, where the synthesis gas is added to
the synthesis loop before the product ammonia is separated, see Sect 6.4.1.

6.3.5.5 Methanolation
Methanolation [907, 908] may be considered an alternative to methanation.
The residual carbon oxides are converted in the methanolation process to
methanol, preferably at high pressure, i.e. after the synthesis gas compressor.
The methanol formed is removed by washing with water and may be returned to
the feed stream to the reformer or separated as a product. Full conversion of
.carbon oxides is not possible, and a clean up methanation unit must be installed
after the methanolation unit.

6.3.6 Synthesis Gas from Unconventional Sources

By far the largest part of the world's ammonia production is made from
synthesis gas which is produced by one of the processes described above, i.e. by
steam reforming oflight hydrocarbons, by partial oxidation of hydrocarbons, or
by gasification of solid feedstocks, in all cases followed by proper conversion,
purification, and adjustment of composition.
There are, however, also other sources of synthesis gas, although their
importance is limited.
In the early days of the ammonia industry the predominant feedstock was
coke. Synthesis gas was produced in so-called water gas units, which were cyclic
units. Coke was heated by partial combustion with air, whereby a mixture of
essentially nitrogen and carbon dioxide, so-called generator gas, was produced.
The hot coke was thereafter contacted with steam whereby a mixture of
hydrogen and carbon oxides,so-called water gas, was produced. The two gas
 Ammonia Production Processes 219

streams were purified and mixed, and carbon monoxide was catalytically
converted to carbon dioxide in a high temperature shift unit operating at low
pressure. The converted gas was compressed to, say, 25 bar, carbon dioxide was
removed by washing with water, and the product gas was further compressed to,
say, 200 bar, whereafter residual carbon oxides were removed in a copper liquor
wash before the gas was sent to the synthesis unit. The process, which was used
in the first ammonia plants in Oppau and Leuna, is described in [50, 366]. An
alternative process for synthesis gas preparation, which was used mainly in
Europe and Asia, was purification of coke oven gas [50, 367]. In this process,
coke oven gas was first purified by the removal of tar, benzene, ammonia, sulfur-
compounds etc., after which hydrogen was recovered in a cryogenic unit.
Nitrogen was obtained from an air separation plant, and the pure gases were
mixed and sent to synthesis.
There is still a limited production of ammonia based on coke oven gas,
although its importance is dwindling. The prospects for the future are reviewed
in [368, 369].
Hydrogen for ammonia production may also be produced by hydrolysis
[370-374] or it may be obtained from refinery off gas [375]. Plants have been
constructed from such feedstocks in the past [376, 377] and it may still be
economical in special circumstances. In modern times plants have been con-
structed on the basis of off gases from petrochemical plants such as ethylene
plants and methanol plants [378, 379], but such cases will most likely remain
rare and be economical only in special circumstances. Production of ammonia
from gases produced in steel plants is suggested in [380-383].

6.3.7 Synthesis Gas Compression

Whatever the production process for synthesis gas has been, the gas must be
compressed before it is added to the ammonia synthesis loop. Up to around
1950, synthesis gas was generally produced at atmospheric pressure. The
pressure has since then been gradually increased, and today the synthesis gas is
normally available for compression at 25-30 bar in plants using steam refor-
ming and at even higher pressure, up to about 70 bar [166] in plants based on
partial oxidation of heavy oil. Synthesis pressure was typically 200-400 bar in
early ammonia plants with extremes of 1000 bar in the original Claude process
[384] and around 80 bar in the Mont Cenis process (385). Capacities were low
compared to today's standard - increasing through the years to a maximum of
300-400 MTPD around 1960 - in part because of capacity limitations in the
reciprocating compressors used in the early plants. In the 1960s the capacity of
ammonia plants increased to about 1000 MTPD (metric tons per day) when
centrifugal compressors became available; synthesis pressures were around 150
bar in the first ammonia plants due to limitations in pressure available from the
centrifugal compressors.
220 Ib Dybkjaer

The technology has since developed further, and today plants with a
capacity of 1500 MTPD and a synthesis pressure of 220 bar are operating. In
fact the availability of compressors is no longer imposing any limit either on
capacity or on synthesis pressure in ammonia plants.
As indicated above, two types of compressors are important for the com-
pression of ammonia synthesis gas. In old plants - and in some small modern
plants - reciprocating compressors dominate, whereas centrifugal compressors
are used exclusively in all large modern plants. Reciprocating compressors are
described in [386, 387] and centrifugal compressors in [388-397]. Extensive
discussions on the use of centrifugal compressors are given in [392-397].
Reciprocating compressors are generally driven by synchronous electric motors
or in a few cases by steam turbines with special speed reduction gears. They are
more expensive than centrifugal compressors, require more maintenance, and it
is normal practice to install two machines in parallel so that excessive down time
caused by the compressor is avoided. They are lubricated so that there is a risk
for contamination of the synthesis loop with lubricating oil, they require much
space and heavy foundations for installation; and their capacity limitation
makes multi train installations necessary above a certain plant size. On the other
hand they are highly efficient, and the efficiency is rather independent of pressure
and load.
Centrifugal compressors can handle very large gas volumes; they are reliable
so that only one machine is normally installed even in very large ammonia
plants; and they have the advantage of being oil free. There is a practical limit to
the minimum clearance between an impeller and the stator and therefore to the
minimum flow (in actual volume) which can be handled by a centrifugal
compressor. Technological developments are constantly pushing this limit, but
there will always be a certain capacity limit below which centrifugal compressors
can not be used. Present technology (early 1990s) probably allows a flow
corresponding to around 350 actual cubic meters per hour at the compressor
outlet to be economically compressed in a centrifugal compressor at ammonia
plant conditions. This corresponds to an upper limit in synthesis pressure of
around 100 bar in a plant producing 300 MTPD ammonia.
Centrifugal compressors operate with very high rotational speed and are
most often driven directly by steam turbines, although motor drivers are also
used in some cases, especially for relatively small capacities. The centrifugal
compressors are inherently less efficient than reciprocating compressors, al-
though technology development may be closing the gap.
Direct steam turbine driver also means that generator loss, distribution loss,
and motor and gear loss are avoided, so that the overall efficiency of a steam
turbine driven centrifugal compressor may well compare favourably with the
efficiency of a motor driven reciprocating compressor, at least in the upper end
of the capacity ranges. Motor drives and steam turbine drivers for the large
power consumers in ammonia plants are compared in [398].
Gas turbines have also been considered as drivers for compressors in
ammonia plants. General discussions are given in [399, 400]. The exhaust gas
 Ammonia Production Processes 221

from the gas turbine can be used for steam production, for preheat duties, or as
combustion air in the primary reformer [401,402]. Experience from an actual
installation is reported in [909].

6.4 Synthesis of Ammonia

The formation of ammonia from hydrogen and nitrogen

N2 + 3 H2 ~ 2 NH3 ( - LlH~98 = 92.44 kl/mol) (8)
is an exothermic reaction, which under all practical conditions is limited by
equilibrium in such a way that only partial conversion can take place during the
passage through a synthesis reactor. Furthermore, the product ammonia is in all
practical cases separated from the unreacted gas by cooling to condense and
separate the liquid ammonia from the gas; at realistic pressures relatively low
temperatures are required to recover the product with reasonable efficiency.
These fundamental properties of the reaction and the product, together with the
properties of the synthesis catalyst and mechanical restrictions in equipment
design, determine the design of the synthesis unit and the synthesis reactor.

6.4.1 The Ammonia Synthesis Loop

The synthesis of ammonia takes place in a recycle loop which always contains
the following elements:
- A reactor system comprising one or more catalytic synthesis reactors with
associated temperature control and heat recovery equipment. The design of
the reactors and of reactor systems is discussed in Sect 6.4.2 below. The
catalysts employed are discussed in Chapter 2. Thermodynamics of the
synthesis reaction are dealt with in Chapter 1. Reaction kinetics are discussed
in Chapter 4.
- Unit(s) for cooling the product gas to recover heat and to condense the
product ammonia. The layout of these units, which are integrated with other
parts of the synthesis loop, is dealt with in the discussion of individual
processes in Sect 6.4.6.
- Unit(s) for the separation of product ammonia from unreacted gas and for the
adjustment of product properties (degassing, adjustment of temperature and
pressure). The gasjliquid equilibrium in the separation stage is dealt with in
Chapter 1.
- Units for preheating the reactor feed. These units are most often integrated
with the cooling/condensing units mentioned above.
222 Ib Dybkjaer

- Equipment for the addition of make-up gas and, if the make-up gas contains
inerts, for removal of purge gas from the recycling gas to prevent the build up
of inerts in the loop. The addition of make-up gas and removal of purge gas
may be done at various points in the recycle loop as discussed in Sect. 6.4.1.1
below.
- Equipment for the recirculation of non converted gas and make-up gas to the
synthesis reactor. The recirculation equipment is in many cases integrated
with equipment for compression of make-up gas to synthesis pressure as
discussed in Sect 6.4.1.1 below.

6.4.1.1 Types of Ammonia Synthesis Loops

It is possible to classify ammonia synthesis loops in various ways. One view-
point [23J is that "since the only proprietary data in commercial ammonia
synthesis plants are the catalyst formulation and the design of the synthesis
converter, which uses the catalyst - the rest of the process design is simply
chemical engineering for the particular conditions chosen for the plant" then
ammonia synthesis processes should be classified according to converter design.
It is, however, also possible to classify the synthesis loops according to the
arrangement of the sequence of process steps and the purge and make-up gas
points [403-405]. The most important types of synthesis loops according to this
classification are shown in Fig. 6.2A-E.
Fig. 6.2 A shows a loop with both recirculator and make-up gas addition
point after the ammonia separator. This layout is from several points of view the
most advantageous layout. Ammonia condensation and separation are done
before the converter exit gas is diluted with fresh make-up gas, and consequently
at the highest possible partial pressure of ammonia. Purge gas may be taken
from the point in the synthesis loop where the ammonia concentration is lowest
and the concentration of inert components is highest. The recycle gas from the
separator is diluted with the fresh make-up gas, so that the lowest possible
ammonia concentration is obtained at converter inlet. Also the volume of gas
which must be recompressed in the recirculator is the lowest possible, since
product ammonia has already been separated.
The disadvantage of this layout is that all impurities present in the synthesis
gas or introduced by the make-up and recycle compressors are passed directly to
the synthesis converter. Therefore, this layout is used mainly in cases where
catalyst poisons are removed very efficiently from the synthesis gas, and where
non-lubricated compressors such as centrifugal compressors are used for make-
up gas compression and recirculation. This loop layout has been used in many
commercial processes. Some early examples are the Fauser process [406J, the
Fauser Montecatini process [407J and the Casale process [408J (both with
ejector recycle), and the modified Claude process [384J (in the original Claude
process a series of converters with intermediate cooling and product recovery
was used; this is the only process which has ever been used commercially without
 Ammonia Production Processes 223

.-----4

A: "Dry Make-up Gas"

B: "4 Nozzle Design"

c: "3 Nozzle Design"

0: "3 Nozzle Design with

Two Separators"

L-_-IIIoL-_. . 6

E: "4 Nozzle Design with

Two Separators"

Fig. 6.2. Important types of am-

monia synthesis loops (1: Synthesis
Converter, 2: Separator, 3: recircul-
ator; 4: make-up gas, 5: purge gas (if
present), 6 ammonia)
224 Ib Dybkjaer

recycle of unconverted gas). In modern processes the concept with the addition of
make-up gas after the separator has been used or proposed in a number of
processes where the synthesis gas is especially purified before addition to the
synthesis loop. Examples are the MDF process by Humphries and Glasgow
[341-343], the Linde process for small ammonia plants [344] the PARC process
by KTI [345-349], and other processes using pressure swing absorption for
purification of synthesis gas. Additional examples are the C. F. Braun process
[326-331,409-412], the ICI AMV process [412-418] and other processes using
cryogenic units for gas purification, and the new M. W. Kellogg process [419],
the LEAD process by Humphries and Glasgow (420), and other processes where
the gas is dried in molecular sieve units before being added to the loop.
Descriptions of all these processes are given in Sect 6.5.3.2.3.
In cases where the synthesis gas is not especially purified, and where it may
therefore contain traces of water vapour and carbon dioxide, the make-up gas is
added before the separator. In this way the make-up gas is in contact with liquid
ammonia, which absorbes the traces of water vapour and carbon dioxide so that
these poisons do not reach the synthesis converter.
This latter layout means that the ammonia is no longer condensed and
separated at the highest possible concentration, and it further means that the
inlet concentration of ammonia to the synthesis converter is equal to the outlet
concentration from the separator, since the gas after the separator is not diluted
with fresh synthesis gas. As a consequence, there is a certain increase in the
energy consumption in the loop because of a slightly increased recirculation rate
and/or a certain increase in the energy required for condensation of ammonia.
Synthesis loops with make-up gas addition upstream of the separator are often
referred to as the "3 nozzle design" and the "4 nozzle design" depending on the
location of the recirculator [404, 405, 421, 422]. These designations refer to the
layout ofthe compressor section. In the "4 nozzle design" illustrated in Fig. 6.2B,
the circulator is located after the separator. This means that make-up gas and
recycle gas cannot be mixed in the compressor, so that 4 nozzles are required,
inlet and outlet for the make-up gas and inlet and outlet for the recycle gas. In
the "3 nozzle design" illustrated in Fig. 6.2 C, the recycle compressor is located
before the separator. In this way the make-up gas can be mixed with the recycle
gas in the compressor, and therefore only 3 nozzles are required, inlet for make-
up gas and recycle gas and outlet for the combined stream.
The "3 nozzle design" has a disadvantage compared to the "4 nozzle design"
in that the volume of gas passing through the recycle compressor is increased
because the product ammonia also has to be compressed here. The difference
between the energy consumption in the two concepts is discussed in [404, 405].
Arrangements corresponding to the "4 nozzle design" have been used in
many commercial processes including the Grande Paroisse process [423] and
processes by Uhde [424,425] and Topsoe [426,427]. The "3-nozzle design" has
e.g. been used in the Casale process (with recycle done by an ejector) [384, 408,
428] and notably by M.W. Kellogg [429-432].
 Ammonia Production Processes 225

In loops operating at high pressure (above about 250 bar) it has been found
advantageous to install two ammonia separators (404), (405). Fig. 6.2D and 6.2E
show two ways of doing this. Fig. 6.2 D is equivalent to the "3 Nozzle Design"
shown in Fig. 6.2 C, whereas Fig. 6.2 E represents the corresponding "4 Nozzle
Design". The characteristics of the designs with one separator are maintained in
each case, and in addition energy is saved, because part of the product ammonia
bypasses the last, energy consuming refrigeration stage. The "3 nozzle design
with 2 separators" (Fig. 6.2 D) has been used by M. W. Kellogg [407, 433J,
Chemico [434J, BASF [435J, Fauser-Montecatini [436J, Pritchard [437].
Topsoe [438J, and Esso [439]. The "4 nozzle design with 2 separators"
(Fig. 6.2 E) has been used in the Haber-Bosch-Mittach process [440J and the
equivalent NEC-process [441J, by Uhde [442, 443J, and by ICI [444].
In a few cases, loop designs have been suggested which cannot be described
by any of the types represented by Fig. 6.2 A-E. This is notably so for certain
complicated loops with double conversion and double separation, e.g., in late
versions of the Claude process [445-447]. A similar concept has been proposed
by PDIL [448]. It has also been proposed [449, 450J to place two synthesis
loops in series to improve overall efficiency, and to install a "preconverter" - i.e.
a synthesis converter in the make-up gas stream - in order to improve
conversion and to protect the synthesis loop against poisons [451]. Pre-
conversion has also been used in revamp of existing synthesis loops to increase
capacity [452].

6.4.1.2 Mathematical Models

The conditions in the ammonia synthesis loop can be described by a number of
parameters, which may each be in some cases independent variables, and in
other cases a function of other parameters. Important parameters are:
- Operating pressure
- Separator temperature
- Recycle ratio (ratio between recycle flow and make-up gas flow)
- Purge ratio (ratio between recycle flow and purge gas flow)
- Conversion per pass (fractional conversion of synthesis gas to ammonia by
passage through the reactor system).
The relationship between these parameters (and other parameters such as space
velocity, inert level, concentrations and temperatures at various points in the
synthesis loop, etc.) may be described in mathematical models which can be used
for design, simulation, and optimization. The models must contain procedures
to calculate the performance of each of the elements in the synthesis loop (at
least synthesis reactor and separator, and for complete models also the com-
pressor and recirculator, the heat exchangers, etc.) as well as procedures to
describe the sequence of operations and the interaction between them. Descrip-
tions of loop models are given in [453-458J; complete mass - and energy
226 Ib Dybkjaer

balances are given in [453-457]; optimization studies using computer models

are reported in [454-457]. Simulation and optimization of complete ammonia
plants are dealt with in Sect. 6.5.3.2.2.

6.4.1.3 Effect of Individual Parameters

The synthesis pressure has, as mentioned above, an important influence on the
performance of the ammonia synthesis loop because of its influence on the
reaction equilibrium, reaction kinetics, and gas/liquid equilibrium in the prod-
uct separation. A wide range of operation pressures has been used in practice,
from less than 100 bar (in the early Mont Cenis process and recently in the ICI
AMV -process) to 1000 bar (in the early Claude and Casale processes). The trend
in modern plants has been to select operating pressure in the low to medium
pressure range; typical operating parameters for modern synthesis loops at two
different pressures are given in Table 6.1.

Table 6.1. Operating conditions, NH3 loops at 220 bar

and 140 bar (1000 MTPD*) [459]

Inlet pressure, bar 220 140

Inlet flow, N m 3 jh 407,000 500,000
Inlet mol% NH3 3.8 4.1
Outlet mol% NH3 19.9 17.1
Inert level, inlet mol % 12 8
Separator temp., °C -5 -5
Relative catalyst volume, % 60 100

* 1000 metric tons per day

The operating pressure is, of course, partly imposed by considerations

concerning catalyst stability and activity. But the combined effects of the
unfavorable equilibrium conversion at low pressure and the low temperature
required to condense the ammonia product may be more important. With
reference to Fig. 1.1 and 1.4 in chapter 1 this can be explained as follows:
- at a low operating pressure, a very low operating temperature is required to
obtain a reasonably high concentration at the reactor outlet. Note that at the
relevant temperature level, a decrease in temperature of 100°C may cause a
reduction in reaction rate of about a factor of 10 (at the same ammonia
concentration).
- at a low operating pressure, a very low temperature is required to obtain
reasonable recovery of ammonia. Refrigeration for product condensation is
normally achieved by cooling with ammonia, and this limits the applicable
separator temperature to about - 25°C (ammonia boils at atmospheric
pressure at - 33 0C).
It might be argued that one could allow a higher concentration of ammonia in
the recycling gas and by increasing the concentration in the converted gas still
obtain a reasonable conversion. This would, however, increase the required
 Ammonia Production Processes 227

catalyst activity (or volume for constant activity) very significantly since the
ammonia synthesis reaction on existing types of catalyst is strongly inhibited by
the product ammonia (see Chapter 2).
A further point which favours an elevated operating temperature in the
catalyst bed, and thereby - through the equilibrium relations - an elevated
operating pressure, is the risk of poisoning. Ammonia synthesis catalysts are
extremely sensitive to poisoning by oxygen-containing compounds such as H 2 0
and CO. The poisoning is caused by a reversible adsorption of oxygen species on
the active sites, and the equilibrium is such that at temperatures below about
350 DC, almost complete coverage - and therefore almost complete deactivation
of the catalyst - is obtained even at concentrations of oxygen-containing
compounds about or below 1 ppm [238]. Such low concentrations are very
difficult (and expensive) to achieve, and as a consequence, the risk of poisoning
sets a practical lower limit to the catalyst temperature and thereby the operating
pressure.
The power consumption for synthesis gas compression, for recirculation,
and for refrigeration depends critically on the operating pressure. An extensive
study on the influence of operating pressure (in the range 140-315 bar
(2000-4500 psig)) on total power consumption is reported in [404, 405]. Loops
with single and double condensation and with 3- and 4-nozzle compressor
configuration (see Sect 6.3.1) are considered, and the power consumption is
given for a number of cases. The main conclusions are that when realistic
compressor efficiencies are used, then the total power consumption for a
constant loop pressure drop depends only marginally on operating pressure in
the range considered.
There seems to be a flat minimum at about 155 bar (2200 psig); a lower loop
pressure drop means, of course, lower power consumption; the 4-nozzle design is
slightly more energy efficient than the 3-nozzle design over the entire pressure
range; and the design with double condensation (considered only at high
pressure) is more efficient than single condensation. Some of the results are
shown graphically in Fig. 6.3 (from [405J).

Catalyst Activity/Volume and Converter Configuration. An analysis of the

effect of various parameters on overall energy consumption in an ammonia
synthesis loop is reported in [460, 461J. The main purpose of the study was to
determine the potential benefit from the installation of superactive catalysts (this
is, of course, equivalent to the installation of a large catalyst volume). It was
found that when optimum values of inert level, separator temperature, etc. are
selected, then the energy consumption in the loop depends only on the equilib-
rium temperature which is established at converter exit. (see Fig. 6.4 (from
[460J). The energy consumption at a constant equilibrium temperature was
found to be almost independent of loop pressure in the range of 80-220 bar
(8-22 MPa). The influence of various parameters on the conversion per pass
through the synthesis converter system is discussed in Sect. 6.4.2. Various
converter configurations are discussed in Sect. 6.4.3.
228 Ib Dybkjaer

32 --------=
total recycle refrigerant
and syn. gas compressors 0

!-3stagedesign
a abQ\le3200 p.s.i.g.

'- 24
~0.
Q.I
III
'-
0

/.
.r;
"6 16
.!:!
~ /' refrigerant
0
Q.I
.r;
l-
--- compressor

S ,,
,
Fig. 6.3. Theoretical horsepower versus syn-
thesis pressure for base case design (early M. W.
Kellogg process) for plant capacity of 1500 short
tons of ammonia per day (from [405])

Nominal synthesis pressure. p.s.i.g.

( discharge of compressor)

M
1.5
J:
Z
I-
~
...,
<!)
III 1.0
g'
.~
>-

.n~ 0.5

Fig. 6.4. Energy saving as function of equilibrium temper-

ature at outlet last catalyst bed. Basis: Haldor Top0e S-200
synthesis loop (from [460])

Equilibrium temperature (OC)

The separator temperature, together with operating pressure and location of

make-up gas addition point, determines the ammonia concentration at the
converter inlet. A low temperature means low ammonia concentration, which
again means either a low catalyst volume or high conversion. But low separator
temperature can only be obtained by increased power consumption and cost of
the refrigeration units, so that the choice of separator temperature is a com-
 Ammonia Production Processes 229

promise between energy consumption and capital costs in various parts of the
synthesis unit. For an existing synthesis loop, a reduction in separator temper-
ature will mean an increase in production capacity [449, 462] since the
converter with its given catalyst volume will be able to make more ammonia due
to the lower inlet concentration.
The inert level in the synthesis loop (most often measured at converter inlet)
depends on the inert level in the make-up gas, the production of ammonia per
unit make-up gas (the loop efficiency), and the purge rate. It may easily be seen
that when two of these four parameters are given, then the two others can be
calculated, if steady state (including constant pressure) is maintained in the loop.
The inert level in the make-up gas is, of course, determined by the conditions in
the synthesis gas preparation unit. The ammonia production is determined by
conditions around the converter, the gas flow (which may be expressed by the
recycle ratio) the inlet temperature and pressure, catalyst volume and activity,
and converter configuration (see Sect. 6.4.3).
This means that for a given situation neither the purge rate nor the inert level
can be changed without affecting operating pressure. It also means that if the
operating pressure or the catalyst activity (by change of catalyst) is changed,
then the other parameters will change until a new steady state is obtained.

6.4.2 Design of Ammonia Synthesis Reactors

6.4.2.1 General
Ideally, the design of a chemical reactor is a strictly logical process, in which the
optimum solution to the problem is determined from experimental and other
data by well-defined procedures. However, the real world does not work in this
way. This is clearly shown, if not by other evidence, then by the multitude of
different solutions which have been found in reactor designs for the same
problem. All the solutions are normally technically and economically very
reasonable - and none of them can claim to be clearly superior in every case.
The term "reactor design" is often interpreted as meaning "calculation of
necessary reactor (catalyst) volume". There are, however, [457] several other
factors to consider in reactor design - at least when this is interpreted as all the
information required as an instruction to a workshop to manufacture a reactor
for a specific practical application. Some of the more important points are listed
below (from [459]).
Chemistry and thermodynamics
- the reaction including possible side reactions
- thermodynamics
- equilibrium
- heat of reaction
- properties of reactants and products
230 Ib Dybkjaer

Reaction kinetics
- intrinsic + diffusion parameters
- pellet kinetics
Properties of catalyst
- thermal stability
- particle size and shape
- pore system
- mechanical strength
- change in properties during activation and
operation
Process optimization
- energy recovery
- pressure drop
limitations on volumetric flow
Process control
- temperature control
- flow distribution
- safety aspects, start-up/shut-down
Mechanical design
- materials of construction
- workshop manufacture
- sea and land transport
- erection
The basic data - the chemistry and the thermodynamics for the process - will
define limits for a broad range of conditions at which the reaction is possible.
The kinetics of the reaction and the properties of the catalyst, especially its
thermal stability, will further narrow the range of possible reaction conditions
and define a "window" of possible operating parameters. Process optimization,
energy efficiency, and safety aspects will then determine at what conditions
within the "window" the reactor should operate to give the optimum result. And
then mathematical models are used to determine how big the reactor must be to
obtain the performance (conversion and pressure drop) determined by the
process optimization.

6.4.2.2 Reactor Calculations

In mathematical models for design, simulation, and/or optimization of am-
monia synthesis reactors, the normal principles for reactor calculations are, of
course, applied. Reference is made to [463-465]. A general purpose computer
 Ammonia Production Processes 231

program for reactor calculations is described in [466, 467]. Application of the

program for analysis of converter performance is described in [462, 468].
General discussions about modelling of ammonia synthesis reactors and com-
parison of different converter types may be found in [469-472]. Models of
multi bed quench converters and optimization studies for such reactors are
reported in [473-479]. In [473] is described a study on conversion of tubular
converters to quench cooled converters inside the same pressure shell. In [474]
is discussed the effect of variation of operating parameters such as feed
temperature, pressure, inert level, feed rate, and hydrogen to nitrogen ratio, on
conversion and converter stability. Similar studies are reported in [476, 477].ln
[475] is compared the performance of a quench cooled and a TVA type
converter with emphasis on analysis of the quench converter performance.
Models for tubular converters are discussed in [480-486]. In [480] is discussed
the effect of variations in operating parameters such as feed rate, feed gas
concentration of inerts and ammonia, catalyst activity, and heat conductivity on
conversion and converter stability. In [481] is analyzed the transient behaviour
of the converter. In [483] is contained a critical review of the work reported in
[455, 456, 482]. A proposal to optimize converter performance by optimizing
the heat transfer coefficient between the catalyst bed and cooling tubes is
presented in [484]. In [485] are reported results from the use of a model to
control an existing ammonia plant, and in [486] is analyzed the effect of catalyst
deactivation on converter performance.
In addition to the above, the effect of operating conditions on conversion in
synthesis reactors is discussed in [487-489]. The general results are:

- Pressure: Increasing pressure increases conversion due to the increasing

reaction rate and the increasing equilibrium conversion.
- Inlet Temperature: Has a dual effect since increasing temperature increases the
reaction rate, but decreases the adiabatic equilibrium concentration.
- Space velocity: Increasing space velocity will normally decrease outlet am-
monia concentration, but increase total ammonia production.
- Inlet Ammonia Concentration: Increasing inlet ammonia concentration de-
creases conversion since reaction rate decreases and adiabatic equilibrium
conversion decreases.
- Inert Level: Increasing inert level decreases conversion since both the reaction
rate and the adiabatic equilibrium concentration decrease due to the lower
partial pressure of hydrogen and nitrogen.
- Hydrogen/Nitrogen ratio: Exhibits a true maximum since reaction rate has a
maximum at a certain hydrogen to nitrogen ratio. The position of the
maximum depends on space velocity, but will generally be between 2.0-3.0. It
also may be noted that the equilibrium conversion of hydrogen-nitrogen
mixtures to ammonia is maximum at a hydrogen to nitrogen ratio below 3
(see Chapter 1).
- Catalyst Particle Size: Smaller catalyst particles give a higher conversion due
to lower diffusion restrictions (higher efficiency factor, see [459]).
232 Ib Dybkjaer

6.4.2.3 Principal Reactor Types

In ammonia synthesis, the desirable conditions around the reactor system
as described above are such that it is normally not possible to go from the
inlet to the outlet conditions in the reactor in one adiabatic step. Some type of
cooling is required. In principle the cooling can be applied in three different
ways:

- internal cooling with cooling tubes in the catalyst bed or with catalyst in tubes
surrounded by a cooling medium. The internal cooling can be effected either
with gas flow in the same direction in the catalyst bed and cooling channels
(co current flow), with gas flow in opposite directions in the catalyst bed and
cooling channels (countercurrent flow), or with gas flow in the catalyst bed
perpendicular to the flow in cooling channels (cross flow). The cooling
medium can be either synthesis gas or some other medium, for example
boiling water.
- quench cooling by injection of cold gas. The injection of quench gas can be
either between adiabatic beds or into a catalyst bed at different locations.
Flow in the catalyst beds can be either axial or radial in vertical converters or
downwards in horizontal converters.
- external cooling by heat exchange between catalyst beds. The cooling medium
can be either synthesis gas or some other medium, for example boiling water.
Flow in the catalyst beds can be either axial or radial in vertical converters or
downwards in horizontal converters.
It is of course possible to combine several cooling methods in the same converter
~ystem. Furthermore, the catalyst beds and/or heat exchangers including feed
effluent heat exchanger may be arranged in one pressure shell or in individual
pressure shells. It is clear that this leads to a very significant number of possible
converter configurations.
The various converter types may be characterized by the temperature profile
through the catalyst bed(s) or by the temperature/concentration profile (plots of
temperature vs ammonia concentration for the gas passing the converter) (see
Fig. 6.6a-d below). Such profiles are often compared to maximum reaction rate
profiles, see Fig. 6.5 (from [460]). It is seen from this figure that when the
temperature is increased (at otherwise constant conditions, including constant
ammonia concentration), then the reaction rate will increase up to a maximum
value; when the temperature is further increased, the rate decreases until it
becomes zero at the equilibrium temperature. The temperature/concentration
points where maximum rate is achieved describe a curve, the maximum rate
curve, which will normally be roughly parallel to the equilibrium curve, but at
30-50 °C lower temperature. It is clear that the minimum catalyst volume would
be obtained in a converter where this maximum rate curve were followed. In the
early days of ammonia production, available technology limited the obtainable
size of the converter pressure shell, and the physical dimensions of the converter
 Ammonia Production Processes 233

optimum operating
:z:'" line
z lines with constant
reaction rate
(5
>

Fig. 6.5. Optimum reaction path or maximum rate

curve. The difference between the lines with con-
stant reaction rate is a factor of 10. Lowest reaction
rate at highest NH3 concentration (from [460J)

Temperature

thus limited the achievable production capacity. Great emphasis was therefore
given to the ammonia production per unit converter volume. Quite complicated
mechanical constructions were used to maximize the production capacity of a
given volume, and the converters were compared to the "ideal" converter where
the temperature/concentration plot follows the maximum rate curve (see e.g.
469)). As technology developed, other considerations such as optimum heat
recovery, reliable mechanical construction, easy catalyst loading and unloading,
etc. became more important than maximum production per unit volume, and
different - mechanically simpler - converter configurations have, therefore,
become more popular.
In connection with "revamps", i.e. modification of existing plants for in-
creased capacity and/or increased energy efficiency (see Sect. 6.4.3.4), where the
pressure shell of existing converters is reused, it has, however, again become a
major consideration to maximize the production capacity of the volume avail-
able inside the pressure shell.
Ammonia converters most often consist of two separate parts, an outer
pressure vessel and an internal "basket" containing the catalyst bed(s), internal
piping for gas distribution, heat exchangers (when applicable) for control of
catalyst temperatures, and in some cases a feed-effluent heat exchanger, so that
all high temperatures are contained inside the pressure shell.
Early types of ammonia synthesis converters are described in [4, 50, 385].
More recent developments are discussed in [23, 490-492]. Good overall reviews
of different converter designs may be found in [493,494].
A discussion of the types of heaters used for start-up of ammonia synthesis
converters (fired vs. electrical heaters) is given in [910].
234 Ib Dybkjaer

60)
0

U 550
~
E
~
::I:
c5
Qj 450
0-
E
~

120 3500
0.1 0.2
.,.
200 400 600
Temperature (Oe)
6b)
0

4 U 550
~
g ~
"
"
::I:

Qj 450
0-
E
{!!.

3500
0.2
.,.
200 400 600 0.1
Temperatu re (Oe)

6e) I I
2 U 550
~
~
E e" 450
::I:
6 ~
E
{!!.

3500
.,.
400 600 0.1· 0.2
Temperature (Oe)

6d)

U 550
:§ ~
~
:x:
"
"Qj 450
0-
E
{!!.

120 3000
.,.
200 400 600 0.1 0.2
Temperature (Oe)
 Ammonia Production Processes 235

6.4.3 Commercial Ammonia Synthesis Converters

A very significant number of different ammonia synthesis converter designs have

been used in industrial practice. In the following survey the different designs
have been characterized, first by the cooling principle applied, and thereafter
mainly by flow direction through the catalyst bed(s).

6.4.3.1 Converters with Internal Cooling in Catalyst Beds

In converters with internal cooling the required cooling is supplied by a cooling
medium which flows in most cases in cooling tubes in the catalyst bed or, in a
few designs, around tubes containing the catalyst. The flow in the cooling tubes
may be countercurrent or cocurrent (in the opposite direction or the same
direction as the flow in the catalyst bed).

6.4.3.1.1 Countercurrent Flow in Cooling Tubes

The most important converter type using internal cooling with countercurrent
flow in cooling tubes is the TVA-converter. A schematic drawing of the converter
and the corresponding temperature- and temperature/conversion profile is
shown in Fig. 6.6a. Feed gas enters at the top and passes in the annulus between
the pressure shell and the basket to the bottom of the converter. In this way the
pressure vessel is cooled (the feed gas is used as "shell cooling gas") so that a
lower design temperature is applicable for the expensive pressure vessel.
The feed gas passes through the feed/effluent heat exchanger, which is
contained in the same pressure shell as the catalyst bed (when installed in this
position the heat exchanger is often referred to as "the lower heat exchanger"),
on the shell side and further up through the cooling tubes in the catalyst bed.
After passing in up flow through the cooling tubes the gas turns and flows in
down flow through the catalyst bed and further through the tube side of the
lower heat exchanger to the exit. A bypass steam (often called "cold shot") is
added for temperature control to the feed gas between the lower heat exchanger
and the catalyst bed cooling tubes. It is seen from the temperature profile that
the gas is at relatively low temperature at converter outlet. This means that the
waste heat can be used only for preheating boiler feed water or producing low
pressure steam.

Fig. 6.6. Schematic drawing, typical temperature profile, and operating curve (temperature/am-
monia concentration plot) for four important converter types. (from [471])
a Internal cooling, countercurrent flow (TV A-converter)
b Internal cooling, cocurrent flow (NEe-converter)
c Quench cooling
d Indirect cooling (heat exchange)
236 Ib Dybkjaer

If a design is used where the feed/effluent heat exchanger is located in a

separate pressure shell, then the converter outlet temperature becomes equal to
the catalyst bed outlet temperature, and the converter outlet system must be
designed for this higher temperature. In such systems the waste heat may be
recovered at higher temperature so that it can be used, for example, for
production of high pressure steam.
The temperature/concentration profile illustrates how the gas reacts almost
adiabatically in the first part of the catalyst bed. As the temperature difference
between the cooling gas and the reacting gas increases, and the reaction rate
decreases due to the approach to equilibrium, the temperature is decreased so
that the approach to the equilibrium line increases, and the profile moves close
to the maximum reaction rate line. In the bottom of the catalyst bed the reaction
rate becomes low, and the temperature drops below the maximum rate line. It is
evident that in order to obtain the best performance of this converter type, it is
important that the cooling must match the heat evolution. Too efficient cooling
will lead to a low reaction rate or even to the loss of reaction, whereas too little
cooling will lead to a too close approach to equilibrium and therefore to
inefficient use of the catalyst. A TV A-type converter of special design is
described in [495]. In this design, core rods are installed in the cooling tubes to
enhance heat transfer, thereby reducing the required heat transfer area.
Stability problems and performance optimization of TVA-converters are
discussed in [480,481,484,485]. A classical account of operating problems in a
synthesis unit using TV A-converters is given in [496J which also gives mechan-
ical details on the construction of a TVA-converter. Drawings showing mechan-
ical details of TV A-converters may also be found in [23, 385, 497, 498]. TVA
. type converters have been used extensively, and may are in operation today.
This converter type has been suggested quite recently for installation in new,
relatively small plants (up to about 300 MTPD of ammonia) by process
licensors such as Tos0e [499J and ICI [500].
Another converter type using countercurrent flow in cooling tubes is the
SBA-converter [501]. In this converter the feed gas enters at the bottom of the
pressure shell and passes upwards in the annulus between the pressure shell and
the basket, flows down through tubes which serve both as heat exchanger tubes
in a gas/gas heat exchanger located in the top of the converter and as cooling
tubes in the catalyst bed. The preheated gas passes in up flow through the
catalyst bed and through the gas/gas exchanger on the shell side before it leaves
the converter at the top. Converters of this type have recently been revamped
into Tube Cooled Radial Flow converters [502].
Old converter types with countercurrent flow in cooling tubes are the Mont
Cenis reactor [385, 503J, the original Haber-Bosch converter [504J, the Claude
converter [505J, and the "old" Fauser converter [506, 507J. These early
converter types were all used in relatively small plants; they are not used in
modern processes.
Combinations of countercurrent cooled catalyst beds with adiabatic beds
have been suggested by ICI [508J, Uhde [509J and in [510, 511].
 Ammonia Production Processes 237

6.4.3.1.2 Cocurrent Flow in Cooling Tubes

The most important converter type using internal cooling with cocurrent flow in
cooling tubes in the NEC converter or the Chemico converter. A schematic
drawing of the converter and the corresponding temperature and temper-
ature/conversion profile is shown in Fig. 6.6b. Process gas enters the converter
at the top, passes downwards in the annulus between the catalyst basket and the
pressure shell as shell cooling gas to the shell side of the lower heat exchanger
and therefrom to cooling tubes in the catalyst bed. The tubes are bayonet tubes,
and the gas flows up through the inner tube, down in the annular space between
the two concentric tubes, up through a center pipe and finally down through the
catalyst bed and through the tube side of the lower heat exchanger to the exit.
The cocurrent flow makes it possible to obtain a temperature profile which
is close to the optimum; the gas is heated adiabatically to a temperature close to
the maximum reaction rate curve, and the temperature/conversion profile then
follows closely the maximum rate curve for the test of the catalyst bed length.
The NEC- or Chemico converter with cocurrent flow in the cooling tubes
has -like the TVA-converter - been used since the early days of the ammonia
industry, and many modifications have been suggested. Descriptions of various
versions may be found in [23, 385, 492, 512-516]. A converter with cocurrent
cooling in flat cooling "fins" in the catalyst bed is described in [517]. A converter
type which combines adiabatic catalyst beds with a catalyst bed using cooling
tubes with cocurrent flow has been developed by the Japan Consulting Institute
[518,519]. The flow pattern in this converter is as follows: Feed gas enters at the
top and passes as shell cooling gas to the bottom of the converter. It passes then
on the shell side of a lower heat exchanger and further in an annulus between the
tubtllar cooled catalyst bed and the outer wall of the basket to the top of the
tubular cooled bed, in down flow through cooling tubes in the catalyst bed, up
through a central pipe to the top of the converter, down through two adiabatic
catalyst beds in series with quench cooling between the beds and finally through
the tubular cooled catalyst bed and through the tube side of the lower heat
exchanger to the exit. Cold gas is added for temperature control after the first
passage of the lower heat exchanger and (as quench gas) after the passage of the
first adiabatic bed.

6.4.3.1.3 Cross Flow

Ammonia synthesis converters with radial flow in tubular cooled catalyst beds
have been suggested by Toyo Engineering Corp. [520] and in [502]. The Toyo
concept has, so far, not been used industrially, while the concept described in
[502] has, as mentioned above, been demonstrated in revamps of converters
originally designed by SBA. It is claimed that the cross flow makes it possible -
through proper design of the cooling tube bundles - to optimize the temperature
profile so that it follows very closely the maximum reaction rate curve. It is
furthermore reported that the heat transfer coefficients obtained in practice in
238 Ib Dybkjaer

the cross flow configuration are significantly higher than in axial flow. A
converter with radial flow and with two zones in each of a series of catalyst beds,
an adiabatic zone and a zone with cooling tubes, is described in [521].

6.4.3.2 Converters with Quench Cooling

In quench cooled synthesis converters the required cooling is supplied by
injection of cold, unconverted feed gas either between separate catalyst beds or
through proper distributors directly into a catalyst bed at different levels.
A schematic drawing of a quench cooled converter and of the typical
temperature and temperature/conversion profile is shown in Fig. 6.6c. Only part
of the feed gas passes through the first catalyst bed where the temperature
increases adiabatically from the inlet temperature to a temperature normally
quite close to the equilibrium temperature. After the first bed an amount of cold
feed gas is added so that the temperature is decreased to the desired inlet
temperature to the second bed. At the same time the ammonia concentration
decreases because the exit gas from the first bed is diluted with unconverted gas.
The quench cooling concept is adaptable to large capacity converters and was
almost exclusively used in the early generation of large single stream units
constructed in the 1960s and 1970s.
Even today quench converters are extensively used although they are
increasingly being replaced by indirectly cooled converters, both for appli-
cations in new plants and by replacement or modification of converters in
existing plants in revamp projects (see Sect. 6.4.3.4).
Quench converters appear to have been used first in plants constructed in
the USA during the second world war with M. W. Kellogg as the engineering
contractor [522]. The early Kellogg quench converter is also described in [385,
523].
The converter is described as a modified NEC converter in which the
catalyst cooling tubes have been replaced by cold gas injections between layers.
10-12 catalyst beds were used in this early generation of quench converters.
Other early designs of quench converters were the Uhde converter [524, 525]
and the so called "modified Haber-Bosch" reactor [385, 526] or the BASF
reactor [435, 490].
As mentioned above, quench cooled converters were typical for the first
generation of the large single stream units constructed in 1960s and 1970s. The
most important quench type converter designs used in these plants may be
characterized by the arrangement of the quench cooling and by the flow
direction in the catalyst bed.

6.4.3.2.1 Axial Flow, Quench Cooling Between Separate Catalyst Beds

The most important example of this converter type is the M. W. Kellogg 3- or 4-
bed Quench Converter which was used in a large number of plants. The
converter is described in [490,491, 527, 528]. A simplified drawing of a 3-bed
 Ammonia Production Processes 239

converter is shown in Fig. 6.7 (from [491]). The feed gas enters the converter at
the bottom and passes as shell cooling gas in the annulus between catalyst
basket and pressure shell to the feed-effluent heat exchanger mounted in the top
of the converter.
The feed gas passes on the shell side of the exchanger to the first catalyst bed.
Between the exchanger and the first catalyst bed and between the following
catalyst beds cold gas is added via quench gas distributors for temperature
control. The converted gas passes from the last catalyst bed through a center
pipe to the tube side of the feed-effluent heat exchanger and leaves the converter
at the top.
This converter type has given excellent service in industry. It is, however, due
to the inherent weaknesses of the quench cooling system, not very efficient, and
the performance has in many cases been improved by revamping, see Sect.
6.4.3.4.
Designs similar to the Kellogg design have been used by others, e.g. Grande
Paroisse [385,492,529], Kubec [530, 531] and Casale [490]. A design in which
quench cooling is applied between catalyst beds contained in separate vessels
has been used by Pritchard [437].
A later development is the M. W. Kellogg horizontal quench converter, see
Fig. 6.8 (from [492]). This converter design is described in [533-536]. The
catalyst beds are arranged in a basket which fits into a horizontal pressure shell;

outlet

catalyst bed

catalyst

catalyst

shell

Fig. 6.7. Three-bed Quench Cooled Axial

Flow Converter. M. W. Kellogg design.
(from [491])
240 Ib Dybkjaer

cartridge gas transfer line internal heat exchanger

inlet
manhole

Fig. 6.8. Quench cooled horizontal converter. M. W. Kellogg design (from [492])

in this way catalyst loading and unloading are facilitated and there is no need for
overhead structure or crane. The basket may be removed from the pressure shell
simply by drawing it out on tracks.
In operation, gas flow is downwards through the catalyst beds. Because of
the large area, shallow beds, a low pressure drop is obtained even with small
catalyst particles. This means that the horizontal design offers some of the same
advantages offered by radial flow converters (see Sect. 6.4.3.2.3.).

6.4.3.2.2 Axial Flow, Quench Cooling by Injection of Gas into the Catalyst Bed
This converter type has mainly been used by ICI; it is described in [23,490,491].
A schematic drawing is shown in Fig. 6.9 (from [491J). Synthesis gas enters at
the bottom of the converter, flows as shell cooling gas in the annulus between the
catalyst basket and pressure shell to the top of the converter, down on the shell
side of a centrally mounted feed-effluent heat exchanger, upwards in an annulus
between heat exchanger and catalyst bed, down through the catalyst bed, and up
on the tube side of the feed effluent exchanger to the outlet in the top of the
converter. Quench gas is added at various levels in the catalyst bed through gas
distributors. A special version of this converter type - the opposed flow
converter - was developed for very large capacities [537, 538]. In this design the
converted gas is collected in the middle of the catalyst bed, and there is down
flow in the upper half and up flow in the lower half of the bed. The special feature
of the ICI design - one uninterrupted catalyst bed which can be emptied
through an opening in the bottom of the converter - is maintained in the
opposed-flow converter.
A design similar to the ICI-design has been suggested by Chemico [539].

6.4.3.2.3 Radial Flow or Axial/Radial Flow

A common problem for all converters with axial flow is that when the capacity
increases, the pressure drop increases due to the increased depth of the catalyst
bed. It is of course possible to some extent to compensate by increasing the bed
 Ammonia Production Processes 241

gas outlet
quench gas inlet quench gas inlet
G

quench gas
KD»+++-- distributors

heat exchanger

Fig. 6.9. Quench cooled conver-

catalyst ter with injection of gas into the
discharge catalyst bed. leI design. G is the
nozzle synthesis gas inlet cold gas inlet for temperature con-
trol (from [491J)

diameter, but above a certain size this becomes technically and economically
unacceptable. In order to partially overcome the pressure drop problem it is
normal in axial flow converters to use relatively large catalyst particles.
However, these have lower activity than small particles due to diffusion
restrictions, and a larger catalyst volume is therefore required [459, 540].
In converters with radial flow the above mentioned disadvantages do not
exist. Radial flow converters can be designed for very large capacities without
excessive reactor diameter, and a low pressure drop can be maintained even with
very small catalyst particles. The advantages of radial flow reactors for ammonia
synthesis is discussed in [459, 502, 541, 542]. The pressure drop as function of
catalyst particle size is discussed in [488]. Radial flow has been used in
converters with cooling tubes [520], in quench cooled converters (see below)
and in converters with indirect cooling (see Sect. 6.4.3.3.2).
The most widely used quench cooled radial flow converter has been the
Tops0e designed S-100 converter, which is a two-bed quench cooled reactor. A
schematic drawing is shown in Fig. 6.10 (from [544]). The converter is described
in [23, 427, 490, 491, 534, 535, 543-547]. The main part of the synthesis gas
enters the converter at the top and passes downwards as shell cooling gas in the
annulus between the pressure shell and the catalyst basket. It then passes
 242 Ib Dybkjaer

~:...::.£.~r=:=~~iniet quench gas

I:' inlet main gas
tJiiiiiii.iiiiilf!l.t¥~ outer internal lid
.-eI-H--inner internal lid
+II*- fl rst catalyst chamber
inner annular space
-H1I-N--perforated center tube
.-RI-H--cotalyst support plate 1
-H!t-kt-- second catalyst chamber
+KI-P+-- tronsfer tube
outer annular space

-e*-cataiyst support plate 2

1:::Rt-H-- heat Il'xcnanger

U>lI"1rr-- ref ractory fiber

pressure sheil
~,,*-- refractory cement Fig. 6.10. Two bed radial flow quench
cooled converter. Haldor Tops0e S-100
Converter (from [543])
I- outlet main gas
inlet cold by-pass gas
and start -up heater

through the feed/effluent heat exchanger ("lower heat exchanger") on the shell
side and up through a central pipe to the first, upper catalyst bed. After passing
the catalyst bed in radial flow in the outwards direction, the gas is cooled by
quenching with fresh synthesis gas before it passes through the second, lower
catalyst bed in inward direction, downwards in an annulus between the catalyst
bed and the center pipe to the tube side of the lower heat exchanger and out
through the bottom of the converter. Cold gas is added through the bottom and
mixed with the inlet gas to the first catalyst bed for temperature control.
Radial flow quench converters have also been used by Chemoproject [548J,
Osterreichische Stickstoffwerke [549J, and Lummus [550]. In these converters,
as well as in the Top0e converter, the flow direction is restricted to be essentially
radial.
An axial-radial flow converter has been introduced by Ammonia Casale
[490,492,551-553]. The special feature of this design is that gas can enter each
catalyst bed both from the top (in axial direction) and from the side through
perforations (in radial direction), see Fig. 6.11 (from [492J). The gas leaves the
catalyst bed through perforations in the inner wall. There are no perforations in
an upper part of the inner wall, and the gas entering at the top of the bed is
therefore forced to flow through part of the catalyst in partially axial flow before
it can leave the catalyst bed. This flow principle can be used in quench cooled
 Ammonia Production Processes 243

radial
gas flow zone

) Fig. 6.11. Principle of axial-radial

flow converter. Ammonia Casale de-
sign (from [492])

converters and in indirectly cooled converters (see Sect. 6.4.3.3.2). The design has
found widespread use in modification of existing converters, see Sect. 6.4.3.4.

6.4.3.3 Converters with Indirect Cooling

In converters with indirect cooling, the cooling required to reach a desired high
conversion in the converter is provided between the catalyst beds by some
cooling medium, which most often is synthesis gas or boiler feed water (in steam
boilers). The heat exchange most often takes place in exchangers installed
together with the catalyst beds inside one pressure shell, although heat ex-
changers installed between separate, adiabatic converters have also been used.
Typical temperature and temperature/conversion profiles for an indirectly
cooled converter are shown in Fig.6.6d. In contrast to the quench cooled
converter (see Fig. 6.6c) all gas passes through all catalyst beds, and there is no
dilution of the partially converted gas between catalyst beds. This means that at
otherwise identical conditions, a higher conversion can be obtained in the same
number of catalyst beds in the indirectly cooled converter than in the quench
cooled converter.
Converters with indirect cooling are used in almost all large, new ammonia
plants constructed today. The cooling principle is also introduced in many
existing plants by converter revamps, whereby quench cooled converters are
modified into converters with indirect cooling by reconstruction ofthe converter
internals (see Sect. 6.4.3.4).

6.4.3.3.1 Axial Flow

Converters with indirect cooling have been used since the early days of
ammonia synthesis, for example by Fauser-Montecatini [23, 385,490,492, 506,
554-556]. In this converter, cooling is provided by circulating water which in
turn delivers the heat to an external high pressure steam boiler.
244 Ib Dybkjaer

In the converter design developed by Osterreichische Stickstoffwerke (OSW)

the cooling between the catalyst beds is provided by the converter feed gas. The
OSW converter - also known as the Uhde-Chemie Linz converter - is described
in various versions in [23,490-492,557- 568]. A schematic drawing is shown in
Fig. 6.12 (from [568]). The feed gas enters at the top of the converter, flows as
shell cooling gas in the annulus between the pressure shell and catalyst basket,
up through the shell side of the lower (feed/efHuent-) heat exchanger and - via a
center pipe - through the interbed heat exchangers to the top of the converter.
The gas then passes downwards in axial flow through the catalyst beds and on
the tube side ofthe interbed heat exchangers and the lower heat exchanger to the
exit at the bottom of the converter. Cold synthesis gas ("cold shot") is added
through the bottom of the converter and mixed into the feed gas to the
individual catalyst beds for temperature control. The OSW /Uhde-design with
interbed heat exchange was further developed into a horizontal converter [563]
and into several Uhde designs with radial flow (see below).
A horizontal converter design with indirect cooling also was proposed in
[569] and has been used by M. W. Kellogg in their low energy process [492, 570,
911,571]. A schematic drawing of this converter is shown in Fig. 6.13 (from
[492]). Gas enters at one end of the horizontal pressure shell, passes as shell
cooling gas between the pressure shell and the catalyst basket to the shell side of
the interbed heat exchanger which is installed at the other end. After preheating

Fig. 6.12. Converter with indirect cooling, axial flow. OSW design
(from [568])
 Ammonia Production Processes 245

inlet

---=----f----=----
bed 2

outlet
interbed heat exchanger

Fig. 6.13. Horizontal converter with indirect cooling between catalyst beds. M. W. Kellogg design
(from [492])

in the heat exchanger and mixing with cold gas ("cold shot") for temperature
control, the gas passes downwards through the first catalysed bed, through the
interbed heat exchanger on the tube side, downwards through the second
catalyst bed (which is divided into two sections) and out in the same end of the
pressure shell as the main inlet. This design offers the same advantages as a
horizontal quench cooled converter - a low pressure drop with small catalyst
particles and the easy removal of converter internals without requiring a crane
or overhead structure.
A design with indirect cooling between catalyst beds is also used by C. F.
Braun. In this case the catalyst beds and the heat exchangers are each in their
separate pressure shell. In an early design [401, 411, 572, 573J two adiabatic
converters were used, and the cooling between the converters was by heat
exchange between feed and product gas to/from the first converter. The
arrangement and the converter design is shown in Fig. 6.14, and Fig. 6.15 (from
[572J). In a later design [573- 576, 952J three reactors are used, and the cooling
between the reactors is provided partly by gas/gas heat exchange, partly by
external cooling, for example by raising or superheating steam. The C. F. Braun
converter (see Fig. 6.15) contains one adiabatic catalyst bed with axial flow. Gas
enters at the bottom, flows as shell cooling gas to the top, down through the
catalyst bed and out at the bottom. The design of the bottom part of the
converter is critical due to the relatively high temperatures prevailing there.
leI has proposed a split flow converter with indirect cooling between beds
[577]. Feed gas is introduced in the top of the converter and flows to the shell
side of an interbed heat exchanger, where it is preheated to the inlet temperature
of the first conversion stage. It is then passed to at least two parallel axial flow
beds installed above the heat exchanger, where the first conversion takes place.
After these "upstream" beds the gas is cooled by passing through the interbed
heat exchanger on the tube side. It then reacts further in at least two parallel
axial flow beds installed below the heat exchanger. A somewhat similar system
has been proposed by M. W. Kellogg [578]. In this case the beds are arranged in
separate vessels, the heat exchanger is external, and the beds are with opposed
flow.
246 Ib Dybkjaer

Fig. 6.14. Separate converters with axial flow

and indirect cooling between the converters.
C. F. Braun design (from [572])

Fig. 6.15. C. F. Braun converter design (from [572])

 Ammonia Production Processes 247

6.4.3.3.2 Radial Flow or Axial/Radial Flow

The radial flow principle (see Sect. 6.4.3.2.3) has also been used in indirectly
cooled reactors.
The Tops0e S-200 reactor [459-461, 492, 579-581J is illustrated in Fig. 6.16
(from [492J). Two versions are shown, with and without a lower heat exchanger.
In both cases the converter consists of a pressure shell and a basket. In the
converter without a lower heat exchanger, the feed gas enters at the bottom and
flows as shell cooling gas to the top of the converter and through the interbed
heat exchanger, which is installed centrally in the first bed. After the passage
through the inter bed heat exchanger the gas is mixed with cold gas ("cold shot")
for temperature control before it passes through the first bed radially from
outside in, through the tube side of the interbed heat exchanger, to and through
the second bed, also from outside in, and finally out at the converter bottom. In
the converter type with lower heat exchanger the gas flow is modified to allow
the feed gas and the exit gas to pass through this heat exchanger.
The S-200 converter may be considered a further development of the S-100
converter, and it has largely replaced the latter in industry. The S-200 converter
design has also been used in the revamping of existing converter installations,
see Sect. 6.4.3.4. In a similar way the axial/radial flow principle of Casale has
been used in indirectly cooled converters [582, 583]. This design is also being
offered both for new plants and for revamps. Other radial converter designs
have been suggested by Uhde [584-586J, by M. W. Kellogg [587, 588J and by
Snamprogetti [589]. In the last case the cooling is provided by a boiler centrally
installed in the first radial flow catalyst bed.

6.4.3.4 Revamp of Ammonia Synthesis Converters

In many cases it is of interest to an ammonia producer to improve the
production economics of an existing plant by the addition of new equipment or
modification of existing equipment, by optimization of the operation, for
example by installation of new and more efficient catalysts, or in some cases by a
change of feedstock, etc. An important element in such a modernization of an
ammonia plant, which is most often referred to as a revamp of the plant, is in
many cases a modernization of the converter system.
Three obvious options exist for revamp of ammonia synthesis converters (see
[590J):
- Replace the old converter by a new, larger and/or more efficient type.
- Add an extra new converter in parallel or in series with the existing converter.
- Replace or modify the internals of the existing converter with new internals to
obtain higher conversion.
The first option is mainly of interest when the existing converter is in poor
condition so that complete replacement is indicated. The two other options
require at least that the pressure shell is in good condition.
 tv
pressure shell pressure shell ---V~ff//7A ~
00
indirect quench main gas
gas inlet inlet
interbed or
interbed heat heat exchanger t!
exchanger '<
1st catalyst bed C"
5.
I\>
(I)
1st catalyst bed .
~ULYl~n~
annulus around
1""'111 \ I •• .. , 111J'..1
catalyst bed
annulus around
catalyst bed - l'lll ~ 1/ II I: 1/ IIIt-.1
2 nd catalyst bed

2nd catalyst bed RlI: / \i Iii / \:Itj cold bypass pipe

lower heat exchanger

, /\.loa....=-- gas outlet

cold bypass cold bypass

a) b)
Fig. 6.16 Two bed radial flow converter with indirect cooling between catalyst beds. Haldor Tops0e S-200 converter. a
with lower heat exchanger. b without lower heat exchanger (from [492])
 Ammonia Production Processes 249

A survey of converter revamp technologies is given in [590, 591]. The largest

number of converter revamps have been modifications of the M. W. Kellogg 3-
or 4-bed quench converters - simply because a very large number of plants using
these converter types were constructed in the 1960s and 1970s. But also Tops0e
S-lOO quench type radial flow converters, ICI and Uhde quench converters, as
well as other types of converters have been modified.
Three options for revamp of the M. W. Kellogg quench converters by
modification of the converter internals (the so-called "in-situ revamp") are
illustrated in Figs 6.17,6.18,6.19 (from [591]). M. W. Kellogg has proposed a
solution whereby the four bed quench cooled reactor configuration is changed
into a kind of two bed configuration with split flow through parallel beds in the
second conversion step and with indirect cooling between the conversion steps.
Haldor Tops0e proposes to change the internals into a "S-200 configuration", a
two-bed radial flow converter with indirect cooling between the beds. Ammonia
Casale proposes to maintain a four bed quench cooled configuration, but with
axial/radial flow. More detailed descriptions of these converter modification
technologies as applied to M. W. Kellogg quench converters and to other
converter types may be found for the M. W. Kellogg technology in [592-594],
for the Haldor Tops0e technology in [595, 596], and for the Ammonia Casale
technology in [597-602], [913]. Other converter revamp technologies are the
Uhde technology [586], which changes axial flow converters partly or com-
pletely into radial flow converters and in some cases also changes quench
cooling into indirect cooling, and the technology described in [502] where a
kind of radial flow principle is introduced into tubular cooled reactors. The

intercooler'-++-ttlil

Fig. 6.17. Four bed Kellogg quench cooled converter after

\
gas inlet
modification to Kellogg's split flow concept (from [591])
 250 Ib Dybkjaer
gas outlet

gas

Fig. 6.18. Four bed Kellogg quench cooled converter after modifica-
tion to Haldor Tops0e's 8-200 two bed radial flow concept (from
gas inlet [591])

gas outlet

Fig. 6.19. Four bed Kellogg quench cooled converter after modifica-
tion to Ammonia Casale's four bed quench cooled axial-radical flow
/' concept (from [591])
gas inlet
 Ammonia Production Processes 251

M. W. Kellogg and the Uhde revamp technologies have only been applied to
converters originally designed by M. W. Kellogg and Uhde respectively, where-
as the Haldor Tops0e and the Casale technologies have been used to modify a
number of different converter designs (see [591]).
Revamps consisting of the replacement of old converters with completely
new converters are described in [409], (c. F. Braun: in one case a complete new
synthesis loop replaced 5 identical old loops; studies are reported on the
addition of a third reactor to an existing two reactor configuration) and in [603]
(Haldor Tops0e: an S-200 converter replaced an existing M. W. Kellogg 4-bed
quench converter). Revamp by addition of an extra converter in series with the
existing converter has been proposed by Tops0e [955], and by M. W. Kellogg
using their new KAAP technology (see Sect. 6.5.3.2.3) [949].

6.4.4 Waste Heat Recovery in Ammonia Synthesis Loops

The heat of reaction in ammonia synthesis is 46.22 kJ /mol ammonia at standard

temperature and pressure or, in the units most often used in the industry,
1168 Btu/lb or 649 kcal/kg of ammonia. The efficient use of this heat, preferably
for the production of high pressure steam, is of major importance in modern
ammonia production. In almost all cases, some heat must be removed from the
converter to ensure sufficient conversion, and this heat is used either to preheat
the feed or for steam production as outlined in Sect. 6.4.3.
In early plants the effluent from the ammonia converter was often cooled
with cooling water without any recovery of heat (see [604]). However, it soon
became evident that the economics of ammonia production could be improved
by better heat recovery, and steam production and/or boiler feed water pre-
heating directly by the reactor effluent was introduced. The temperature of the
reactor effluent depends on the reactor inlet temperature for each type of
reactor. In the early 1960's it was suggested to increase the reactor inlet
temperature by heat exchange between the outlet gas from the waste heat
recovery and the reactor feed [605-607]. This corresponds to moving the lower
heat exchanger (which in many converter types exchanges heat between the feed
to the first catalyst bed and the effluent from the last bed) partly or completely to
a position outside the reactor and downstream of the waste heat recovery. In
this way the temperature at which the heat can be recovered is increased,
ultimately to a point where the inlet temperature to the waste heat recovery is
equal to the outlet temperature from the last catalyst bed (converters without
lower heat exchangers). The effect is to allow steam production at a higher
pressure. In modern large ammonia plants, steam is most often produced at
100-120 bar; the waste heat downstream from the synthesis converter is typical-
ly available between about 450°C and about 260 °C with the exact value
depending on the layout and the conversion within the converter system. This
means that the heat must be recovered partly in a boiler (down to about 350°C),
and partly in a boiler feed water preheater. In low energy plants the tendency is
252 Ib Dybkjaer

to increase the conversion by increasing the outlet concentration of ammonia

from the last catalyst bed, which requires a reduced outlet temperature (at
constant pressure). At the same time, maximum high pressure steam production
in the synthesis loop is required in order to satisfy the overall energy balance in
the plant. In such cases it may be necessary to recover part of the heat before the
reaction in the converter system has been completed. This may be accomp-
lished [425, 460, 461, 585] in a converter system with two or more converters by
cooling by steam production, not only after the last converter, but also between
the converters. It may, of course, also be done by indirect cooling by steam
production inside the converter pressure shell [554, 589], but this solution is not
preferred in modern plants because of the safety risk involved.
Synthesis loop heat exchangers, and especially the waste heat boilers, are
critical items in ammonia plants. The ammonia content and the high partial
pressure of hydrogen in the reactor effluent makes proper boiler design and
selection of construction materials essential. Several designs are available.
Reviews and examples may be found in [558, 559, 608-615, 919].
It has been proposed [616,617] to recover heat, not by converting water to
steam, but by evaporating liquid ammonia at high pressure. This could take
place inside the ammonia converter, since it does not involve the safety risks
involved in the evaporation of water in such equipment. The hot, high pressure
ammonia would be used for power production by expansion in a turbine in a
way similar to the production of power from high pressure steam. The proposed
system has, however, not been used in industrial practice.

6.4.5 Construction Materials in Ammonia Synthesis Loops

The corrosion phenomena causing the most problems in ammonia synthesis

loop equipment are hydrogen attack, nitriding, and stress corrosion. A survey of
the types of corrosion is given in [618, 619]. Steel for service at different
hydrogen partial pressures and temperature may be chosen on the basis of the
so-called Nelson-Curves [620,621]. A survey of materials problems in ammonia
plants is given in [622].
Hydrogen attacks steel by dissociation and diffusion of atomic hydrogen
into the metal. The hydrogen atoms react with carbon in the steel causing the
formation of methane. Methane cannot diffuse in steel, and very high pressure
may build up in "pockets" in the steel, ultimately causing destruction of the
material. The reaction can be counteracted by alloying the steel with carbide-
forming elements such as Cr and Mo. Hydrogen is to some extent soluble in all
construction materials. Too rapid cooling may trap the dissolved hydrogen,
whereby the material becomes brittle. Slow cooling will allow the hydrogen to
escape, and the material stays ductile. Examples of hydrogen attack are given in
[623-625].
 Ammonia Production Processes 253

Nitriding is caused by ammonia reacting with the surface of the steel forming
a hard and brittle surface layer of nitride [626,627]. It has been found by both
experience and experiments that low alloy steel can be safely used in gas
containing ammonia at temperatures below approximately 400°C, whereas
stainless steel, Inconel and Incoloy can be used at any temperature relevant for
ammonia synthesis loop equipment. Formulas for prediction of the rate of
nitriding of stainless steels in ammonia synthesis loops are given in [628].
Stress corrosion cracking may be caused by a number of reasons including
presence of trace compounds, especially chlorine, poor heat treatment after
welding, etc. Some examples of stress corrosion on loop equipment are given in
[629-633]. Stress corrosion cracking of ammonia storage tanks is dealt with in
Sect. 6.5.6.1.
The above indicates that low alloy steel, for example 21/2 Cr 1/2 Mo, can be
used for most loop equipment including the converter pressure shell. Critical
parts are the reactor internals (the "basket"), which are most often made from
stainless stell (SS 304), and the inlet part of the waste heat boiler which must
either be made from or cladded by corrosion resistant material, most often
Incoloy.

6.4.6 Product Recovery from Ammonia Synthesis Loops

Ammonia is without exception recovered as a product from synthesis units by

cooling at synthesis pressure to condensation followed by separation of the
liquid product from the gas, which is recirculated to the converter. The
arrangement of the ammonia separator(s), the recirculator, and the addition of
make-up gas and extraction of purge gas is discussed in Sect. 6.4.1.1. In high
pressure synthesis loops the cooling can be done by water cooling and/or air
cooling [634]. In modern plants, which operate at moderate pressures, this is in
most cases supplemented by refrigeration. The refrigeration is typically supplied
from a mechanical ammonia refrigeration cycle with one, two or several
refrigeration levels. Refrigeration down to - 25°C has been used which corres-
ponds to cooling by evaporation of ammonia at about atmospheric pressure.
The liquid ammonia from the separator contains a small amount of dis-
solved gases. These are partly released by pressure reduction in a "let-down
vessel" normally to about 20 bar. The gases released are normally used as fuel.
Recycle to the synthesis gas compressor has been proposed [635]. Flashing at an
intermediate pressure with recycle of the released gas to the process feed has
been used, e.g. in revamp projects [452].
After the let-down vessel the ammonia is further flashed to cool to the
temperature required for product storage. The ammonia vapour is removed
from the let-down gas by washing with water, and the gas is used for fuel, while
the ammonia water is distilled together with ammonia water obtained from
254 Ib Dybkjaer

washing of the purge gas. The ammonia vapour from the last flash is normally
sent to the refrigeration compressor.
Other fluids than ammonia have been used in the refrigeration circuit (see
[636J), and absorption refrigeration (normally based on ammonia/water) [637,
638J has been used instead of mechanical refrigeration. It has been proposed to
remove the ammonia from the partly cooled reactor effluent by absorption in
water. This procedure makes it necessary to dry the recycle gas before it reenters
the converter. The drying may be done with molecular sieves or by injection of
liquid ammonia into the gas stream and separation of the resulting ammonia -
water mixture [32, 639].
Ammonia recovery in part by adsorption of ammonia vapour on a solid
adsorbent has also been proposed [640-642]. The warm gas from heat recovery
passes through a cold adsorption vessel in which the adsorbent is saturated with
ammonia. The ammonia is des orbed and goes with the product gas to further
refrigeration cooling and product separation. The cold gas from the separator
passes through another adsorption vessel which has been heated and depleted
from ammonia by warm converter effluent. This vessel is now cooled and
ammonia is adsorbed from the separator effluent. In this way the adsorption
beds act as both heat and mass exchangers, transferring heat from the reactor
effluent to the reactor feed and at the same time ammonia from reactor feed to
reactor effluent. Both energy savings and increased production capacity are
claimed.
Separation of ammonia through semipermeable membranes has been sug-
gested. Use of a polyethylene membrane was proposed in 1954 [643J, whereas a
more recent development based on anion-exchange membranes is described in
[644,645J.

6.4.7 Purge Gas Recovery

The synthesis gas produced by any of the methods described in Sect. 6.3. will
contain certain concentrations of "inerts", i.e. compounds which are not con-
sumed by the ammonia synthesis reaction and do not interfere with the catalyst
performance. The inerts are typically methane, argon, and traces of other rare
gases. In some cases significant amounts of helium may be present originating
from He-containing natural gas.
In plants using a liquid nitrogen wash for final purification (see Sect. 6.3.5.2)
the concentration of inerts may be so low - about 100 vol ppm or less - that they
are dissolved in the liquid ammonia leaving the separator. In all other cases it is
necessary to purge a gas stream from the synthesis loop in order to prevent
excessive build up of the inerts.
The purge gas is ideally extracted at a point, where the inerts concentration
is highest and the ammonia concentration lowest, i.e. after product separation
and before make-up gas addition. However, for various reasons this is not
 Ammonia Production Processes 255

always possible, as discussed in Sect. 6.4.1.1. The purge gas will, wherever it is
taken from the loop, contain~a few percent ammonia, 10-30% inerts, mainly
methane and argon, and 60-85% of a 3 to 1 mixture of hydrogen and nitrogen.
Ammonia is normally recovered by condensation after cooling to a very low
temperature in a purge gas chiller and/or by washing with water. The remaining
gas may be used as fuel in the reformer or in auxilliary installations.
The hydrogen and nitrogen content of the purge gas represents, however,
potential raw materials for ammonia production, and energy has been invested
in the production of especially the hydrogen and in bringing the gases to the
synthesis pressure. As energy cost increased, it became increasingly attractive to
separate at least the hydrogen from the purge gas and return it to the synthesis
loop. In this way the yield of ammonia from hydrogen in the make-up gas in a
typical synthesis loop may be increased from about 94% to above 98%.
Recovery of nitrogen is less attractive, especially in plants based on steam
reforming, because preferential recovery of hydrogen will shift the required
hydrogen/nitrogen ratio in the synthesis gas to lower values, from 3.0 to in some
cases as low as 2.8. This makes it possible to transfer some of the reforming duty
from the primary to the secondary reformer (see Sect. 6.3.1 and [646, 647]),
which means a saving in the capital cost of new plants and possibilities for a
capacity increase in existing plants. The addition of purge gas recovery is an
attractive revamp option, partly because it can result in both a reduction in the
specific energy consumption and in an increased capacity, partly because in
most cases it can be done as a separate undertaking with minimum interference
to other parts of the plant.
Several types of processes have been developed for purge gas recovery. The
most important are discussed in the following. Reviews and comparisons may be
found in [648-650].

6.4.7.1 Cryogenic Separation

The inerts may be separated from the purge gas in a cryogenic unit similar to the
process used in some cases for synthesis gas purification (see Sect. 6.3.5.2). The
purge gas is washed with water and then dried over molecular sieves to prevent
freezing of water and ammonia on the heat e~changer surfaces. It is thereafter
cooled to partial condensation in a heat exchanger. Liquid and gas are
separated, and the hydrogen-rich gas is reheated by heat exchange with the
incoming gas. The liquid, which contains most of the inerts, is depressurized, and
the resulting gas is reheated by heat exchange with the incoming gas. It is further
used for regeneration of the molecular sieves in the drier, before it is used as fuel.
Refrigeration is supplied by the expansion of the off gas and, if required, by
additional chilling from the ammonia refrigeration system.
The allowable operating pressure in cryogenic purge gas recovery units
depends on the heat exchanger type. If plate heat exchangers are used, the
maximum pressure is about 90 bar. With other heat exchanger types, a higher
pressure is possible. The pressure drop on the hydrogen rich stream is marginal,
256 Ib Dybkjaer

whereas the off gas becomes available at low pressure; typical recoveries (the
fraction of a compound in the incoming gas which is found in the hydrogen rich
gas) are about 90-95% for hydrogen, 25% for nitrogen, 25% for argon, and 4%
for methane. Cryogenic purge gas recovery is discussed in [650-659]. Results
obtained in industrial plants are described in [660-662]. Argon recovery in a
cryogenic purge gas recovery unit is discussed in [663].

6.4.7.2 Separation Through Membranes

A very rapid development has taken place in the use of membranes for gas
separation. Reviews are given in [664-666]. The most popular process for purge
gas recovery using membranes is based on the use of hollow fibres, although
other types such as spiral wound membranes and stacks of flat elements have
also been used.
In the processes using hollow membranes, the feed gas, which has been
chilled and normally washed to remove ammonia to a low level, is passed to the
"shell side" of an assembly which contains a large number of hollow fibres. The
fibres pass through a "tube sheet" at one end while they are sealed at the other
end. Hydrogen will- together with smaller amounts of other compounds - pass
through the membrane and pass out on the "tube side". Hydrogen recovery and
purity depend on the membrane material, on the pressure differential, and on
space velocity, and typical values are difficult to give. Feed pressure can be high
and is often equal to synthesis pressure. It is of fundamental disadvantage of the
membrane processes that the pressure drop is on the hydrogen rich gas, while
the off gas is obtained essentially without pressure drop. In some cases two stage
processes are used so that two hydrogen rich streams are obtained, one at
. intermediate pressure and one at lower pressure. The process is, however, simple
and reliable, and it has become popular in the ammonia industry both for new
plants and for revamps. Descriptions of membrane processes may be found in
[667-670]. An example of industrial applications is given in [671].

6.4.7.3 Separation by Pressure Swing Adsorption

Pressure swing adsorption (PSA) is in many respects a rather attractive process
for hydrogen recovery in ammonia plants. High recovery of hydrogen (about
90%) can be obtained with 100% rejection of inerts, and the pressure loss
through the PSA unit is on the off gas, while the hydrogen is available
approximately at feed gas pressure.
However, the feed gas pressure is limited to about 40 bar with today's
technology so that the overall pressure loss of the hydrogen is high, and this has,
together with investment considerations, limited the utilization of the techno-
logy in the ammonia industry. Descriptions of the technology may be found in
[358-361]. A comparison between membrane separation and separation by
pressure swing adsorption is given in [672]. Cryogenic separation and PSA are
compared in [673].
 Ammonia Production Processes 257

6.4.7.4 Use of Metal Hydrides

It has also been considered to use metal hydrides for hydrogen recovery from
ammonia synthesis purge gas. Hydrogen may react with certain metals with the
formation of hydrides, and the reaction can be reversed by increasing the
temperature and/or reducing the pressure. The process has not yet found
acceptance in the ammonia industry. Description of the technology may be
found in [674, 675].

6.5 Complete Ammonia Production Processes

6.5.1 General

In the foregoing sections the individual process steps involved in the production
of ammonia from various feedstocks have been described. However, it is very
important how these "building blocks" are combined with each other, and with
the steam and power systems, to form a complete facility for the production of
ammonia. The way this is accomplished has a major impact on plant efficiency
and reliability, and much of the difference between the several ammonia
processes and much of the development in ammonia production technology
may today be found in these areas. It may be said that while "Ammonia
Technology" was in the early days of the industry most often understood as
"Ammonia Synthesis Technology" or even "Ammonia Converter and Catalyst
Technology", it is today interpreted as the complete technology involved in
transformation of the primary feedstock to the final product ammonia.
The most important factor for the choice of process layout is, of course, the
type of feedstock. Before the Second World War coke dominated (see Sect.
6.3.6). During the war several plants based on natural gas were constructed in
the USA [522], and natural gas has since then been the preferred feedstock in
the USA as well as in other parts of the world. There has, however, also been a
significant production based on partial oxidation of heavy fuel oil or gasification
of coal, especially in Europe and in countries like India and China; also naphtha
has been a preferred feedstock in some areas. During the 1970s there was, due to
the oil crises, a renewed interest, especially in the USA, in coal as feedstock for
ammonia production, but an expected major change to coal-based production
of ammonia did not materialize. Comparisons of the economics of ammonia
production from different feedstocks may be found in [160, 676-691].
Another important factor is plant capacity. Until 1960 the capacity of
individual installations had increased to a maximum of about 400 MTPD. The
units were often multi-train units meaning that several parallel trains were
installed in synthesis gas preparation and the synthesis loop not necessarily the
258 Ib Dybkjaer

same number of trains in each part of the plant. In the early 1960s technological
developments made it possible to construct large capacity single train plants,
[692-695J, and since then this has been the dominating concept. In special
situations small plants may, of course, be of interest [696-700J; but in "normal"
conditions plants with capacities below about 1000 MTPD (metric tons per day)
are rarely considered [958]. On the other hand, it seems that logistics and the
dwindling economy of scale have limited the maximum capacity to below 2000
MTPD. Today there are no technical limitations which prevent the construction
of single train ammonia plants with a larger capacity.
In the following section, energy balances in ammonia production will be
discussed. The difficulties related to the comparison of energy consumption in
different process schemes will be dealt with, and finally a description will be
given of the main concepts in the integration of ammonia production units and
steam and power systems.

6.5.2 Energy Balances

6.5.2.1 Thermodynamic Analysis

In an energy analysis a reference value for energy must be selected. Normal
practice in the industry is to select the lower (or "net") heating value, which is
generally used as a basis for pricing of fuels. (The lower heating value (LHV) of a
hydrocarbon is the enthalpy released when the compound is converted by
reaction with air to carbon dioxide and water vapour at atmospheric pressure
and 25 QC). If a production process consumes liquid water, the energy balance
can only be fulfilled if the energy for evaporation of the water is taken into
consideration. In this case, therefore, it is more correct to use the higher heating
value of the gas, where the product of the combustion is liquid water. The use of
the higher heating value (HHV) is, however, a concept which is not generally
accepted in the gas processing industry. If steam and/or power is imported or
exported to or from the production process, it should be converted to the
equivalent amounts of fuel by using standard conversion factors. Examples of
energy balances based on lower heating value may be found, for example, in [16,
246, 426]. In [246J balances bas;;:d on higher heating value are also given. The
difficulties associated with balances based on lower heating value are briefly
discussed in [701].
The balances based on a heating value are based on the first law of
thermodynamics. A more stringent analysis should, however, also take the
second law of thermodynamics into consideration. The first law simply states
that the energy is conserved, and an energy analysis using this law can, therefore,
only be used to estimate where and how the energy is converted. It cannot be
used to estimate where energy is degraded. This can, however, be shown by a
second law analysis which shows where and to what extent energy is wasted
through irreversibilities in the process. The combination of the first and second
 Ammonia Production Processes 259

laws of thermodynamics leads to a concept called ideal work, chemical avail-

ability or exergy. The latter notation will be used in the following. For literature
on Second Law Analysis and Exergy, see [702-708]. Second Law Analysis of
ammonia production is given in [703,709-711] and for CO 2 removal systems in
[291].
The first law of thermodynamics is a simple conservation balance for energy.
In a steady-state flow system it is
~H = Q - Ws (9)
where ~H is the enthalpy difference between the outlet and inlet streams, Q the
heat added to the system, and Ws the shaft work delivered by the system.
In any process requiring or delivering work, there is a limiting value called
the ideal work or maximum work, which results when all changes in the system
are carried out in a completely reversible process. This implies that all heat
transfer between the system and the surroundings is carried out reversibly and at
the temperature of the surroundings, Te. The second law of thermodynamics
then requires:
Qrev = Te~S (10)
where ~S is the entropy change in the system.
Insertion of (10) in (9) then gives the equation for the ideal work:
W ide •1 = - (~H - Te~S) (11)
The exergy is now defined as the ideal work which can be obtained from a
stream when it is converted to a standard state, an equilibrium state of the
surroundings where no more work can be extracted from it, i.e.:
(12)
where 1 is the actual state, e the equilibrium state of the surroundings, and Ex is
the exergy. The exergy is therefore also the theoretical minimum work that must
be used to produce the material from the surroundings in a steady-state flow
process.
The change in exergy from a state 1 to a state 2 is:
(13)
The change is, therefore, independent of the selected equilibrium standard state
except for the temperature Te.
It is also seen that the exergy change is equal to the change in Gibbs free
energy when the process is carried out isothermally at T e.
As an example, the differential exergy change for a gas which is heated with
no change in composition and pressure can be seen from the following equation:
dEx = d(~H) - Ted(~S)

= CpdT - T:Cp/TdT
= d(~H)* (1 - Te/T) (14)
260 Ib Dybkjaer

In this case the exergy change is consequently equal to the change in enthalpy
multiplied with the ideal Carnot efficiency of a heat engine. The exergy change is,
therefore, small when all heat is transferred to or from the surroundings at a
temperature close to Te.
The maximum work a system can perform is found by combining Eqs. (11)
and (13).
Wide a] = - ~Ex (15)
i.e. the maximum work is equal to the negative value of the exergy change, which
is the sum of the exergies of the leaving streams minus the sum of the exergies of
the entering streams. The lost work is then found as the difference between the
ideal work the system can perform and the actual work it delivers. In the actual
work should also be included work which could be extracted from useful heat
transferred to or from the system by using an ideal Carnot engine. The lost work
is then:
(16)
where Ts is the temperature of the source delivering the heat Qs to the system.
In processes with chemical reactions it is necessary, if one is interested in
absolute values of the exergy, to define a convenient standard state where, in
principle, the compound can do no more work. Different standard states have
been defined in the literature depending on how many 'spheres' (atmosphere,
hydrosphere, lithosphere) are included in the analysis.
In gas processing systems such as the production of ammonia from natural
gas, liquid water and air at 25°C and atmospheric pressure are normally
selected as standard states for the environment. This implies that these raw
-materials are assigned zero exergies. This may not always be realistic, in
particular in places where clean water is expensive to provide.
There is also the problem of carbon dioxide which is only present in a very
small amount in the atmosphere. It can, therefore, be argued whether it is correct
to assign a value of the exergy of carbon dioxide equal to the theoretical
minimum work to extract it from the atmosphere.
For this reason, some authors prefer to define the standard state for each
pure component at atmospheric pressure, and some prefer to assign a partial
pressure to carbon dioxide as it exists in a typical flue gas. The difference
between different standard states is mainly the value assigned to air. For other
gas processing streams the difference in exergy is less than 1-2%.
In the following, the standard state selected for each component in air is the
partial pressure in dry air at atmospheric pressure. The standard state for water
is liquid at 1 atmosphere. The temperature of the surroundings is 25°C. The
exergy of a stream of arbitrary composition can now be calculated by use of
Eq. (12).
This is done in two steps. The first contribution EX 25 is the Gibbs energy of
formation of the mixture from the compounds in the surroundings and consider-
 Ammonia Production Processes 261

ing the work required to transfer the individual reactants and products to 1 atm
partial pressure, which is the standard state for the Gibbs energy.
The second contribution Ex is the transfer of the compounds in the mixture
from 1 atm partial pressure at 25°C to the actual temperature and the actual
partial pressures in the mixture. This is done by use of Eq. (13), where the
enthalpy and entropy differences are calculated by standard methods. An
example of results obtained by exergy analysis of an ammonia production
process is given in Sect. 6.5.3.1.2.

6.5.2.2 Comparison of Energy Consumption for Different Cases

Ammonia can be produced from many different feed materials and with different
process design. In order to compare and study all these cases, it is necessary to
have a common basis for comparing the energy consumption.

6.5.2.2.1 Theoretical Minimum Energy Consumption

In all processes there is a theoretical minimum energy consumption which is
equal to the chemical 'value' of the pure ammonia.
The simplest raw materials are air and liquid water, which are both present
in sufficient quantities, and the 'value' of the ammonia can then be expressed as
the ideal chemical work that can be obtained when ammonia is converted to air
and water. This work is the exergy of ammonia.
For gaseous ammonia at 25°C and 1 atm, and for liquid ammonia at
-- 33°C, 1 atm, the exergies in Gcal/MT (MT = metric tons) NH3 are shown in
Table 6.2 together with the higher and lower heating values, which are the first
law "values" of ammonia.
It is interesting to note that whereas the heating values both drop when
ammonia is condensed, this is not the case for the exergy. The reason is that the
exergy also considers the entropy change. It is also seen that the exergy lies
between the higher and lower heating values; the reason being that the amount
of work obtained by condensation of water is much smaller than the corres-
ponding enthalpy change.

Table 6.2. Energy value of ammonia, Gcal/MT

Exergy HHV LHV

Gaseous 4.73 5.37 4.44

25°C, 1 atm
Liquid 4.81 5.01 4.08
- 33°C, 1 atm
262 Ib Dybkjaer

6.5.2.2.2 Influence of External Conditions

Even when the thermodynamic basis has been defined, it may be quite difficult
to compare the energy consumption for different process schemes on the basis of
information available in the literature. The only safe way is to prepare, on the
same basis and using the same calculation procedures, complete mass and
energy balances for the different schemes within well defined battery limits. This
will obviously not always be possible - due to lack of information about the
processes or due to lack of resources to do the calculations - and it may
therefore not be possible to compare figures gives in the literature for different
cases. There are, however, a number of external factors which should in any case
be considered, and which may have a significant influence on the calculated
energy consumption. Important among these are the following.
Cooling Water Temperature. The temperature of the cooling water has an
influence on the power consumption for air compression and make-up gas
compression and more importantly on the power required for the refrigeration
compressor because a larger amount of the ammonia is condensed in the water
cooler when the cooling water temperature is low. The overall effect in a natural
gas based plant is shown in Fig. 6.20 (from [701]). It is seen that the effect is
quite significant, especially at low cooling water temperature. However, it
should be noted that a significant part of the effect is due to a reduced
temperature and, therefore, a reduced pressure in the turbine condensers. This
effect mayor may not be obtained in the specific case depending on the choice of
type of turbines and turbine condensers.
Feed Composition. It is evident that the type of feedstock has a major
influence on energy consumption in an ammonia plant. But even for the same
type of feedstock, differences in composition may have an effect. As an example,
consider natural gas, which can be characterized by its content of "bulk"
impurities (most often carbon dioxide and/or nitrogen), "trace" impurities

c
o
'5. M 0.3r--r-----r-------,-------,
EJ:
::Jz
Ill ....
C :::E
8 ... 0.2
618
liil!>
C -

=c c3\
<II
u
0.1

Fig. 6.20. Influence of cooling wa-

lii 3\
~ .8 o ter temperature on overall energy
.£ E consumption. Basis Haldor
<II 0
Cl ....
Tops0e's low energy ammonia pro-
c - - 0.1'----:'-:-----'-----::'-::----'----' cess (from [701])
.E
u
20 30 40
Cooling water temperature (DC)
 Ammonia Production Processes 263

(sulfur compounds, etc.), and by the mix of hydrocarbons which can be

expressed as the hydrogen to carbon ratio of the gas.

- Trace impurities will normally have no influence on the energy consumption.

They may be catalyst poisons in which case they must be removed carefully
before the gas is introduced in the process.
- Carbon dioxide may be present in certain natural gases in significant concen-
trations. If the concentration is very high, it will normally be reduced in a
separate purification step before the ammonia process proper. But natural gas
with up to 15-20% carbon dioxide can be used without problems as feed gas
for an ammonia plant. The presence of such amounts of carbon dioxide will
have an effect on the operation in several sectors of the plant, but the effect on
overall energy consumption is modest, especially when a physical carbon
dioxide removal process is used. A typical increase in energy consumption
caused by the presence of 10% carbon dioxide is about 0.05 Gcal/MT
ammonia.
- Nitrogen could, in contrast to carbon dioxide, be regarded as a desirable
impurity since it is a component required for the synthesis of ammonia. It has,
however, only a very limited influence on energy consumption. A concentra-
tion of 10 vol % nitrogen in methane will cause a decrease in the overall
energy consumption of about 0.02 Gcal/MT ammonia and a slight change in
the operating conditions in the reformer due to the decrease in the amount of
process air.
- The hydrogen to carbon ratio of the gas has an influence on the overall
stoichiometry of the theoretical synthesis reaction. When the hydrogen to
carbon ratio decreases, more water is consumed in the reaction and more
-carbon dioxide is produced together with the ammonia (see Section 6.5.4.1).
The effect on the energy consumption of the process is modest. For pure
n-butane as a feed, the consumption expressed as the lower heating value will
increase by about 0.1 Gcal/MT of ammonia. If the higher heating value is
considered, the same consumption corresponds to a decrease in energy
consumption of about 0.07 Gcal/MT of ammonia compared to a case with a
pure methane feed. The hydrogen to carbon ratio of the feed will also have an
influence on the acceptable feed preheat temperature due to the risk of steam
cracking in the preheat coil when higher hydrocarbons are present.

Ammonia Supply Temperature. A part of the energy consumption in am-

monia plants is used to condense the product ammonia. If the ammonia could
be supplied as a gas, this energy could be saved. On the other hand, extra energy
must be used if the product must be supplied to an atmospheric storage as liquid
at - 33°C. The differences compared to supply of liquid ammonia at ambient
temperature are:
- Ammonia supplied as gas at 3 atm, savings: 0.14 Gcal/MT
- Ammonia supplied as liquid at - 33°C, extra consumption: depends on
cooling water temperature; 0.07 Gcal/MT with cooling water at 30°C.
264 Ib Dybkjaer

6.5.2.3 Steam and Power System

Fundamentally an ammonia production process is a combination of end-
othermic and exothermic processes. Heat is required at high temperature for the
initial conversion of feedstock, e.g. by steam reforming; this heat is supplied by
external combustion of fuel in the primary reformer and by internal combustion
in the secondary reformer (or by internal combustion alone in partial oxidation
or gasification processes).
Heat is also required for the generation of process steam, for the preheat of
process streams, and in most cases for the regeneration of solvent in a carbon
dioxide removal unit.
Furthermore, motive power is required for compression of feed streams and
make-up gas for the synthesis loop, for recovery and refrigeration of product
ammonia, and for certain other purposes, mainly pumps and blowers.
Heat is generated at various temperature levels by exothermic processes,
notably in gasifiers or secondary reformers, by shift conversion, and by the
ammonia synthesis reaction. Heat at high temperature is also available in the
flue gas from the primary reformer.
In the design of an energy efficient ammonia production process a number of
considerations are relevant:
Efficient use of the feedstock must be ensured. In modern natural gas based
plants with recovery of hydrogen from the loop purge gas, ammonia yields
above 95% of the theoretically possible (corresponding to stoichiometric
conversion of feedstock with air and steam to ammonia, see Eq. (17), Sect.
6.5.3.1.1) may be achieved.
- Non-converted feedstock must be collected and used as fuel with the highest
possible efficiency.
- Excessive firing in primary reformers and/or preheaters must be avoided by
efficient equipment design and by maximum use of heat recovery for pre-
heating of feed streams and, in some cases, combustion air.
- Consumption of power for compressors, pumps etc. must be minimized by the
choice of optimized operating conditions and by the choice of efficient
machines including drivers. In all large modern plants this involves use of
centrifugal compressors driven by stream turbines or gas turbines.
- Heat available in various process streams must be recovered to the maximum
possible extent for useful purposes. In all large modern plants this involves
recovery for the production of high pressure steam.
- The release of waste heat must be adjusted to various temperature levels as
required by the steam generation, so that matching amounts of heat are
available for preheating boiler feed water, steam generation, and steam
superheating.

The above points indicate that the steam and power balance of an ammoma
plant is of major importance for the overall efficiency.
 Ammonia Production Processes 265

The concept of recovering heat for the production of high pressure steam
and using the steam for driving compressors via steam turbines in plants based
on steam reforming was developed during tne 1960s [426, 712-714]. Descrip-
tions of systems for heat recovery by high pressure steam production may be
found in [709, 715-718]. See also Sect. 6.4.4. A discussion of the energy and
money value of waste heat and steam is given in [719].
An example of a steam generation system in a large, modern, natural gas
based ammonia plant is shown in Fig. 21 and 22 (from (720}). Fig. 6.21 shows
how heat is recovered for preheating boiler feed water and for generating and
superheating high pressure (110 bar, 510°C) steam. The total heat recovery in
the steam production system is 3.00 Gcal/MT of ammonia corresponding to
4.52 tons of high pressure superheated steam per ton ammonia. It could be
possible by adjusting the design - for example by introducing combustion air
preheat - to reduce the heat available for steam production to about 2.4 Gcal
corresponding to about 3.6 tons of steam per ton ammonia. (It is of course
possible by extra firing to increase the steam production to any desired level).
Fig 6.22 shows how low grade heat is recovered in the same plant; as much
as possible - in this case slightly above 0.4 Gcal/MT of ammonia - is used to
preheat demineralized water before deaeration. Other uses of low grade heat
include regeneration of solvent in the carbon dioxide removal unit and produc-
tion of low pressure steam which is used in deaeration and in the carbon dioxide
removal unit. It may also be noted that condensate stripping is in this plant
integrated with deaeration and carbon dioxide removal. General descriptions of
process condensate treatment are given in [721, 722, 920].
Fig 6.23 shows schematically how steam is used in the same plant. Super-
heated high pressure steam from the process units generated as shown in
Fig. 6.21 is passed through the back pressure part of the turbine driving the
synthesis gas compressor; part of the steam is condensed in the condensing part
to provide sufficient power, while the main part is extracted as medium pressure
steam. The medium pressure steam is partly used for process steam, partly for
other turbines driving the air compressor, refrigeration compressor, and flue gas
fan. Low pressure steam to balance the needs in carbon dioxide removal, boiler
feed water deaeration, and other uses is generated in the back pressure turbine
driving the flue gas fan and extracted from the turbine driving the air com-
pressor. Total power production in the four large turbines is about 23 MW in
this 1000 MTPD ammonia plant.
Figs. 6.21, 6.22, and 6.23 show only one possible layout of the steam system
in large ammonia plants. Other similar systems are described in [31, 615, 718,
723, 724, 949]. Other types of drivers for ammonia plant compressors than steam
turbines are discussed in [398] (electrical motors) and [399, 400] (gas turbines).
Steam and power production for ammonia plants in cogeneration units is
described in [725-727,921]. Steam and power balances in ammonia plants and
ammonia-urea complexes are discussed in [959] with special reference to the use
of gas turbines.
 t..)
superheater superheater a,
a,
BFW preheater downstream in reformer
downstream HTS sec. ref. WHS

superheated HP steam ~
I •
at 110 at. 510°C U
'<
r::r
.E.
~
>-;

BFW from BFW pump !1¥J •

BFW preheater BFW preheater
downstream LT5 in loop
boiler downstream sec. ref.

BFW preheater
in reformer WH5

boiler downstream HTS

boiler in loop

Fig. 6.21. Steam production in a large, modern natural gas based ammonia plant (from [720])
DMW from polisher condenser

2nd separator

condensate from
3 rd separator
excess condensate
from Benfield

lP steam from grid reformer ~

feed and fuel 3
o
preheater ::I
CW stripped s·
condensate
to polisher ~
o
- 0-
s:l.
'"
o·
::I
HP BFW to preheater
..,~
o
Fig. 6.22. Recovery and use of low grade heat in a large, modern natural gas based ammonia plant (same plant as Fig. 21) (from [720]) g
~
&'l

N
~
.....
 MP steam grid IV
HP steam from process 0-
00
110 at. 510°C 38at.375°C process steam
MP steam
to consumers 0=
MP steam t:I
'<
export cr"
2r.
~
....

synthesis gas BFW pump steam to urea unit

compressor driver -1 MW 22 at. 310°C
driver -11 MW
LP steam
export

CW LP steam
to consumers

turbine condensate

Fig. 6.23. Use of high pressure steam in a large modern natural gas based ammonia plant (same plant as Figs. 21 and 22)
 Ammonia Production Processes 269

6.5.3 Processes Based on Steam Reforming of Light Hydrocarbons

6.5.3.1 Theoretical Considerations

6.5.3.1.1 Stoichiometric Production of Ammonia from Methane
Consider the production of pure liquid ammonia and pure carbon dioxide from
pure methane, pure liquid water, and dry air. By pure in the context is meant
that the component is present in its pure form at 1 atm partial pressure.
The reaction scheme may be written as:
0.4422 CH 4 + 0.6155 H 20(t) + 0.6407 (0.2099 O 2 + 0.7804 N2
+ 0.0003 CO 2 + 0.0094 Ar) = NH 3 (t, - 33°C)
+ 0.4424 CO 2 + 0.0060Ar (17)
The thermodynamic properties of the components are given in Table 6.3.
It can be seen from the table that the exergy change of the reaction is much
smaller than the corresponding enthalpy change. This is mainly due to the
difference caused by condensation of ammonia. It also appears from the table
that it is necessary to assign a negative heating value to the liquid water if the
heat balance is to be fulfilled in the case using the lower heating value.

6.5.3.1.2 Energy Analysis of a Low Energy Ammonia Process

Afirst law analysis based on both lower and higher heating values was reported
in [246].
This analysis was repeated in a slightly updated version for one of the cases,
i.e. for a low energy process with a net energy consumption (lower heating value)
of 7.01 Gcal/MT of ammonia, and an exergy analysis was also made. The
process concept is an "optimized conventional" scheme with the following
features:
- Reforming at a steam to carbon ratio of 2.5 with combustion air preheat.
- Shift conversion in two stages with copper based catalysts.
- Carbon dioxide removal with the Selexol process.
- Final purification by methanation.

Table 6.3. Thermodynamic values of components in ammonia production (Gcal per ton
Ammonia).

CH 4 H 2 0(t) Air NH 3 (t, - 33) CO 2 Ar ~React

Exergy 5.15 0 0 4.81 0.12 0 - 0.22

LHV 4.98 - 0.38 0 4.08 0 0 - 0.52
HHV 5.53 0 0 5.01 0 0 - 0.52
270 Ib Dybkjaer

~ Synthesis at 140 kg/cm 2 g using a two bed radial flow converter with indirect
cooling.
Use of centrifugal compressors driven by steam turbines.
The energy balance was based on the following assumptions:
~ Feed and fuel: 100% methane available at the required pressure.
Cooling water available at 30 DC at the required pressure.
Ambient conditions: 30 DC, 65% relative humidity, 1 atm.
Product ammonia supplied at - 33°C to atmospheric storage.
No import or export of steam or power across ammonia plant battery limits.
The results of the analysis are summarized in Table 4.
Losses are calculated as the energy content (heating value or exergy) of the
ingoing streams minus the content in the outgoing streams including in- or
outgoing work. The total loss, therefore, includes the heat loss to the surro-
undings and the heat loss to the cooling water. For the exergy, these two losses
are in most cases much smaller than the losses due to the irreversibilities

Table 6.4. Energy analysis, low energy ammonia plant (Gcal/MT Ammonia)

HHV LHV Exergy

Natural Gas Consumption

Reformer Feed 5.89 5.32 5.56
- Fuel 1.79 1.62 1.69
Fuel, Aux. Boiler 0.08 0.07 0.08
Total Consumption 7.76 7.01 7.33
Energy to HP Steam System
Primary Reformer 0.20 0.20 0.11
Secondary Reformer 1.06 1.06 0.53
Shift CO 2 -removal 0.73 0.73 0.32
Synthesis 0.61 0.61 0.29
Aux. Boiler-Misc. Recovery 0.17 0.17 0.07
Total HP Steam Production 2.77 2.77 1.32
Steam Use
Steam for Turbines 1.56 1.56 0.84
Process Steam 0.97 0.97 0.35
Various Consumers 0.24 0.24 0.13
Total Steam Use 2.77 2.77 1.32
Losses
Reforming 0.09 0.09 1.18
Steam Generation 0.08 0.08 0.57
Shift + CO 2 -removal + Methanator 0.31 0.31 0.16*
Synthesis 0.41 0.41 0.37
Turbines + Compressor 1.56 1.56 0.13
Misc. Losses 0.08 0.08 0.05
Stack 0.23 0.08 0.06
Total Losses 2.76 2.61 2.52
Energy in NH3 Product 5.01 4.09 4.81

* Mainly C02 Product

 Ammonia Production Processes 271

produced in the process. The reason is, of course, that the heat loss is multiplied
by the Carnot efficiency.
The largest transfer of heat to the cooling water occurs in the steam system,
where the cooling needed to condense the exit streams from the turbines and for
cooling in compressor interstage coolers is equal to 1.56 GcaljMT. But since
most of the heat is transferred at about 50 DC, the Carnot efficiency is only 8%,
and the exergy loss to the cooling water is only 0.13 Gcal/MT.
The heat loss from the primary reformer occurs at a high temperature, hence
giving a high Carnot efficiency. The exergy value is best evaluated by consider-
ing the extra fuel necessary to compensate for the heat loss. This is equivalent to
about 5% of the total exergy loss from the reforming section or about 3% of the
heating value of the total fuel to the reformer.
Almost 70% of the total exergy loss is lost in reforming section and the
associated steam generation, mainly due to irreversibilities present in the
combustion.
This result can be compared with an ordinary energy or first law analysis,
which shows that almost all energy is transferred to the cooling water.
The exergy analysis can be used to calculate a thermodynamic efficiency for
production of ammonia. It is simply the ratio between the exergy of product and
the total feed which is:
Efficiency = exergy ratio = 4.81/7.33 = 0.66 or 66%

6.5.3.2 Process Schemes

6.5.3.2.1 Early Developments
The overall process scheme for production of ammonia from light hydrocar-
bons, i.e. natural gas, LPG, or naphtha, has changed little since the early days of
the industry. In most cases the plants have comprised the following process
steps:
- feed purification
- primary and secondary reforming
- shift conversion
- CO 2 -removal
- final purification of synthesis gas
- compression and ammonia synthesis
- ammonia recovery and refrigeration
but there has been, of course, a significant development within each of the
process steps, and in addition a dramatic increase has been seen in the capacity
of the units and in the integration ofthe process units with the steam and power
system.
The development in design of natural gas plants from the Second World War
up to 1954 is summarized in [25] and in [141]. The developments are sum-
marized in Fig. 6.24 and Table 6.5 (adapted from [25]).
 tv
~ ~

Feed
St -I p.
nm. ......
cr
= ~ 0
'<
Pressure, barg ~
Temp., ·C 'j;;.
Carbon Oxides 1.5-2.0% CO 0.Q1-1% COz 5 ppm ~
Compressor Stages 2 2 1
Air

Feed
2 Steam : Prim. Sec. r--1J~~v.
Ref. Ref. (HTS)
1 H
Pressure, barg 4.2 3.2
Temp., ·C 770 930
Carbon Oxides 1.7% CO 0.01% COz 5 ppm
Compressor Stages 1 2 1
Air

3 SFeed :1· H Pnm. Sec. H CO-

Cony CO z- Pur·f NH3
team Ref. Ref. (HTS) M Rem. ~ (rn~i;n \ ~ Synt.
Pressure, barg 8.8 7.8
Temp., ·C 790-800 950-960
Carbon Oxides 1.4-1.7% CO 0.01% COz 5 ppm
Com pressor Stages 1 2 1
Air

4 Feed __ , . I I Final
Steam Pnm. ~ Sec. Purif. Lr...-I NH3
(Meth.) 1..:r11 Synt.
Pressure, barg 14.5-18.4
Temp., ·C 800-820
Carbon Oxides 0.8-1.0% CO 0.01% COz 5 ppm
Compressor Stages 3
 Air

Feed
5
Steam

Pressure, barg 21-28 20-27 25 23 330

Temp., ·C 790-850 950-980
Carbon Oxides 05% CO 0.01% CO 2 5 ppm
Compressor Stages 3
Air

Feed
6
Steam

Pressure, barg
Temp., ·C
Carbon Oxides 0.5% CO 0.01% CO2 5 ppm
Compressor Stages 4 +2
Air

Feed
7
Steam

Pressure, barg
i
o
i:l
Temp., ·C 0;'
Carbon Oxides 0.5% CO 0.01-0.1% CO2 5 ppm '"c:I
....
Com pressor Stages 2 to 220 bar o
0-
3 to 330 bar
ao·
Fig. 6.24. Some steps in the development of ammonia process layout between 1940 and 1972 (efr. also Table 5) (adapted from [25]). i:l
'"c:I
....
o
g
'"~
'"
tv
-.l
W
 N
....,
""'"

5i
t:l
'<
cr'
5.
%
....
Table 6.5. Comments to Fig. 6.24

No. Year Capacity COr Removal No.ofNH 3 Process Features and Improvements
short tons/day Converters

1940-52 150 Water wash 3 - low pressure steam reforming

20% MEA - poor heat recovery
- high energy consumption
2 1953 162 20% MEA 2 - first increase of reforming pressure
- significant reduction of power consumption
- deletion of gasometer and associated equipment
3 1955 300 20% MEA - further increase of reforming pressure
hot potassium carbo - reduction of raw material and power consumption
vetrocoke - first introduction of turbocompressor for com-
pression of raw synthesis gas
- first single stream 300 tons/day unit
- use of raw materials such as refinery off gas and
naphtha
- use of hot potassium carbonate wash for CO 2
removal
4 1960-62 360 20% MEA - further increase of reforming pressure, improved
tube materials
- Cu liquor wash replaced by methanation
- deletion of compression between gas preparation
and purification
- preheat of process air
- modification of reformer for reduction of heat loss
- improved heat recovery at all catalytic process steps
5 1963 320 20% MEA - increase of reforming pressure
- high heat recovery
- reduced fuel consumption
- improved LT shift catalyst
- reduction of purge from synthesis loop
6 1964 600 20% MEA - tirst 600 tons/day single stream unit
- tirst use of turbocompressor for synthesis gas com-
pression to 150 atm
- about 80% of power for synthesis gas compression
by turbocompressor
7 1965-72 600-1700 20% MEA - further increase of reforming pressure combined
Vetrocoke with low pressure synthesis
Catacarb - all synthesis gas compression up to 200 atm by one
Carsol two-stage turbocompressor, to 300 atm by three-
TEA/MEA stage machine
Sultinol - introduction of steam turbines for compressor and
pump drivers. Production of high pressure steam
- reduction of electric power consumption (in some
cases to practically zero)
-low (for the time) raw material and operation costs
- high preheat temperature of process air in spite of
high reforming pressure
>
3
- tirst single stream 1700 tons/day unit 3o
- improved converter construction gives lower pre- i:l
;;;.
ssure drop in synthesis loop '"t:I
....
o
0-
"~
o·
i:l
'"t:I
....
o(")
Dl
'"
'""

tl
'-"
276 Ib Dybkjaer

Descriptions of plants designed in the various periods may be found in the

following references:
1940-1952: [139, 522, 728-730]
1952-1960: [143,433, 731-734]
1960-1965: [144, 145,735]
1965-1972: [426,430-432, 736-738]
The "end product" of the development described above (and in more detail
in [25]) was, as mentioned, the large capacity single stream ammonia plant.
Fig. 6.25 from [430] shows a simplified process flow diagram for an am-
monia plant of this type. This process design - by M. W. Kellogg - was used in a
major part of the very rapid expansion of the ammonia industry in the late 1960s
and early 1970s. A detailed description of the process including composition of
major process streams is given in [739].

6.5.3.2.2 Recent Developments

A major concern of process licensors, engineering contractors, and plant owners,
especially since the changes in energy price levels in 1973 and thereafter, has
been to develop process schemes with better energy efficiency than the above
mentioned classical process.
The effort has been quite successful; the overall energy consumption has
been reduced from about 10 Gcal/MT of ammonia in the earliest large capacity
single train units to less than 7 Gcal/MT in the most modern plants. Energy
saving developments are discussed e.g. in [30-32, 35, 38, 40, 740, 741, 954].
The most important improvements and modifications which have been
introduced in the individual process steps as compared to the classical process
scheme are:
- In feedstock purification - mainly desulfurization - the development was from
adsorption of sulfur-compounds on active carbon to catalytic hydrogenation
of sulfur-compounds followed by adsorption of hydrogen sulfide on hot zinc
oxide. This concept has been of major importance for the development of
processes based on reforming at severe conditions including reforming of
naphtha. See also Sect. 6.3.1.
Feedstock purification as such has no influence on the energy consumption.
But efficient purification is a prerequisite for other energy saving measures,
especially in reforming and shift conversion.
- In the reforming section, which is the most critical section of the synthesis gas
preparation (the "front end"), the most important developments have been the
increase in operating pressure, decreased steam addition, use of more sophisti-
cated alloys for reformer tubes, increasing exit temperatures from the primary
and secondary reformer, increasing preheat temperatures for feed, process air,
and combustion air, and the use of catalysts with higher activity and improved
resistance to poisoning and coking. See also Sect. 6.3.1.
 primary reformer
natural gas-------~
boiler feedwoter steam

fuel

I A ..... 1\ 1\ J\ .... I
VVv""v'4

steam "
a i r - - - -........- - - - - j

heat recovery
heat recovery
water

NH3 converter
----.-----.
:>
§o
::>
0.;'
'"0
...,
o
0-
~
o·
::>

~g
'"
ri
Fig. 6.25. The classical single-train M. W. Kellogg ammonia process. Simplified process flow diagram (from [430])
'"
N
...,
...,
278 Ib Dybkjaer

Reforming at higher pressure saves energy due to reduced power requirements

for synthesis gas compression. Reduction of the steam to carbon ratio also
saves energy due to savings in process steam consumption. In both cases, more
severe conditions are required in the reformer to obtain the required high
conversion of the feedstock. This has led to the requirement for better catalysts
and better materials for the reformer tubes.
High preheat temperatures for feed and combustion air lead to reduced fuel
consumption and thereby to savings, especially when the steam production in
the plant must be minimized. Reduction of the flue gas stack temperature
reduces the heat loss to the atmosphere. A similar effect has, perhaps more
importantly, been obtained by improved insulation in the reformer.
In some processes excess air is added to the secondary reformer. This decreases
the duty and the temperature level in the primary reformer, so that firing is
reduced. But the energy required to compress the process air is increased, and
an extra process step becomes necessary to remove excess nitrogen from the
synthesis gas.
- In shift conversion, only one stage-high temperature shift conversion-was
used in early plants. In the early 1960s two stage shift units were introduced,
originally with two stages of high temperature shift.
Later low temperature shift catalysts were developed. Since then more efficient
catalysts have facilitated operation at the reduced steam content and the
correspondingly reduced temperature required to obtain high conversion at
the modified conditions. The risk of byproduct formation at a low steam to
dry gas ratio has led to the development of new types of catalysts which avoid
these problems. See also Sect. 6.3.3.
- In carbon dioxide removal, the water wash was very early replaced with more
efficient systems using MEA or hot potassium carbonate as the absorbent.
Later, the efficiency of these processes was gradually improved, and also new
processes based on physical or combined physical and chemical absorption
were developed. See also Sect. 6.3.4.
The energy requirements have been reduced in part by the introduction of
better solvents which reduces both the power requirement for solvent
recirculation and the heat requirement for regeneration. Savings have also
been obtained by process modifications which have improved the heat
recovery from the process streams. The effect of this has been to reduce the
requirement for low grade heat which has mainly been supplied by condensa-
tion of excess process steam. Therefore, the improvements in carbon dioxide
removal have further added to the advantages obtained by reforming at
reduced steam to carbon ratio.

- Final purification was in the early plants most often done with a copper
liquor wash. With the advent of the low temperature shift, it became feasible
to do the final purification by methanation, and since then this has been the
preferred method. See also Sect. 6.3.5.
 Ammonia Production Processes 279

In Gases where excess air is used in the secondary reformer, it is necessary to adjust
the composition of the synthesis gas by the removal of excess nitrogen. This has
been done in a cryogenic purification unit or in a Pressure Swing Adsorption
unit. See also Sect. 6.3.5.
In the ammonia synthesis, improved reactor design, increased converter
volume, and to some extent improved catalysts have improved the conversion
efficiency. Improved heat recovery has been introduced with the production
of high pressure steam instead of preheating of boiler feed water, and single
train capacity has been dramatically increased. Drying of make-up synthesis
gas and recovery of hydrogen from the purge gas have been introduced to
improve performance. See also Sects. 6.4.3,6.4.4 and 6.4.7, where the improve-
ments in loop and converter design have been extensively reviewed.
In the steam and power system, developments in compressor and turbine
technology have led to significant energy savings. The use of gas turbines as
direct drivers for compressors or as part of cogeneration units producing
steam and power has also been introduced. Heat recovery has been improved
by increasing the steam pressure, by introducing high pressure steam produc-
tion in the synthesis loop, and by combining steam production and steam
superheating downstream of the secondary reformer. Production of process
steam from low level heat by feed gas saturation has been used in some cases,
and stripping of process condensate with low or medium pressure steam for
reuse as boiler feed water has been introduced. See also Sects. 6.3.7,6.4.4, and
6.5.2.3.
Developments in instrumentation and computer science have led to the
increased use of advanced control systems and computerized optimization of
plant operations. Advanced control systems may be used to keep key
operating parameters (e.g. steam to carbon ratio at inlet reformer, reformer
exit temperature, hydrogen to nitrogen ratio of synthesis gas, synthesis loop
pressure, etc.) constant in spite of variations in feedstock properties or other
external changes [742-750, 922, 965]. The control systems can be in open
loop, where the computer system evaluates data collected by the plant
instrumentation and gives new set points to the plant operator who then
manually changes the controls. The system can also be in closed loop where
the set points are directly controlled and changed by the computer to
optimum values calculated in a computer model on the basis of operating
data and external (economic) data.
Computer simulation of plant operation may also be made off line real time using
computer models of the complete installation, which are capable of simulating
dynamic plant response to changes in operating parameters, plant upsets, etc.
Such systems may be used for off line optimization studies and for operator
training in handling emegencies, start-up- and shut-down situations, etc. Without
risk to plant or personnel. Simulators are described in [751-758, 923].
280 Ib Dybkjaer

6.5.3.2.3 Commercial Processes

A modern ammonia plant consists, as described in the previous sections, of a
combination of several more or less independent sections each with their own
options for choosing basic technology and for optimizing process layout and
operating parameters. The value of the plant and of the product is very
significant and certainly large enough to justify individual design and optimiza-
tion in each case. This has lead to a situation where all plants are different, and
where it may be difficult to give a general description of what is called an
"ammonia process".
A limited number of companies possess the basic knowledge and the
experience which makes them able to prepare the basic engineering
information-process flow diagrams with heat and mass balances, equipment
specifications, specification of instrumentation and process control including
safety precautions, operating manuals etc. -which is required for the design of
an ammonia plant. These companies have also developed their own proprietary
design of critical equipment, most often the ammonia synthesis converter and
the primary reformer.
The sum of the basic engineering information and the detailed design of
proprietary equipment may then be said to represent an "Ammonia Techno-
logy", which is made available to others under a license agreement. The basic
information is then further developed to a detailed engineering specification by
an engineering contractor, which may be the process licensor or another
company. A discussion of the contribution of the engineering contractor to the
plant design may be found in [36].
In the past, many different companies have offered ammonia technology to
the market. The competitive situation has, however, been such that today (1992)
only five licensors of ammonia technology are active, namely Braun (original
name: C. F. Braun, later Santa Fe Braun, and most recently Brown & Root
Braun), ICI, M. W. Kellogg, Haldor Tops0e AjS, and Uhde GmbH.
In the following, each of these commercially important technologies will be
discussed in some detail. In addition, a somewhat more limited description will be
given of other process schemes which have less commercial importance today. The
discussion will be limited to "stand alone ammonia processes", i.e. processes
producing ammonia without special considerations for production of other
products. Simultaneous production of ammonia and other products, especially the
simultaneous production of ammonia and a stoichiometric amount of carbon
dioxide to convert all of the ammonia to urea, is treated in Sect. 6.5.4.

The Braun Purifier process [28, 32, 33, 37, 324-330,409-412, 759, 952] has
been available for more than 20 years. A simplified process flow scheme of the
updated version is shown in Fig. 6.26 (from [330]). Key features of the process
are:

- Reforming with a large ( ~ 50%) excess of air. Steam to carbon ratio at inlet
of the primary reformer approximately 2.8. Relatively high methane leakage
( ~ 1% dry) from the secondary reformer.
 gas turbine
, . primary reformer C02 CO 2 removal
, aIr compressOr r.:-:l secondary
alr--, .--. ~ reformer

I I

steam I
LL::::r:J • I

water
methanator

•• cryogenic purifier

:>
ammonia product
~
::s
iii'
..,'"0
o
Po
§
g.
Fig. 6.26. The Braun purifier process, simplified process flow diagram (from [330]) ::s

~
~
~
'"
to..>
00
......
282 Ib Dybkjaer

-. Air compressor driven directly by gas turbine. Exhaust from the gas turbine is
used as combustion air in the primary reformer.
- Conventional high temperature and low temperature shift conversion.
- Carbon dioxide removal, normally by BASF's MDEA process.
- Final purification by methanation followed by adjustment of synthesis gas
composition (removal of excess nitrogen and part of the inerts) in a cryogenic
unit (referred to as the Braun purifier). Gas drying upstream of the purifier.
- Synthesis of ammonia in a "dry" synthesis loop (with addition of make-up gas
after the ammonia separator) at approximately 180 kg/cm 2 g with the Braun
converter system consisting of two or three single-bed, adiabatic, axial flow
converters. Cooling between the converters is by synthesis gas preheating
and/or by steam production.
Proprietary items are the purifier and the converters. The process scheme as
such is unique and is not used by other licensors. The use of excess air in the
secondary reformer and the high methane leakage both contribute to a signific-
ant reduction in the size of the primary reformer compared to more conven-
tional process schemes and thereby to savings in fuel consumption. On the other
hand, the power consumption in the air compressor is higher than in conven-
tional plants; the whole front end up to the purifier must handle more gas, and
the purifier adds pressure drop. Energy consumption in the synthesis loop is low
due to the dry and relatively pure synthesis gas. The use of a gas turbine as the
driver for the air compressor adds extra firing and thereby increases the amount
of waste heat generated in the plant. The extra heat can be recovered for steam
production, but the plants will by necessity generate a surplus of steam for
export.
About fifteen plants have been designed using the Braun process. Experience
from operating plants using the Braun technology is reported in [329, 330, 332,
333, 760-763, 952]. A net energy consumption of 6.7 Gcal/MT of ammonia has
been reported [330] for a specific case, operating in a cold climate with partial
production of gaseous ammonia and with full credit for a large export of
18 kg/cm 2 g steam. In [952] energy consumptions of 7.18 and 6.92 Gcal/MT of
ammonia are reported for two plants, also in a cold climate.
I CI has been active in the design and operation of ammonia plants since
before the Second World War. In recent years, they have commercialized two
processes namely the AMV process, and the LCA process.
The AMV process[37, 413-418, 764] was introduced in the early 1980s. A
simplified process flowsheet is shown in Fig. 6.27 (from [416]). Key features of
the process are:
- Reforming with an excess of process air. Steam to carbon ratio at inlet
reformer approximately 2.8. Relatively high methane leakage ( ~ 1% dry)
from the secondary reformer. Addition of part of the process steam via a
saturator. Combustion air preheat.
- Process air compressor driven by a steam turbine (all steam produced in the
 .-------<.~ CO 2 product

.s
~
CII
2G>
c: .0
CII

~ Ja

C02removat

methanation
>
§o
t:l
Pi·
'"C
...o
0-
§.
o·
t:l

...o'"C
~
1 • I: :: 1and drying I.. ~ '"
r.l
rec ered ammonia '"
ammonia svnl SIS NH3 product
tv
00
....,
Fig. 6.27. The lei AMV process. Simplified process flow diagram (from [416])
284 Ib Dybkjaer

. plant goes to this turbine) with a turbo-alternator on the same shaft. All other
power consumers, including synthesis gas compressor, use electric motors.
- Conventional high temperature and low temperature shift conversion.
- Carbon dioxide removal by the Selexol process; compression to synthesis
pressure before final purification.
- Final purification by methanation and drying at synthesis pressure. After
drying, the gas is mixed with recycle gas and off-gas from a cryogenic purge
gas recovery unit. The mixed gas goes via a recirculation compressor and a
gas-gas heat exchanger, the hot exchanger, to the synthesis converter. Be-
tween the recirculator and the exchanger, a part of the gas is taken to a
cryogenic purge gas recovery, where inerts are removed and the gas composi-
tion adjusted by removal of excess nitrogen.
- Synthesis at about 90 kg/cmZg in a three-bed converter with quench between
the two first beds and indirect cooling (converter feed preheat) between the
second and third bed.
As in the Braun process, the size of the primary reformer is reduced and the size
of the air compressor increased in the AMV process compared to more
conventional process schemes. This is due to the operation with excess process
air and with high methane leakage. Power consumption in the synthetic gas
compressor is low because of the low synthesis pressure and the low suction
temperature (gas direct from the low temperature Selexol COz removal unit),
but this is compensated by increased power consumption for the compression of
excess nitrogen and high power consumption in the refrigeration section.
The AMV process scheme may be considered as a variation of the classical
scheme with some unique features. It has been used as described in the design of
{)ne plant and- with some modifications to accommodate integration with a
urea plant-in another plant. Experience from operation of a plant using the
process has been reported in [417]. The expected energy consumption has been
given as less than 7.0 Gcal/MT of ammonia (for a plant located in a cold
climate), but actual data are not available.

The LeA process was announced by ICI in the late 1980s [350-357,
924-927, 951]. A simplified ftowsheet is shown in Fig. 6.28 (from [350J). Key
features of this process are:
- Reforming in gas-heated reformer (unfired, heat supplied by heat exchange
with product gas from the secondary reformer) at a steam to carbon ratio of
2.5. Process steam supplied via a saturator. Large excess of process air; high
methane leakage (1 %). Process air preheat in fired heater.
- One-stage shift conversion in a cooled reactor; exit temperature approxim-
ately 265°C.
- Gas purification: PSA with adjustment of the hydrogen to nitrogen ratio.
Carbon dioxide removal (optional) in a low pressure MDEA wash on the off-
gas from PSA unit. Methanation upstream of the synthesis gas compressor.
- Synthesis in a low pressure loop (approx. 80 kg/cmZg) with a tubular conver-
ter (TV A type).
 air

30 De
gas driers

2582 kmol/h

~1330kmol/h 33.8 bar

::J
a.
~
ao
:;
~ pj'
..,'"t:I
8-
n
'"
g.
Fig. 6.28. The ICI LCA process. Simplified process flow diagram (from [350]) :;
..,'"t:I
o
~
'"
ri
'"
N
00
v.
286 Ib Dybkjaer

Proprietary items are the gas-heated reformer [117, 889], the PSA unit [765]
and the synthesis converter. The process scheme as such is unique. It is,
according to literature, suitable only for small-capacity plants. The process has
so far only been used in one installation comprising two units, each with a
capacity of 450 MTPD.
The amount of process air is adjusted so that sufficient heat is available in
the effluent from the secondary reformer to supply all heat required for the
primary reformer; the excess of process air is approximately 75% compared to
the conventional process. The extra power for compression ofthis excess air and
the loss of hydrogen in the PSA unit increase the energy consumption. On the
other hand, firing in the plant is greatly reduced, and the natural gas consump-
tion is low, only 6.44 Gcal/MT ammonia. The process produces very little steam
(the only boiler is a 60 kg/cm 2 g boiler in the synthesis loop), and extra power
must be supplied from outside battery limits for compressor drivers etc. It is
suggested to raise the required power in a gas-fired, combined cycle power plant
with high efficiency. If this system-which is not incorporated in the operating
plants-is considered, a net energy consumption of 6.95 Gcal/MT ammonia is
claimed. The energy consumption in the actual plants is reported to be
7.6 Gcal/MT [952].
M. W. Kellogg has played a major role in the development of the modern
ammonia industry (see Sect. 6.5.3.2.1). Their most recent Low Energy Process
[37,419,766-768,960] represents a further development of the classical process
maintaining basically the same process scheme, but introducing updated tech-
nology and energy saving features in the various process steps. A simplified
process flow sheet is shown in Fig. 6.29 from ([419]). Key features of the process
are:
- Reforming with the stoichiometric amount of process air (i.e. the amount of
air required to give a hydrogen to nitrogen ratio of 3.0 at the ammonia
synthesis converter inlet without dedicated installations for adjustment of the
gas composition). Steam to carbon ratio about 3.3; low methane leakage
(approximately 0.3% dry).
- Air compressor driven by a steam turbine; preheat of process air to a very
high temperature in the reformer waste heat section.
- Gas purification by conventional shift; carbon dioxide removal normally by
Selexol or Benfield; methanation; drying of synthesis gas with molecular
sieves.
- Synthesis at 140 kg/cm 2 g or 180 kg/cm 2 g (depending on plant capacity) in a
"dry" synthesis loop (make-up gas added after ammonia separator) using a
two-bed horizontal converter with indirect cooling by gas/gas exchange.
Refrigeration in so-called unitized heat exchanger. Purge gas recovery.
Proprietary items are the primary reformer, the synthesis converter, and the
unitized heat exchanger. About 6 plants have been designed using various
versions of this technology. (The total number of ammonia plants designed by
M. W. Kellogg is much higher). Operating experience from a plant using the
 to process steam
high temperature
shift converter
heat
recovery

condensate to
cooling boiler feedwater
process~ ;, system
steam

ammonia
to I ~t
'atmosphere to c;:ondensate
stripper
i
o
2.
po
cooling purge gas
to fuel ~
c.
~
'"
o·
:;
::,0
o
@
lean pump ~
Fig. 6.29. M. W. Kellogg's low energy process. Simplified process flow diagram (from [419])
N
00
-..l
 288 Ib Dybkjaer

Selexol process for carbon dioxide removal and located in a cold climate is
reported in (766). Energy consumption is reported to be below 29GJ (LHV)jMT
of ammonia.
M. W. Kellogg has also suggested a process (the KAAP process) based on a
new catalyst [928, 949]. This technology has been used in a revamp project
where a new reactor was installed downstream of the existing reactor in an
ammonia synthesis loop (see Sect. 6.4.3.4). In addition, more radically new
process schemes deviating from the traditional route have been described. An
example is a scheme based on so-called parallel reforming [30, 32, 120, 121J and
a low pressure loop with ammonia recovery by water absorption [769]. None of
these new developments have been implemented in practice.
Haldor Tops0e AjS has supplied technology for ammonia production for
more than 30 years. Unlike the other process licensors, Tops0e is also a catalyst
supplier, and the development of catalysts and process has gone hand in hand.
The most modern, the Tops0e low energy ammonia process is, as is the M.W.
Kellogg low energy process, based on the traditional process layout with new
energy saving developments introduced in the individual process steps. Descrip-
tions are given in [28, 29, 32, 33, 37, 246, 595, 770, 929]. A simplified flow sheet is
shown in Fig. 6.30 (from [246J). The process is available in two main versions,
the main difference being the steam to carbon ratio at the reformer inlet. The
first version (with a steam to carbon ratio about 3.3) has been used in about 15
industrial plants. The second version (with a steam to carbon ratio about 2.5 and

primary shift
desulphurization reformer secondary conversion CO 2 removal methanation
reformer
air

steam

natural gas

fuel

product ammonia

main
compressor '--------....=.:.:.=:..::=='-'

Fig. 6.30. Haldor Tops0e's low energy process. Simplified process flow diagram (from [246])
 Ammonia Production Processes 289

other energy saving features} is not yet proven in any industrial plant. But each
process step is individually proven, and two plans based on the concept are
under construction.
Key points in the two versions of the Tops0e process are:
First version:
Reforming with stoichiometric air and low methane leakage. Steam to carbon
ratio about 3.3.
- Process air compressor driven by a stream turbine (or gas turbine).
- Gas purification: conventional shift; carbon dioxide removal by Benfield or
Vetrocoke; methanation.
- Synthesis at 220 or 140 kgjcm 2 g with Tops0e 2-bed radial flow converter (S-
200).
Second version:
- Reforming with stoichiometric air and low methane leakage. Steam to carbon
ratio 2.5.
- Process air compressor driven by a steam turbine (or gas turbine).
- Gas purification: shift conversion with medium and low temperature cata-
lysts, both copper-based; carbon dioxide removal by MDEA or Selexol;
methanation.
- Synthesis at 140 kgjcm 2 g with Tops0e 2-bed and I-bed radial flow converters
(S-250).
Proprietary items are the primary reformer, the ammonia synthesis converter,
and certain catalysts. Operating experience in plants designed by Tops0e has
been described in [720, 771-774, 956, 966, 967]. In [967] a net energy consump-
tion of 6.97 GcaljMT ammonia was reported for a plant located in a warm
climate. For the most energy efficient process concept, which has so far not been
demonstrated in an industrial installation, a net consumption of 6.67 GcaljMT
ammonia is claimed for a stand alone plant [920].
Uhde GmbH is one of the pioneers in the ammonia industry. Even before the
Second World War they constructed ammonia plants (based on the Mont Cenis
process). Later they built plants in cooperation with many other companies,
such as 6sw, Tops0e, and ICI.
Uhde has recently commercialized a new low energy process scheme [41,
585, 775, 930]. A simplified process flow is shown in Fig. 6.31 (from [41]). The
process scheme is, like the Tops0e and Kellogg process schemes, a further
development of classical process schemes. Key features of the process are:
- Reforming at a slightly increased pressure with stoichiometric air and high
methane leakage ( :::::; 0.6%). Steam to carbon ratio about 3.0.
- Process air compressor driven by a steam turbine (or gas turbine).
- Gas purification: conventional shift; carbon dioxide removal by MDEA;
methanation.
- Synthesis at approximately 180 kgjcm 2 g in two converters, a two-bed and a
one-bed radial flow converter with steam production between the beds.
290 Ib Dybkjaer

natural gas feed

fuel rocess steam

combustion
aIr

process air

HP-steam
superheated
BFW

C02

s ngas compressor

HP-steam to
superheater

product NH3
Fig. 6.31. Uhde's low energy
fuel process. Block diagram (from
L!!1C!!~~J---"""'"""- [585])

Proprietary items are the primary reformer and the synthesis converters. Two
plants are operating with a design according to this process scheme, while a
third is under construction. An expected net energy consumption of 6.67-7.20
GcaljMT ammonia is claimed, depending on the conditions.
As mentioned above, several other companies have proposed low energy
process schemes, some based on the conventional process route, some represen-
ting a more innovative approach. In the following the most important of these
process schemes will be briefly discussed.
The Exxon Chemical Low Energy Ammonia Process has been used in one
plant. Key features of the process are:
Reforming with stoichiometric air and low methane leakage. Steam to carbon
ratio about 3.3.
- Air compressor driven by a gas turbine. Exhaust from the gas turbine to the
reformer as hot combustion air.
Conventional shift conversion (with low temperature shift guard vessel);
carbon dioxide removal with the Catacarb process; methanation.
Drying of synthesis gas with molecular sieves installed between the first and
second compressor stages.
 Ammonia Production Processes 291

- Ammonia synthesis at 140 bar with a two-bed radial flow converter with
indirect cooling. Make-up gas addition after the separator.
- Cryogenic purge gas recovery.
The energy consumption of the process is given as 29 GJ/MT (6.93 GcaljMT)
ammoma.
The Exxon Chemical process is a good example of processes which are
designed for a specific case in cooperation between the plant owner (Exxon
Chemical), the contractor (Bechtel), and process licensors (Haldor Tops0e for
the synthesis loop and converter). The process and experience from operation of
the plant is described in [776, 777].
The Fluor Ammonia Process [28, 30, 32, 298, 778] is based on a conventional
process scheme. The main features are:
- Use of Fluor's proprietary polypropylene carbonate process (see Sect. 6.3.4)
for carbon dioxide removal. It is suggested [778] that carbon dioxide removal
and methanation be installed after synthesis gas compression.
- Use of heat downstream of the low temperature shift converter for absorption
refrigeration which is used in the ammonia recovery section. In conventional
plants this low level heat is used for reboiling in the carbon dioxide removal
process.
- Air compressor driven by a gas turbine; exhaust from the gas turbine to the
reformer as hot combustion air.
- Primary reforming in a special combination of adiabatic reactors and rehea-
ting elements [125, 126].
It is claimed that the above modifications will reduce the energy consumption
from 32.0 x 199 Btu/short ton (8.89 GcaljMT) to 29.5 x 109 Btu/short ton
(8.19 Gcal/MT) of ammonia [778].
Foster Wheeler's AM 2 Process [779-781] belongs to the group of processes
using an excess of process air to the secondary reformer, so that a specific
process step is required to separate excess nitrogen from the synthesis gas. Key
features of the process are:
- Only 20-50% of the total hydrocarbon feedstock is treated in the primary
reformer which operates at high pressure and a relatively low temperature.
- The remaining feedstock is, together with product gas from the primary
reformer, reacted in the secondary reformer with a large excess of process air.
Methane leakage from the secondary reformer is high, 2.5 vol %.
- Gas purification is by conventional shift conversion followed by a physical
carbon dioxide removal (e.g. Selexol) and final purification in a cryogenic
unit. The hydrogen to nitrogen ratio of the synthesis gas at the cryogenic unit
inlet is 1.0-1.8.
- Synthesis loop conditions have not been specified.
It is claimed that the energy consumption in the process may be as low as 25.1
x 109 Btu/short ton of ammonia (7.0 GcaljMT).
292 Ib Dybkjaer

F oster Wheeler has also suggested a process where the excess of process air is
so high that the primary reformer is deleted. This process (see Sect. 6.5.3.3) can
operate on all types of hydrocarbon feedstock including feedstocks which can
not be used as feed for a primary reformer.
Humphreys & Glasgow have suggested a number of process schemes i.e. the
MDF-process, the BYAS-process, and the LEAD-process.
The MDF-process [28, 30, 32, 341-343,782] consists basically of a hydrogen
plant with PSA purification, an air separation plant for production of nitrogen,
and a synthesis loop converting the hydrogen and nitrogen to ammonia. Key
features of the process are:
- Steam reforming (primary reforming only) at a pressure of about 25 bar.
- Purification of hydrogen by high temperature shift conversion and Pressure
Swing Absorption. Optional carbon dioxide-removal located upstream of the
PSA unit in case carbon dioxide is required.
- Mixing of the pure hydrogen with pure nitrogen which is imported or
produced in an air separation plant.
- Compression and synthesis in an inert-free, dry synthesis loop.
Energy consumption including energy for air separation is given as 32.8
GJjMT of ammonia (7.84 GcaljMT). Significant savings in capital costs are
claimed [782].
The BYAS-process [783-785] differs from the conventional process scheme
mainly in the reformer section and in the final gas purification. Key features of
the process are:

- A significant part of the process feed is sent direct to the secondary reformer
bypassing the primary reformer.
- An excess of air is used in the secondary reformer; the amount of excess
depends on the accepted methane leakage, which may be relatively high.
- Conventional shift conversion and carbon dioxide removal.
- Final purification and adjustment of the hydrogen to nitrogen ratio in a
cryogenic unit.
- Synthesis in an inert-free, dry synthesis loop.

Energy consumption is claimed to be as low as 6.85 GcaljMT of ammonia. The

process is well suited for increasing the capacity of existing plants. An example of
a revamp case is given in [786].
The LEAD Process [32, 420] is basically a variation of the classical process
scheme. Key features are:

- Reforming with a stoichiometric amount of process air using high preheat

temperatures for process feed, process air, and combustion air. Low methane
leakage. As a special feature, natural gas feed is preheated downstream of the
high temperature shift converter, and reformer feed is preheated downstream
of the process gas boiler after the secondary reformer.
 Ammonia Production Processes 293

- Conventional high temperature and low temperature shift; carbon dioxide

removal by the Selexol process; methanation; and synthesis gas drying.
- Synthesis at about 125 kg/cm 2 g in a dry synthesis loop using two converters,
the first a two-bed quench-cooled converter, the second a one-bed converter.
Absorption refrigeration for ammonia recovery. Cryogenic pure gas recovery.
Energy consumption for the process is claimed to be 7.0 Gcal/MT of ammonia
(32).
The PARe Process by KTI [32, 345-349, 785J is very similar to the MDF-
process by Humhpreys & Glasgow as described above. Key features of the
process are:
- Reforming of natural gas (primary reformer only) at 2.9 MPa. Enriched air
from the nitrogen production unit (air separation plant) is used as com-
bustion air in the reformer.
- High temperature shift conversion.
- Production of power in a Rankine cycle with trichlorofluoromethane using
heat available downstream of the shift converter.
- Optional carbon dioxide removal before final purification.
- Final purification of feedstock and nitrogen addition in a specially designed
Pressure Swing Absorption unit.
- Synthesis in an inert-free, dry synthesis loop at 350 kg/cm 2 g without refriger-
ation.
An energy consumption of 29.21 x 10 9 Btu/short ton ()f ammonia (8.11
Gcal/MT) is given for the PARC process [345-348J; in [349J a figure of 27.86
x 10 9 Btu/short ton (7.74 Gcal/MT) is given. Experience from an operating
plant is reported in [787].
Linde has also suggested a process scheme [344J similar to the MDF-process
and the PARC-process. Key features are:
- Production of pure hydrogen by tubular reforming; high temperature shift
conversion; and purification by proprietary Pressure Swing Absorption
technology.
- Production of pure nitrogen by proprietary air separation technology.
- Mixing of hydrogen and nitrogen in the corr~r;t ratio and synthesis in a dry,
inert-free synthesis loop.
The process is said to be especially suited for small plants. Energy consumption
is claimed to be 7.60 Gcal/MT of ammonia.
The Lummus Process [28, 30, 32, 788J is an optimized conventional process
scheme. Key features are:
- Reforming with stoichiometric air at a steam to carbon ratio of approxim-
ately 3.0. Combustion air preheat.
- Conventional high and low temperature shift conversion; carbon dioxide
removal by a physical process; methanation.
- Synthesis at 231 kg/cm 2 g using the Lummus quench cooled converter.
294 Ib Dybkjaer

Energy consumption values from 8.0 Gcal/MT [32] to 7.06 Gcal/MT [28]
are quoted.

Montecatini has suggested a process involving a so-called tertiary reformer

[30,123] and low pressure synthesis, in many ways similar to process schemes
suggested by M. W. Kellogg [120, 121]. Key features of Montecatini's process
scheme are:

- Reforming with stoichiometric air in a special reforming process involving

split flow to two primary reformers, one heated by direct firing, the other
heated by heat exchange with the effluent from the secondary reformer.
- Air compressor driven by a gas turbine; exhaust from the gas turbine is used
as hot combustion air in the fired primary reformer.
- Conventional high temperature and low temperature shift; carbon dioxide
removal using a chemical absorption process; methanation.
- Synthesis of ammonia in a low pressure loop (at 60-70 kg/cm 2 g). Drying of
the mixture of make-up gas and recycle gas by molecular sieves. Synthesis in a
proprietary three-bed radial flow converter with indirect cooling between the
beds.
- Ammonia recovery by absorption in water and distillation. Cryogenic purge
gas recovery. The energy consumption is claimed to be 6.7 Gcal/MT of
ammonia [123].

Projects and Development India Limited (PDIL) has proposed a scheme using
enriched air as process air [789,790]. The scheme is very similar to other modern
low energy concepts with the exception of the use of enriched air. Key features
are:

- Air separation to produce pure oxygen and nitrogen; nitrogen is used as plant
inert gas or exported.
- Primary reforming at elevated pressure, about 45 kg/cm 2 g, and low exit
temperature, about 725°C.
- Secondary reforming using enriched air with 24-30 vol% oxygen. Low
methane leakage.
- Air compressor driven by a gas turbine; compressed air for oxygen produc-
tion may be drawn from an inter mediate stage of the air compressor. Exhaust
from the gas turbine used as hot combustion air for the primary reformer.
- Conventional high and low temperature shift conversion; carbon dioxide
removal preferably by a physical absorption process; methanation.
- Compression and synthesis at 220 kg/cm 2 g using a two-bed radial flow
converter.

Energy consumption for the process scheme is given [789] as 7.40 Gcal/MT of
ammonia.
PDIL has also proposed a process [791] where so much oxygen is added to
the process air that the primary reformer duty is reduced to zero, i.e. the process
uses an auto thermal reformer only in the reforming step.
 Ammonia Production Processes 295

6.5.3.2.4 Revamp of Existing Plants

A large part of the world's ammonia demand is produced in plants which were
designed in the late 1960s or early 1970s using the original Kellogg process (see
Sect. 6.5.3.2.1) or other processes from the same generation. These plants will
typically have an energy consumption of 9-11 GcaljMT ammonia, and it has
become attractive to modernize - "revamp" - such plants in order to increase
the energy efficiency and/or production capacity.
Each revamp project must be tailor-made to enable optimization of produc-
tion, consumption and capital investment based on the specific case. In the most
comprehensive and most extensive revamp projects, the energy efficiency may be
increased to a level comparable with that of the new low-energy ammonia
plants. The required capital investment will, however, in most cases be too high
to justify such extensive revamps.
Revamp of an ammonia plant will typically involve a combination of
replacing or upgrading of equipment, replacing of catalysts and chemicals and
changing the operating parameters. Replacing or upgrading of equipment will,
in addition to more efficient operation, also result in an increased lifetime of the
plant as well as reduced maintenance costs. A higher on-stream factor can be
expected. The new or upgraded equipment is typically designed in such a way
that future bottlenecks are eliminated in that part of the plant. Replacing
catalysts and chemicals will improve efficiency of the plants and, in some cases,
also the lifetime of the unit, while changing operating conditions will mainly
improve plant efficiency.
When major improvements in the performance of a plant is desired, it is
obviously not sufficient to consider only one part of the plant. A complete review
of all sections is required in order to ensure that capacity of the individual units
corresponds in all cases to the expected improved efficiency and/or higher
capacity after revamp. Discussions on such "audits" may be found in [792,793],
and general discussions on the revamp of ammonia plants in [38, 595, 784,
794-803]. Descriptions of actual revamp projects are given in [452, 786,
804-806, 933, 934, 957].
Revamp Options-Front End.
The most important options for revamp of the front end of an ammonia
plant-i.e. the part of the plant where the synthesis gas is generated-using steam
reforming are:
- Reduction of the Steam to Carbon Ratio
- Introduction of Combustion Air Preheat
- Introduction of Feed Gas Saturation
- Installation of an Adiabatic Pre-reformer
- Modification of the Shift Unit
- Modification of the Carbon Dioxide Removal System
- Reduction of the steam to carbon ratio will reduce the pressure drop in the
front-end of the plant. The reduced production of steam in the waste heat
hoiler downstream of the reformer is compensated for by the smaller amount
296 Ib Dybkjaer

. of steam required for the process. The reduced flow inside the reformer tubes
will result in a reduced firing required for the process. The methane leakage
will increase slightly due to the reduced steam to carbon ratio, but the overall
results is typically an energy saving of about 0.20 GcaljMT ammonia when
the steam to carbon ratio is reduced from 4 to 3.
Operation of the reformer at a reduced steam to carbon ratio requires
modification of the downstream shift (if the steam to carbon ratio goes below
the critical level for by-product formation) and carbon dioxide removal
sections in order to fully obtain the potential energy savings. The reduced
steam to carbon ratio will also require a highly active reforming catalyst in
order to eliminate the risk of carbon formation and resulting hot bands on the
reformer tubes.
~ Introduction of combustion air preheat will reduce the amount of fuel
required in the reformer by using the low-level energy in the fuel gas. When
the stack temperature is reduced from about 200°C to 100 °C, the corres-
ponding energy saving is about 0.10 GcaljMT ammonia. The lower temper-
ature of the fuel gas may cause corrosion problems in the form of condensing
sulfuric acid, if the sulfur content of the fuel is high. This problem may be
eliminated by desulfurizing the fuel gas together with the process feed gas. The
energy available in the preheated, pressurized fuel gas may be recovered in an
expansion turbine, which will reduce the pressure to that required in the fuel
gas header system.
~ The feed gas saturator is used to transfer the low-level energy in the fuel gas
into process steam. Hot boiler feed water or process condensate is contacted
with cold natural gas in the saturator, whereby the natural gas is preheated,
and, at the same time, saturated with water. The water leaves the saturator,
make-up water or condensate is added and the water is then reheated in a coil
in the reformer waste heat section. About 20% of the required process steam
may be taken up by the natural gas in the saturator. The energy saving is
typically around 0.10 GcaljMT ammonia. The saturator also serves as a
process condensate stripper whereby overall water consumption may be
reduced, and a pollution problem may be solved. The choice between
combustion air preheat and saturator depends on local conditions. Com-
bustion air preheat saves fuel directly whereas the saturator saves steam.
~ An adiabatic pre-reformer is an attractive revamp option, especially when the
aim is to minimize steam export and/or to increase plant capacity in cases
where the primary reformer is the bottleneck.
In the typical reformer waste heat section, the process feed is preheated to
about 520°C before entering the reformer tubes, where reforming takes place
using the additional heat supplied by the burners. The burner fuel could be
reduced, if the process feed were preheated to higher temperatures, but this
may not be feasible due to the increased potential for carbon formation by the
cracking of hydrocarbons. By passing the preheated process feed through an
adiabatic pre-reformer, the higher hydrocarbons in the feed will react
 Ammonia Production Processes 297

quantitatively with the steam causing a temperature drop. The partially

reformed mixture can then be reheated safely in the waste heat section before
entering the reformer tubes. The overall result is a reduction of reformer fuel
consumption and a corresponding reduction in the waste heat available for
steam production. The fuel consumption can, in certain cases, be reduced by
as much as 15%.

When the steam to carbon ratio in the reforming section is reduced, the
conditions in the shift section must be carefully evaluated. If the steam to dry
gas ratio becomes too low, severe problems may arise due to conversion of the
iron oxide in the high temperature shift catalyst to iron carbide, which will
promote formation of undesirable by-products (hydrocarbons and oxy-
genates) (see Sect. 6.3.3.1).
When the steam to carbon ratio in the reforming section is not reduced
below the critical limit for carbide formation, it is possible, by adding an extra
shift catalyst bed after the conventional shift unit, to reduce the carbon
monoxide leakage to around 0.05 dry vol%. This means that less hydrogen
will be lost by reaction with carbon monoxide in the methanator, and that the
synthesis loop will become more efficient due to the lower inert content in the
make-up gas.

Several revamp options are available for modification of the carbon dioxide
removal section depending on the type of carbon dioxide removal process.
The processes mostly used in ammonia plants are chemical absorption
processes based on either hot potassium carbonate (HPC) such as Benfield, or
Vetrocoke, or amine solutions such as MEA. The chemical carbon dioxide
removal processes may be improved or replaced with a physical process in
which the absorbent is regenerated by simply flashing off carbon dioxide. In
this way the need for regeneration heat may be reduced or eliminated. A
physical carbon dioxide removal system may result in energy savings of
0.01-0.35 GcaljMT ammonia.
In certain cases methanolation may be an attractive revamp option
[907,908]. In this process, the process gas after carbon dioxide removal is fed
to a methanol synthesis reactor instead of a methanation reactor. Residual
carbon oxides are converted to methanol, which can be recycled to the
reformer on the process side, whereby hydrogen loss is avoided, and the inert
content of the make-up gas is reduced. An overall reduction in reformer feed
consumption of 0.15-0.30 Gcal/MT NH3 can be obtained depending on
plant layout. There will, however, be an increased consumption of reformer
fuel to compensate for the reduced heating value of the loop purge gas caused
by the reduced methane content. The total saving in energy consumption is
therefore somewhat smaller, in the range of 0.05-0.10 GcaljMT NH 3 . As an
alternative to recycling of methanol, it can be recovered as a product which is
exported or converted to formaldehyde or ureaform, which can be used for
coating of urea to prevent caking.
298 Ib Dybkjaer

Revamp Options-Synthesis Loop.

The main revamp options of the ammonia synthesis loop are the installation
of molecular sieves for drying of the make-up gas, installation of a purge gas
hydrogen recovery system, and improvements in the ammonia synthesis conver-
ter system.
- The installation of molecular sieves to remove water and carbon dioxide from
the make-up gas and the corresponding relocation of the make-up point to
the inlet of the synthesis reactor is especially interesting in loops where the
make-up gas addition point is upstream of the water cooler. In such cases, the
modification will result in a reduced pressure drop as well as a slightly higher
conversion. The energy savings will vary with the loop configuration and may
be in the order of 0.10-0.15 GcaljMT ammonia. In process such as the
Tops0e and Uhde processes where the make-up gas is added just before the
last chiller, the potential savings are much smaller, only about 0.03 GcaljMT
[461]. In such cases installation of make-up gas driers can not be justified.
- Hydrogen loss with the purge gas may be reduced substantially by installing a
purge gas recovery unit such as a membrane unit, a PSA unit or a cryogenic
unit. Energy savings of up to 0.15 GcaljMT ammonia have been demon-
strated (see Sect. 6.4.7).
Converter Revamp options are discussed in detail in Sect. 6.4.3.4.

6.5.4 Processes based on Partial Oxidation of Hydrocarbons

or Gasification of Solid Feedstocks

In certain economic circumstances it may be of interest to produce ammonia

from feedstocks such as fuel oil or coal (see Sect. 6.5.1 and [676-691]). The
technology for the production of synthesis gas from such feedstocks is partial
oxidation or gasification (see Sect. 6.3.2). Partial oxidation of natural gas was
used in a few early plants [137, 138,807], and it was suggested in a few modern
process schemes, for example by Foster Wheeler [808,809], but it is of no
importance today.
Plants based on partial oxidation of heavy fuel oil by the Shell process are
described in [166-168] and by the Texaco process in [163-165, 169]. Coal
based plants using the Koppers-Totzek process are described in [187,188], the
Winkler process in [202], and the Texaco process in [213-219]. Catalysts for
coal based ammonia plants are discussed in [810], and problems in synthesis gas
purification in [811,812]. Complete process concepts are discussed in [179, 182,
813-818].
Two classical process schemes are illustrated in Fig. 6.32 (from [818]). The
schemes as the illustrated can be used both with fuel oil or coal as the feedstock.
In both cases the feedstock is converted to raw synthesis gas by reaction with
oxygen and steam. In scheme A, which is used with gasification processes which
require cooling of the raw gas to near ambient temperature for soot removal,
 Ammonia Production Processes 299

A B

H2 S

fuel gas

Fig. 6.32. Two classical process schemes for production of ammonia from fuel oil (or coal). (from
[818J)
a Scheme with Shell gasification (or other process with cooling of raw gas to near ambient
temperature)
b Scheme with Texaco gasification (or other process with soot removal at elevated temperature)

hydrogen sulfide is removed upstream of the shift conversion, and conventional

shift catalysts are used. In scheme B, which is used with gasification processes
where soot is removed at elevated temperature, sulfur remains in the feed gas to
the shift conversion section, and sulfur tolerant shift catalysts must be used. In
both process schemes the final gas purification is done by acid gas removal and
nitrogen wash, and ammonia synthesis is done in a dry, inert-free ammonia
synthesis loop. In a special process proposed by Topsoe [818,819], gasification
is done with enriched air, the raw gas goes after cooling and soot removal direct
to the shift conversion without sulfur removal. The shift conversion is done in
300 Ib Dybkjaer

two stages, using a special high activity sulfur tolerant shift catalyst. After the
shift conversion, hydrogen sulfide and carbon dioxide are removed in an acid gas
removal unit; the remaining traces of carbon oxides are removed by methan-
ation; and ammonia is synthesized in a normal synthesis loop using the Topsoe
S-200 two bed radial flow converter. A synthesis gas generation process similar
to this scheme and using the same type of catalysts-but based on gasification
with oxygen-is used in a large ammonia plant in Japan [213,214].

6.5.5 Integrated Production of Ammonia and Other Products

In many cases ammonia is supplied direct-or via an intermediate storage-to

another production unit where it is further processed to a final product, for
example urea or ammonium salts (nitrate, phosphate, etc.). In other cases there
may be on the same site production of other products, such as methanol, which
do not require ammonia as raw material, but may be made from the same raw
materials as ammonia. In both situations it may be advantageous to integrate
the production units to obtain savings in investments and/or production costs.
The integration may be at various levels, ranging from shared off-site and utility
installations (cooling water installations, raw water treatment, auxiliary power
production and distribution etc.) through integrated steam systems (with e.g.
boiler feed water preheating in one unit for boilers located in other units) to
complete integration on the process side. Only the last type of integration will be
considered in the following. In addition, the special situations will be discussed
which arise when ammonia and another product must be produced in a
constant ratio because they are used in fixed amounts in downstream processes.
A typical example is a urea complex without facilities for ammonia export. In
such units there must be a fixed ratio between ammonia and carbon dioxide
production in the ammonia plant and this will have an impact on the design of
the production process.

6.5.5.1 Ammonia and Urea

Urea, CO(NH 2 h, is formed by dehydration of ammonium carbamate,
NH 4 C0 2 NH 2 , which is again formed by the reaction between ammonia and
carbon dioxide
(18)
It follows that one half mole of carbon dioxide is required for each mole of
ammonia to obtain complete conversion. It is evident from Eq. (17) (p. 269) that
when natural gas is the feedstock, and the highest possible conversion of
feedstock to ammonia is achieved, there may be a deficit in carbon dioxide
production.
The most important ways to overcome this problem and to obtain stoichio-
metric production of carbon dioxide for complete conversion of ammonia to
 Ammonia Production Processes 301

urea can be illustrated by theoretical equations similar to Eq. (17) [701]:

a 0.4422 CH 4 + 0.6155 H 20 + 0.6407 Air--+

NH~ + 0.4424 CO 2 + 0.0060 Ar (19)
b 0.4998 CH3.0786 + 0.7306 H 20 + 0.6407 Air--+
NH3 + 0.50 CO 2 + 0.0060 Ar (20)
c 0.4998 CH 4 + 0.6956 H 20 + 0.7421 Air--+
NH3 + 0.50 CO 2 + 0.1952 H2 + 0.0651 N2 + 0.0068 Ar (21)
d 0.4997 CH 4 + 0.5006 H 20 + 1.1896 Air--+
NH3 + 0.50 CO 2 + 0.0112 Ar + 0.4284 N2 (22)
e 0.4998 CH 4 + 0.5004 H 20 + 0.6407 Air + 0.1153 O 2 --+
NH3 + 0.50 CO 2 + 0.0060 Ar
Air = 0.2099 O 2 + 0.7804 N2 + 0.0003 CO 2 + 0.0094 Ar. (23)

Case a above corresponds to Eq. (17). It is seen that in this case, which is
based on the use of pure methane as feed, on 100% conversion of the feed and
100% yield of ammonia from hydrogen and nitrogen and of carbon dioxide
from carbon and oxygen, there is a deficit of carbon dioxide of 13 % compared to
the amount required for full conversion of ammonia to urea.
_Case b illustrates that if the feed contains heavy hydrocarbons correspond-
ing to an overall hydrogen to carbon atomic ratio slightly above 3.0, then the
stoichiometric production of carbon dioxide may automatically be achieved.
Case c shows a solution where the extra carbon dioxide is obtained by
production of excess synthesis gas. The excess synthesis gas can be extracted as a
side stream before the synthesis gas compressor and used as fuel in the reformer
or exported as such or as hydrogen after purification. Alternatively, it can be
compressed and passed through the synthesis loop which will then operate with
a high purge rate and correspondingly low inert level. The actual solution will
depend on the specific case, especially on the composition of the feedstock.
Solutions of this type are used by the process schemes (see Sect. 6.5.3.2.3) using a
stoichiometric amount of process air, such as the Kellogg, Topsoe and Uhde
processes.
Case d shows that stoichiometric carbon dioxide production can also be
achieved by using an excess of process air. However, in this case it becomes
necessary to introduce an extra gas separation process step in order to remove
the excess nitrogen which is introduced with the excess air. As explained in
Sect. 6.5.3.2.3, several process schemes exist where excess process air is used, and
the excess nitrogen is removed in dedicated units. Processes using excess process
air are the Braun process, the ICI processes, and several process schemes of
minor importance.
302 Ib Dybkjaer

Case e illustrates that sufficient carbon dioxide can also be produced by

adding pure oxygen, i.e. by using enriched air instead of normal air. In the
theoretical situation illustrated in case e, the oxygen concentration in the
enriched air becomes 33%. Less oxygen is required if the feedstock contains
higher hydrocarbons. Use of enriched air has been suggested in several process
schemes, see Sect. 6.5.3.2.3.
In addition to the above discussed influence of urea production on the design
of ammonia plants due to the required carbon dioxide production, ammonia
and urea plants may interact further on several levels. It is customary to
integrate the steam systems, for example by driving the carbon dioxide com-
pressor turbine with steam produced in the ammonia plant and extracting
process steam for the urea plant from the same turbine or from one of the drivers
in the ammonia plant. A discussion offurther integration may be found in [820].
Integration of ammonia and urea plants on the process side has been suggested
by several companies, for example Toyo Koatsu [821-823], M. W. Kellogg
[824,825] and Snamprogetti [826-828].
In all cases the main point of integration is the carbon dioxide removal,
which is done by washing the process gas with an ammonium carbonate/am-
monia solution. The gas from the shift conversion unit is cooled, compressed to
200-300 kg/cm 2 g, and carbon dioxide is removed by contact with the solution.
The remaining carbon oxides are removed by methanation, and the gas goes to
the ammonia synthesis section. The Snamprogetti process, which has been
described in considerable detail [826], features a falling film absorber for carbon
dioxide absorption. In addition, a falling film absorber is used in the ammonia
synthesis loop to recover the ammonia product by absorption in water. The
aqueous ammonia is sent direct to the urea reactor. Traces of water are removed
from the recycling gas in the ammonia synthesis loop by the injection of liquid
ammonia [32,639]. In [829] a process is described in which the raw ammonia
synthesis gas containing carbon dioxide is introduced into the urea reactor
where the carbon dioxide is removed by reaction with ammonia. Significant
improvements in the urea production process are claimed.

6.5.5.2 Ammonia and Other Products

A review of integrated processes based on steam reforming of hydrocarbons is
given in [825]. It is proposed to integrate production of ammonia and hydrogen
by recovering hydrogen from the ammonia synthesis purge gas (see Sect. 6.4.7).
The hydrogen to ammonia production ratio can, within limits, be adjusted by
varying the purge rate from the synthesis loop. The suggested scheme for
integrated production of ammonia and methanol features the installation of a
methanol synthesis unit (with or without synthesis gas compression or synthesis
gas recycle depending on required ammonia to methanol production ratio)
between the secondary reformer and the shift conversion unit in the conven-
tional ammonia process. Co-production of ammonia and methanol may also be
achieved by installation of a methanolation unit (see Sect. 6.3.5.5) with or
 Ammonia Production Processes 303

without partial by-pass of the shift unit and/or the carbon dioxide removal unit
to adjust the ammonia to methanol production ratio. Conversion of existing
ammonia plants to combined production of ammonia and methanol is de-
scribed in [935]. Integration of ammonia production with the production of
steel has also been described [380-384]. Production of ammonia from off gases
from ethylene plants or methanol plants has also been used [378,379]. Integ-
rated production of ammonia and methanol by modifying the methanol process
by incorporating an oxygen-fired secondary reformer and producing ammonia
from the off-gas and nitrogen from the air separation unit is discussed in [968].
A plant producing both methanol and ammonia from synthesis gas produced by
partial oxidation of heavy oil is described in [166-168].

6.6 Storage of Ammonia

Facilities for storage of ammonia are required at the production facilities either
as buffer capacity to smooth out variations in production and demand between
the ammonia plant and downstream units using ammonia as raw material or to
hold product between shipments from plants exporting ammonia. Required
storage capacity may range from less than one to a few days' production
capacity for intermediate storage and up to one month's production capacity, or
even more, for exporting plants located in remote areas. This means that
ammonia storage tanks in ammonia plants may range in size from less than
100 t to above 40000 t.
In addition to this, very large ammonia storage facilities have in recent years
been constructed at terminals served by ammonia pipelines. Pipeline transport
of ammonia and related storage facilities are discussed in [830,831]. In the
construction and operation of ammonia storage facilities the special properties
of ammonia-an inflammable and toxic compound-must be taken into account.
The properties of ammonia and the hazards involved in handling and storage of
this product are dealt with in chapter 7. General descriptions of the design of
storage facilities are given in [832,833]. Surveys of actual installations may be
found in [834-840, 936]. Safety recommendations are given in [841-843] (see
also chapter 7).
Depending on the capacity, ammonia may be stored by three different
methods:
- In non-refrigerated pressure vessels-perhaps cooled by water spray and/or
painted with reflecting paint. The storage pressure corresponds to ammonia
vapour pressure at ambient temperature. The storage vessels, often referred to
as "bullets", are usually cylinders; capacity is generally low, below about
200 t. Non-refrigerated spherical pressure vessels with capacities up to 1500 t
ammonia have been used [833].
304 Ib Dybkjaer

- In semi-refrigerated tanks where the ammonia is refrigerated-usually by a

refrigeration unit with a single stage ammonia compressor-to a temperature
often around 0 0c. The storage pressure corresponds to ammonia vapour
pressure at the refrigeration temperature and may be as low as 2.5 kg/cm 2 g.
Semi-refrigerated storage vessels, normally spheres or in some cases special
constructions referred to as "Horton Spheres", may have capacities up to
about 3000 t.
- In fully refrigerated storage tanks (cylindrical, insulated tanks with flat
bottom and domed roof) at essentially atmospheric pressure and - 33°C.
The refrigeration is provided by a refrigeration unit, most often with a two
stage ammonia compressor. The refrigeration may also be integrated with the
ammonia synthesis loop refrigeration system. The tanks, most often referred
to as "Atmospheric Ammonia Storage Tanks"-may have capacities up to
above 40000 t. Characteristic features of the three types of storage is given in
Table 6.6 (based on [836,837]).
The choice between the three types evidently depends on the capacity required
and is made on the basis of economics in such a way that fully refrigerated tanks
are used for the largest capacities and non-refrigerated vessels for the smallest
capacities. Semi-refrigerated vessels are used in an intermediate capacity range,
but according to [832, 833] they are becoming less popular.
In addition to the storage capacity, the rate of required supply of ammonia
at different conditions (gaseous, warm liquid, cold liquid) must also be taken
into account. This is discussed with reference to a specific situation in [844]. In
many cases ammonia terminals-both for export from producing plants and for
import from the distribution system-involve a combination of pressurized
spheres and refrigerated storage tanks. As an example, Fig. 6.33 (from [831)]
shows a system capable of delivering product ammonia from a manufacturing
plant to a variety of distribution systems. The figure assumes that the refriger-
ation system for the refrigerated storage is integrated with the refrigeration for
the ammonia synthesis loop, and that ammonia must be supplied at - 33°C to
a ship or barge and at ambient temperature to pipeline or to rail or road
transport cars.

Table 6.6. Characteristic features of ammonia storage tanks

Type Typical Design tAmmonia Capacity Refrigeration

Pressure Temperature per t Steel Range Compressor
kgjcm 2 g °C t ammonia

Non-refrigerated 18 ambient 2.75 < 270 None

Semi-refrigerated 3-5 :::::;0 10 450-2700 Single-stage
Fully refrigerated 1.1-1.2 - 33 41-45 4500-45000 Two-stage
 Ammonia Production Processes 305

r---'I
I

•
I
ammonia
manufacturing
plant

;:
o
.~

-oE
Sa 1----\:><1-- pipe line fill
1----\:><1-- rail tank car
I----,../r-----t><l-- fill
L.o----J~-i><3-- highway
f---t:<1--- truck fill

r----'
I I ship or
river barge
dock
M

Fig. 6.33. Generalized scheme for supply of ammonia from manufacturing plant to various
transport facilities (from [831])

6.6.1 Pressurized Storage (Non-Refrigerated and Semi-Refrigerated)

This type of storage, as mentioned above, is most often used for storage of
relatively small quantities of ammonia as intermediate storage between the
ammonia plant and ammonia consumers or transport systems accepting warm
ammonia. Pressurized tanks are also used extensively in the distribution system
for storage of small quantities of ammonia, for example in field tanks in
connection with direct application of liquid ammonia as a fertilizer. Non-
refrigerated pressurized storage tanks may be protected against overheating in
hot weather by insulation, by water spray, or by circulating ammonia through a
water cooler. The design pressure is typically around 18 bar.
The design of pressurized storage tanks and related safety aspects are
reviewed in [845]. Stress corrosion cracking is a much debated risk in pre-
ssurized storage tanks. The mechanism of stress corrosion cracking (including
the effect of oxygen and water) is discussed in [846-850, 937] and, in an
authoritative review, in [938]. It has been generally accepted that the addition of
water to ammonia inhibits stress corrosion cracking [832,846], and in certain
situations (transport of ammonia in containers made from certain specified
306 Ib Dybkjaer

types of steel) a minimum water content of 0.2 wt% is a requirement [851J.

Further protection may be obtained by cathodic polarization, for example by
aluminum or zinc metal spray [937,938]. The most recent research [849,938J,
however, has shown that water does not give complete protection. It has also
been more or less accepted that stress corrosion cracking does not constitute a
risk in atmospheric storage tanks. Recently, incidents have been reported,
however, [852, 940, 941J in which stress corrosion cracking had occurred in
large atmospheric storage tank. General surveys of cracking incidents are given
in [835, 854]. Specific cases including examples of inspection and repair proced-
ures are given in [855-858, 939, 942, 943].

6.6.2 Atmospheric Storage (Fully Refrigerated)

Atmospheric storage at - 33°C is, as mentioned, used for large-capacity

storage; single tanks with a capacity above 40000 t of ammonia are known. The
storage tanks are cylindrical with a flat bottom and domed roof. The refriger-
ation is provided by recompression of boil-off, normally in a two-stage com-
pressor. Incoming warm ammonia is flashed to - 33°C before entering the
tank, and the vapors from the flash vessel are also handled by the refrigeration
compressor. A sketch of a typical storage unit is shown in Fig. 6.34 (from [832J).
Two types of tanks are used, single and double wall tanks. The single wall
tank consists of a single shell designed to withstand the operating pressure. The
shell is insulated; the insulation must be completely vapour-tight and requires a
very high standard of construction of maintenance to avoid hazards.
Double wall tanks consist of an inner tank designed for the storage
temperature and pressure surrounded by a second tank, with a gap between the
two shells of normally 18 inches or more. The gap is filled with insulation
material, and the main purpose of the outer tank is to contain the insulation. It
is, however, becoming more and more normal to design the outer shell to the
same standards as the inner shell. These so-called "double-integrity" tanks offer
the added safety that the outer tank will hold the ammonia in case of failure of
the inner tank. As a further safety measure, ammonia storage tanks are normally
surrounded by dikes or placed in concrete basins with the capacity to contain
the liquid ammonia in case of complete failure of the tank. Discussions of such
secondary retainment are given in general and with reference to specific cases, in
[859-863J. Owing to the refrigeration, there is a risk of ice formation under the
tank. This must be avoided since it could lead to movements that could damage
the bottom of the tank, and the foundation of the tank is therefore normally
provided with heating, or the tank is constructed on piles with ventilation under
the bottom. Problems relating to tank foundations are discussed with reference
to specific cases in [864-866, 944]. Details about the design of storage tanks are
given in [832, 867, 868]. The construction of a new high-integrity tank is
discussed in [869].
 condenser to stack,
1st stage

purger
condensate
receiver

incoming warm ammonia

storage
f
~.
vapor return line
tank

I
§

heater r
'"
Fig. 6.34. Typical fully refrigerated ammonia storage unit. Simplified process flow diagram (from [832])

~
308 Ib Dybkjaer

Models for tank behaviour and potential hazards during tank loading are
described in [945].
Retrofit of existing tanks for added safety is discussed in [870]. Special design
and operating considerations relating to the influence of climatic conditions are
reported in [871]. Safety aspects including maintenance and inspection proced-
ures are reviewed with reference to specific cases in [872-875, 946-948]. For
risk of stress corrosion cracking in atmospheric ammonia storage tanks see
Sect. 6.6.1.

6.7 References

1. Harding AJ (1959) Ammonia manufacture and uses, Oxford University Press, London
2. Vancini CA (1961) La Sintesi dell'Ammoniaca. Ulrico Hoepli, Milano
3. Sauchelli V, Stevenson FJ, Timm B, Danz W, Axelrod LC, O'Hare TE, Hansen HJ (1964) In:
Sauchelli V (ed) Fertilizer nitrogen its chemistry and technology. Reinhold, New York
4. Vancini CA (1971) Synthesis of ammonia. Macmillan, London
5. LeBlanc JR Jr, Madhavan S, Porter RE (1978) In: Kirk-Othmer Encyclopedia of Chemical
Technology, 3rd ed. Wiley, New York, vol 2, p 470
6. Slack AV, James GR (eds) Ammonia. Marcel Dekker, New York, vol. 1 (1973) vol 2 (1974) vol 3
(1977) vol 4 (1979)
7. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York
8. Bakemeier H, Huberich T, Krabetz R, Liebe W, Schunck M, Mayer D (1985) In: Ullman's
Encyclopedia of Industrial Chemistry, 5th edn, vol A2, VCH, Weinheim, p 143
9. Noyes R (1964) Ammonia and synthesis gas 1964. Noyes Development Coporation, New York
10. Noyes R (1967) Ammonia and synthesis gas 1967. Noyes Development Coporation, New York
11. Brykowski FJ (ed) (1981) Ammonia and synthesis gas recent and energy-saving processes,
Noyes Data Corporation, Park Ridge, p 1
12. Mittasch A (1951) Geschichte der Ammoniaksynthese. Verlag Chemie, Weinheim
13. Timm B (1960) Chern Ind (London) 12: 274
14. Timm B (1963) Chern Ing Tech 35: 817
15. Timm B, Danz W (1964) In: Sauchelli V (ed) Fertilizer nitrogen its chemistry and technology.
Reinhold, New York, p 40
16. Appl M (1976) Nitrogen 100: 47
17. Slack AV (1973) In: Slack AV, James GR (ed) Ammonia, Marcel Dekker, New York 1: 5
18. Waeser B (1939) Chern Fabr 12(31, 32): 370
19. Brown CO (1954) Chern Eng Prog 50(11): 556
20. Nitrogen (1962) 16: 35
21. Strelzoff S, Vasan S, (1963) Chern Eng Prog 59(11): 60
22. Bresler SA, James GR (1965) Chern Eng (NY) 72(13): 109
23. Allen JB (1965) Chern Process Eng (London) 46(9): 473
24. Mayo HC, Finneran JA (1968) Erdol Kohle, Erdgas, Petrochem 21(7): 404
25. Quartulli J, Wagener D (1973) Erdol Kohle, Erdgas, Petrochem Brenns Chern 26(4): 192
26. Lyon SD (1975) Chern Ind (London) 17: 731
27. Quartulli OJ , Buividas LJ (1976) Nitrogen 100: 60
28. Pachaiyappan V (1979) Fert News (August) 41
29. CEER (1979) Chern Econ Eng Rev 11(5): 24
30. Zardi U, Antonini A (1979) Nitrogen 122: 33
31. Honti GD (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphus Corp Int ConfFert Technol,
4th, part 1, 1
32. Saviano F, Lagana V, Bisi P (1981) Hydrocarbon Process 60(7): 99
33. Rai HS (1982) CEER Chern Eng World 17(1): 69
34. Brown FC (1983) Proc Fert Soc London 218: 1
35. Dybkjaer I, (1983) ECN Eur Chern News Fertilizers '83 Supplement (February 21): 15
36. Picciotti M, Pocini CA (1984) Chim Ind (Milan) 66(2): 97
 Ammonia Production Processes 309

37, Agarwal MR (1982) CEW Chern Eng World 17(1): 55

38. James GR, Stokes KJ (1984) Chern Eng Prog 80(6): 33
39. Blanken JM (1987) Proc Fert Soc London 259: 1
40. Blanken JM (1989) Ammonia Plant Saf 29: 273
41. Fertilizer Focus (1987) 4(10): 22
42. Fokdal A, Hansen HJ (1955) Ingeni0ren 47: special issue 902
43. Catalyst handbook (1970) Wolfe Scientific Books, London
44. Twiggs MV (ed) (1989) Catalyst handbook, 2nd ed. Wolfe Publishing Ltc, London
45. Hansen HJ (1964) In: Sauchelli V (ed) Fertilizer nitrogen its chemistry and technology.
Reinhold Publishing Corporation, New York, p 89
46. Rankin JD (1978) Proc. Fert Soc London 168: 1
47. Badische Anilin- & Soda-Fabrik, Ludwigshafen (1908) GER 235421, CA (1911) 5: No. 3137
48. Haber F (1909) GER 238450, CA (1912) 6: No. 1663
49. Slack AV (1973) In: Slack AV, James GR (eds) Ammonia, vol 1. Marcel Dekker, New York,
p20
50. Gmelins Handbuch der Anorganischen Chemie (1935) Stickstoff. Verlag Chemie, Berlin, p 295
51. Habermehl R, Long ND (1974) In: Slack AV, James GR (eds) Ammonia vol 2. Marcel Dekker,
New York, p 333
52. Kjeldgaard K (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New
York, p 341
53. Phillipson 11 (1970) In: Catalyst handbook. Wolfe Scientific Books, London, p 46
54. Carnell JH (1989) In: Twiggs MV (ed) Catalyst handbook, 2nd ed. Wolfe Publishing Ltd,
London, p 199
55. Nitrogen (1971) 72: 34
56. Livingston JY (1971) Hydrocarbon Process 50(1): 126
57. Lee MNY, Collins 11 (1969) Ammonia Plant Saf 11: 59
58. Burklow BW, Coleman RL (1977) Ammonia Plant Saf 19: 21
59. Jensen PE, S0ndergaard K (1984) Ammonia Plant Saf 24: 47
60. Carnell PJH, Denny PJ (1985) Ammonia Plant Saf 25: 99
61. Carnell PJH, Denny PJ, Clark DN (1987) Ammonia Plant Saf 27: 99
62. Chatterjee A, Rajagopal KR (1982) CEW Chern Eng World 17(1): 63
63. Polasek J, Bullin J (1984) Energy Prog 4(3): 146
64. Rostrup-Nielsen JR (1975) Steam reforming catalysts. Teknisk Forlag A/S, Copenhagen
65. Rostrup-Nielsen JR (1984) In: Anderson JR, Boudart M (eds) Catalysis science end technology
vol 5. Springer, Berlin
66. Vancini CA (1961) La Sintesi dell'Ammoniace. Ulrico HeopJi, Milano, p 160
67. Noyes R (1967) Ammonia and synthesis gas 1967, Noyes Development Corporation, Park
Ridge, p29
68. Atwood K, Knight CB (1974) In: Slack AV, James GR (eds) Ammonia. Marcel Dekker, New
York, vo12 p 145
69. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 101
70. Brykowski FJ (ed) (1981) Ammonia and synthesis gas recent and energy-saving processes.
Noyes Data Corporation, Park Ridge, p 20
71. Gard NR (1966) Nitrogen 39: 25
72. Appl M, Giissling H (1972) Chern Ztg 96(3): 135
73. Nitrogen (1962) 18: 24
74. Voogd J, Tielrooy J (1963) Hydrocarbon Process Pet Refin 42(3): 144
75. Borgars DJ (1964) Chim Ind (Paris) 91(3): 276
76. Young RJ (1964) Chern Age India 15(6): 741
77. Bridger GW, Wyrwas W (1967) Chern Process Eng (London) 48(9): 101
78. Marschner F, Renner HJ (1982) Hydrocarbon Process 61(4): 176
79. Rostrup-Nielsen JR (1973) Ammonia Plant Saf 15: 82
80. Agarwal MR, Bajpai T (1977) Chern Age India 28(113): 995
81. Tottrup PB, Nielsen B (1982) Hydrocarbon Process 61(3): 89
82. Bridger GW (1966) Chern Process Eng (London) 47(12): 39
83. Bridger GW (1976) Oil Gas J 74(7): 73
84. Rounthwaite DP (1983) Plant/Oper Prog 2(2): 127
85. Nitrogen (1988) 174: 23
86. Congram GE (1976) Oil Gas J 74(7): 75
87. Rostrup-Nielsen JR (1982) In: Figueiredo JL (ed) Progress in catalyst deactivation. Martinus
Nijhoff Publishers, The Hague, p 127
310 Ib Dybkjaer

88. Rostrup-Nielsen JR (1982) In: Figueiredo JL (ed) Progress in catalyst deactivation. Martinus
Nijhoff Publishers, The Hague, p 209
89. Davies J, Lihou DA (1971) Chern Process Eng (London) 52(4): 71
90. Marsch HD, Herbort HJ (1982) Hydrocarbon Process 61(6): 101
91. Nitrogen (1987) 166: 24
92. Nitrogen (1987) 167: 31
93. Leyel CP (1975) Ammonia Plant Saf 17: 41
94. Thuillier J, Pons F (1978) Ammonia Plant Saf 20: 89
95. Kawai T, Takemura K, Zaghloul MB (1984) Paper presented at International Plant Engr Conf,
Bombay
96. Kawai T, Takemura K, Shibaksi T, Rump L, Danielsen B, Wrisberg J (1982) Paper presented at
AIChE Ammonia Saf Symp. San Francisco
97. Muhlenforth CJ (1989) Finds 4(2): 14
98. Salot WJ (1972) Ammonia Plant Saf 14: 119
99. Salot WJ (1973) Ammonia Plant Saf 15: 1
100. Konoki K, Shinohara T, Shibata K, (1982) Plant/Oper Prog 1(2): 122
101. Imoto Y, Terada S, Maki K (1982) Plant/Oper Prog 1(2): 127
102. Kawai T, Takemura K, Shibasaki T, Mohri T (1982) Plant/Oper Prog 1(3): 181
103. Kawai T, Takemura K, Shibasaki T, Mohri T (1980) Ammonia Plant Saf 22: 119
104. Kawai T, Mohri T, Takemura K, Shibasaki T (1984) Ammonia Plant Saf 24: 131
105. Cockerham RG, Percival G (1957/58) Trans Inst Gas Eng 107: 390
106. Jockel H, Triebskorn BE (1973) Hydrocarbon Process 52(1): 93
107. Rail W, (1967) Erd61 Kohle, Erdgas, Petrochem 20(5): 351
108. Davies HS, Humphries KJ, Hebden D, Percy DA (1967) IGE J (October): 708
109. Jockel H (1969) GWF Gas Wasserfach 110(21): 561
110. Ishiguro T (1968) Hydrocarbon Process 47(2): 87
111. Rostrup-Nielsen JR, Tottrup PB (1979) In: Proc Symp Sci Catal its Appl Ind, Sindri, p 379
112. Clark DN, Henson WGS (1988) Ammonia Plant Saf 28: 99
113. Mosley T, Stephens RW, Stweart KD, Wood J (1972) J Catal 24: 18
114. Stahl H, Rostrup-Nielsen J, Udengaard NR (1985) In: Fuel cell seminar 1985. Tuscon, Arizona,
p 83
115. Udengaard NR, Christiansen LJ, Summers WA (1988) Endurance testing of a high-efficiency
steam reformer for fuel cell power plants. EPRI AP-6071, Project 2192-1. Electric Power
Research Institute, California.
116. Sederquist RA (1978) US 4071330; (1978) CA 89: No. 8711
117. Smith ASP, Doy RJ, Limbach APJ (1986) EP 194067, CA (1986) 105: No. 229260.
118. Ruziska PA (1984) EP 113198; CA 101: No. 113278
119. Pinto A (1988) US 4750986; see EP 124226 CA (1985) 102: No. 27497
120. Crawford DB, Becker CL, LeBlanc JR (1978) US 4079017; CA (1979) 91: No. 213715
121. Crawford DB, Becker CL, LeBlanc JR (1979) US 416229. see BE 856919 CA (1978) 89:
No. 179566
122. Miyasugi T, Kosaka S, Kawai T, Suzuki A (1984) Ammoonia Plant Saf 24: 64
123. Pagani G, Brusasco G, Gramatica G (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur
Corp Int Conf Fert Technol, 4th, part 1, p 195
124. Nitrogen (1989) 179: 16
125. Bogart MJP (1973) US 3743488; CA (1973) 79: No. 106547
126. Bogart MJP (1974) US 3795485; CA (1974) 81: No. 15210
127. Nobles EJ (1973) In: Slack AV, James GR (eds) Ammonia, vol 1. Marcel Dekker, New York,
p275
128. Fuderer A (1987) US 4650651; CA (1987) 106: No. 179566
129. Herbort H-J, Marsch H-D (1987) (GB 2181740; see DE 3532413 CA (1987) 106: No. 158813
130. Marsch H-D, Thiagarajan N (1989) Ammonia Plant Saf 29: 195
131. Noyes R (1967) Ammonia and synthesis gas 1967. Noyes Development Corporation, Park
Ridge, p 78
132. Nitrogen (1962) 17: 35
133. Chern Eng Int Ed (1962) 69(14): 89
134. Chern Eng Int Ed (1966) 73(1): 24
135. Hydrocarbon Process (1984) 63(4): 103
136. Flytzani-Stephanopoulos M, Voecks GE (1981) Energy Prog 1(1-4): 52
137. Reidel JC (1954) Oil Gas J 52(40): 60
 Ammonia Production Processes 311

138. Reidel JC (1954) Oil Gas J 53(5): 86

139. Foster AL (1946) Oil Gas J (June 22): 98
140. Tuttle HA (1952) Chern Eng Prog 48(6): 272
141. Mayland BJ, Comley EA, Reynolds JC (1954) Chern Eng Prog 50(4): 177
142. Blanken JM (1988) Procestechnologie 2: 37
143. Eickmeyer AG, Marshall WH, Jr (1955) Chern Eng Prog 51(9): 418
144. Chopey NP (1961) Chern Eng (NY) 68(8): 158
145. Weyermuller G, Ogden JW (1952) Chern Process (London) (February): 36
146. Fricke H (1972) Chern Ztg 96(3): 123
147. Noyes R (1967) Ammonia and synthesis gas 1967. Noyes Development Corporation, Park
Ridge, p69
148. Lee KT, (1973) In: Slack AV, James GR (eds) Ammonia vol 1. Marcel Dekker, New York,
p293
149. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 165
150. Strelzoff S (1974) Hydrocarbon Process 53(12): 79
151. Nitrogen (1973) 83: 40
152. Shimizu Y (1978) CEER Chern Econ Eng Rev 10(7): 9
153. Jungfer H (1985) Linde Rep Sci Technol 40: 14
154. van den Berg GJ, Rijnaard P, Byrne DJ (1966) Hydrocarbon Process 45(5): 193
155. van den Berg GJ, Reinmuth E, Kapp E (1971) Chern Process Eng (London) 52(10): 49
156. ter Haar LW (1973) CEER Chern Econ Eng Rev 5(12): 22
157. ter Haar LW (1973) Het Ingenieursblad 21: 3
158. Reed CL, Kuhre CJ (1979) Hydrocarbon Process 58(9): 191
159. Banquy DL (1970) Fert News 15(11): 22
160. Banquy DL (1970) Proc Fert Soc London 117: 82
161. Becker PD, Hiller H, Hochgesand G, Sinclair AM (1971) Chern Process Eng (London) 52(11):
59
162. Haese E (1974) Chern Ind (Dusseldorf) 26(11): 732
163. Kammholz G, Muller G-A (1973) Erdiil Kohle, Erdgas, Petrochem Brenns Chern 26(12): 695
164. Butzert HE (1976) Chern Eng Prog 72(1): 56
165. Supp E, (1977) Nitrogen 109: 36
166. Morrison J (1969) Oil Gas J 67(8): 76
167. Morrison J (1969) Oil Gas Int 9(3): 78
168. Auer W (1971) Erdol Kohle, Erdgas Petrochem Brenns Chern 24(3): 145
169. Balz DF, Gettert HF, Gruendler KH (1983) PlantjOper Prog 2(1): 47
170. Happe U (1973) In: Slack AV, James GR (eds) Ammonia, vol 1. Marcel Dekker, New York,
p325
171. Forney AJ, Field JH (1973) In: Slack AV, James GR (eds) Ammonia, vol 1. Marcel Dekker,
New York, p 347
172. Noyes R (1967) Ammonia and synthesis gas 1967. Noyes Development Corporation, Park
Ridge p94
173. Jiintgen H, Heek van KH (1979) GWF-GasjErdgas 120 (12): 559 and (1980) 121(1): 6
174. Vorres KS (1980) Energy Res 4: 109
175. Lacey JA, Roberts GFI (1984) In: Proc World Pet Congr 11th 1983, London, p 1
176. Von Gratkowski H-W (1975) Erdiil Kohle, Erdgas, Petrochem Brenns Chern 28(2): 81
177. Wurzbacher G (1977) Chern Tech 6(8): 317
178. Nitrogen (1980) 126: 32
179. Ammonia Coal Symp (1979) Tennessee Valley Authority, Muscle Shoals, AL USA
180. Teggers H, Jiintgen H (1984) Erdol Kohle, Erdgas, Petrochem Brenns Chern 37(4): 163
181. Brown F (1977) Hydrocarbon Process 56(11): 361
182. Brown FC, Hargreaves HG (1979) Proc Fert Soc London 184: 1
183. Nitrogen (1986) 161: 23
184. Staege H (1979) Chern Eng (NY) 86(19): 106
185. Michaels HJ, Leonard HF (1978) Chern Eng Prog 74(8): 85
186. Staege H (1982) Hydrocarbon Process 61(3): 92
187. Sharpe RA (1976) Hydrocarbon Process 55(11): 171
188. Beck B, (1979) In: Ammonia Coal Symp. Tennessee Valley Authority, Muscle Shoals, AL USA,
p72
189. Engelbrecht AD, Partridge U (1979) In: Ammonia Coal Symp. Tennessee Valley Authority,
Muscle Shoals, AL USA P 176
312 Ib Dybkjaer

190. Patridge LJ (1976) Chern Eng Prog 72(8): 57

191. Engelbrecht AD, Partridge LJ (1981) Oil Gas J 79(6): 113
192. Chaurey KH (1981) In: FAI seminar 1980. The Fertiliser Association of India, New Delhi,
TECH 1-2/1
193. Chaurey KH, Sharma KC, (1979) In: Ammonia Coal Symp. Tennessee Valley Authority,
Muscle Shoals, AL USA, P 166
194. Valkel HK (1978) Energie 30(6): 196
195. Vogt EV, Burgt van der MJ (1980) Chern Eng Prog 76(3): 65
196. Burgt van der MJ (1979) Hydrocarbon Process 58(1): 161
197. Hydrocarbon Process (1984) 63(4): 96
198. IPG International Power Generation (1990) 13(4): 29
199. Hydrocarbon Process (1984) 63(4): 95
200. Schmidt-Traub H, Pohl HC (1983) Chern Ing Tech 55(11): 850
201. Schellberg W, Braun H (1988) Paper presented at symp on trends and developments in the
fertilizer industry. Bursa, Turkey, p I
202. Bailey E, Subramaniam TK, Anwer J, Bogner F (1979) In: Ammonia Coal Symp. Tennessee
Valley Authority, Muscle Shoals, AL USA, p 51
203. Theis KA, Nitschke E (1982) Hydrocarbon Process 61(9): 233
204. Franke FH, Pattas E, Nitschke E, Keller J (1979) In: Ammonia Coal Symp. Tennessee Valley
Authority, Muscle Shoals, AL USA, P 86
205. Hiller H (1975) Erdal Kohle, Erdgas, Petrochem Brenns Chern 28(2): 74
206. Becker PD (1979) In: Ammonia Coal Symp. Tennessee Valley Authority, Muscle Shoals, AI
USA, p44
207. Serrurier R (1976) Hydrocarbon Process 72(9): 253
208. King WEH (1981) Fuel 60: 803
209. Child ET (1979) In: Ammonia Coal Symp. Tennessee Valley Authority, Muscle Shoals, AL
USA, p79
210. Schlinger WG (1980) Energy Res 4: 127
211. CorniIs B, Hibbel J, Ruprecht P, DiirrfeId R, Langhoff J (1981) Hydrocarbon Process 60(1): 149
212. Konkol W, Ruprecht P, CorniIs B, Diirrfeld R, Langhoff J (1982) Hydrocarbon Process 61(3):
97
213. Sueyama T, Tsujino T (1985) In: More AI (ed) Proc Int ConfFertilizer 85 I Bf. Sulphur Corp.,
London, p 83
214. Jpn Chern Week (1984) (Nov. 15): 4.
215. Burnett JA, Jr (1979) In: Ammonia Coal Symp. Tennessee Valley Authority, Muscle Shoals, AL
USA, p27
216. Chern Eng News (1979) 57(23): 27
217. Alves GW, Waitzmann DA (1982) Erdal Kohle, Erdgas, Petrochem Brenns Chern 35(2): 70
218. Watson JR, McClanahan TS, Weatherington RW (1985) Altern Energy Sources 3: 407
219. Lee RG (1981) In: FAI Seminar 1980, The Fertiliser Association of India, New Delhi, part II,
TECH 1-3/1
220. Alpert SB, Gluckman MJ (1986) Ann Rev Energy II: 315
221. Bohlbro H (1969) An investigation on the kinetics of the conversion of carbon monoxide with
water vapour over iron oxide based catalysts, 2nd ed. Gjellerup, Copenhagen
222. Moe JM (1962) Chern Eng Prog 58(3): 33
223. Ting AP, Wan S-W (1969) Chern Eng (NY) 76(11): 185
224. Chern Eng News (1963) 41(7): 46
225. Habermehl R (1965) Chern Eng Prog 61(1): 57
226. Habermehl R, Atwood K (1965) Paper presented at United Nations Interregional Seminar in
the Production of Fertilizers, Kiev, USSR
227. Nitrogen (1966) 40: 28
228. Allen D (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York, p 3
229. Mertzsch N (1979) Chern Tech Umsch 11(2): 5
230. Newsome DS (1980) Catal Rev Sci Eng 21(2): 275
231. Hawker PN (1982) Hydrocarbon Process 61(4): 183
232. Valter J, Berndt H, Lietz G (1976) Chern Tech 28(19): 606
233. Lloyd L, Twigg MV (1979) Nitrogen 118: 30
234. Nitrogen (1987) 168: 29
235. Lombard JF (1969) Hydrocarbon Process 48(8): 111
 Ammonia Production Processes 313

236. Campbell JS (1970) Ind Eng Chern Process Des Dev 9(4): 588
237. Young PW, Clark CB (1973) Chern Eng Prog 69(5): 69 and Ammonia Plant Saf 15: 18
238. Rostrup-Nielsen JR, Hejlund Nielsen PE (1985) In: Diidar J, Wise H (eds.) Deactivation and
poisoning of catalysts. Marcel Dekker, New York, p 259
239. Kitchen D (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf, Geneva, p 127
240. Pedersen PS (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf Geneva, p 111
241. Lundberg WC (1979) Chern Eng Prog 75(6): 81 and Ammonia Plant Saf 21: 105
242. Lorenz E (1967) Ind Chim Beige 32: (Spec No.) (Pt. 2) 377
243. Auer W, Lorenz E, Griindler KH (1971) Paper presented at The 68th National Meeting of the
American Institute of Chemical Engineers, Houston, USA
244. Dybkjaer I (1979) In: Ammonia Coal Symp. Tennessee Valley Authority, Muscle Shoals, AL
USA, p 133
245. Dybkjaer I, Bohlbro H, Aldridge CL, Riley KL (1979) Ammonia Plant Saf 21: 145
246. Dybkjaer I (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp Int ConfFert Technol,
4th, part II, p 503
247. Hejlund-Nielsen PE, Begild-Hansen J (1982) J Mol Catal17: 183
248. Hejlund-Nielsen PE, Begild-Hansen J (1981) Paper presented at Communicacao ao Coloquio
Nacional de Catalise Industrial, Lisboa
249. Hansen JB, Carstensen JH, Pedersen PS (1989) Ammonia Plant Saf 29: 204
250. Kitchen D, Pinto A, Praag H van (1989) Ammonia Plant sar 29: 212
251. Carstensen JH, Begild-Hansen J, Pedersen PS (1990) Ammonia Plant Saf 30: 139
252. Buckthorp CM (1978) Nitrogen 113: 34
253. Bonacci JC, Otchy TG (1978) Ammonia Plant Saf 20: 165
254. Colby JH, White GA, Notwick PN Jr (1979) Ammonia Plant Saf 21: 138
255. Krishnaswami KR, Neelakantan PS (1989) Fert. News (December) 31
256. Cover AE, Hubbard DA, Jain SK, Shah KV, Koneru PB, Won EW (1985) Review of selected
gas removal processes for SNG production. Gas Research Institute, Contract No. 5082-222-
0754, p 293
257. Christensen KG, Stupin WJ (1978) FE-2240-49, Report to the United States Department of
Energy and Gas Research Institute. Contract No. EX-76-C-OI-2240
258. Kohl AL, Riesenfeld FC (1979) Gas Purification, 3rd ed. Gulf Publishing Company, Houston
259. Schmidt HW, Henrici H-J (1972) Chern Ztg 96(3): 154
260. Goar BG (1971) Oil Gas J 69(28): 75, 69(29): 84
261: Hochgesand G (1968) Chern Ing Tech 40(9/10): 432
262. Werner D (1981) Chern Ing Tech 53(2): 73
263. Christensen KG, Stupin WJ (1978) Hydrocarbon Process 57(2): 125
264. Tennyson RN, Schaaf RP (1977) Oil Gas J 75(2): 78
265. Schmidt HW (1968) Chern Ing Tech 40(9/10): 425
266. Thirkell H (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 117
267. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 193
268. Strelzoff S (1975) Chern Eng (NY) 82(19): 115
269. Stokes KJ (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp Int Conf Fert Technol
4th, part II, 525
270. Stokes KJ (1980) Ammonia Plant Saf 22: 178
271. Stokes KJ (1981) Nitrogen 131: 35
272. Brown FC, Leci CL (1982) Proc Fert Soc London 210: 1
273. Jackson JM (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 183
274. Field JH (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 153
275. Butwell KF, Kubek DJ (1977) Hydrocarbon Process 56(10): 173
276. Nitrogen (1975) 96: 33
277. Nitrogen (1976) 102: 40
278. Butwell KF, Kubek DJ, Sigmund PW (1979) Chern Eng Prog 75(2): 75; Ammonia Plant Saf21:
156
279. Pierce JD, Chao EI (1985) Paper presented at Ammonia Saf Symp, Seattle paper 30e, p 1
280. Linsmayer S (1972) Chern Tech 24(2): 74
281. Elberling K, Gabriel W (1977) Chern Tech 29(1): 43
314 Ib Dybkjaer

282. Maddox RN, Burns MD (1968) Oil Gas J 66(4): 91

283. Giammarco G (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New
York, p 171
284. Hydrocarbon Process (1982) 61(4): 95
285. Eickmeyer AG (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New
York, p 165
286. Benson HE, Parrish RW (1974) Hydrocarbon Process 53(4): 81
287. Bartoo RK, Ruzicka SJ (1983) In: Fertilizer 83, Br Sulphur Corp 7th Int Conf, London, p 129
288. Benson HE (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 159
289. Grover BS, Holmes ES (1986) In: Proc Nitrogen 86 Amsterdam, Br Sulphur Corp, London
p 101
290. van Hecke F (1977) US 4035166; CA (1977) 86: No. 173649
291. Crabs LE, Pouilliart R, van Hecke FC (1980) Ammonia Plant Saf 22: 185
292. Bogner F (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 205
293. Zarker KE (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 219
294. Sehrt B, Polster P (1980) Chern Tech 32(7): 345
295. Hochgesand G (1970) Ind Eng Chern 62(7): .17
296. Haase H (1970) Chern Anlagen Verfahren 10: 59
297. Linde W (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 233
298. Chern Eng (NY) (1979) 86(26): 88
299. Lewis JL, Truby HA, Pascoo MB (1974) Oil Gas J 72(25): 120
300. Bucklin RW, Schendel RL (1984) Energy Prog 4(3): 137
301. Cook JP (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p249
302. Dailey LW (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p243
303. Kohrt H-U, Thormann K, Bratzler K (1963) ErdOi Kohle, Erdgas, Petrochem 16(2): 96
304. Judd DK (1978) Hydrocarbon Process 57(4): 122
305. Swanson CG Jr. (1979) Ammonia Plant Saf 21: 152
306. Swanson CG Jr, Burkhard FC (1984) Ammonia Plant Saf 24: 16
307. Shah VA (1988) Energy Prog 8(2): 67
308. Shah VA, McFarland J (1988) Hydrocarbon Process 67(3): 43
309. Hernandez RJ, Huurdeman TL(1988) In: Nitrogen 88 Br Sulhur's 12th Int Conf, Geneva p 147
310. Fertilizer Focus (1988) 5(11): 27
311. Volkamer K, Wagner E, Schubert F (1982) Plant/Oper Prog 1(2): 134
312. Strelzoff S (1981) Technology and manufacture of ammonia. J Wiley, New York, p 203
313. Strelzoff S (1974) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker,New York,
p 201
314. Meissner RE, Wagner U (1983) Oil Gas J 81(5): 55
315. Volkamer K, Wagner U (1983) In: Fertilizer 83, Br Sulphyr Corp, 7th Int Conf, London,
p 139
316. Meissner RE, Parsons RM (1984) Energy Prog 4(1): 17
317. Daviet GR, Sundermann R, Donnelly ST, Bullin JA (1984) Hydrocarbon Process 63(5): 79
318. Nobles IE, Shay LN, Stover JC (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf Geneva
p 243
319. Gerhardt W, Hefner W (1989) Ammonia Plant Saf 29: 73
320. Meissner H, Hefner W (1990) Paper presented at European Conference on Energy Efficient
Production of Fertilizers, Bristol UK
321. Allgood HY (1974) In: Slack AV, James GR (ed) Ammonia, vol 2. Marcel Dekker, New York,
p 289
322. Yeandle WW, Klein GF (1952) Chern Eng Prog 48(7): 349
323. Egalon R, Vanhille R, Willemyns M (1955) Ind Eng Chern 47(5): 887
324. Grotz BJ Jr (1969) (US 3442613: CA (1967) 67: No. 66185
325. Nitrogen (1983) 144: 30
326. Grotz BJ (1967) Hydrocarbon Process 46(4): 197
 Ammonia Production Processes 315

327. Grotz BJ (1976) Nitrogen 100: 71

328. Slack AV (1977) In: Slack AV; James GR (ed) Ammonia, vol 2. Marcel Dekker, New York,
p 361
329. Grotz BJ, Good G (1980) Chern Age (London) (November 14): 18
330. Wilson KC, Grotz BJ, Bhakta ML, Gosnell JH (1984) Nitrogen 151: 31
331. Fertilizer Focus (1987) 4(10): 47
332. Morrison J (1968) Oil Gas Int 8(9): 84
333. Verduijn WD (1979) Ammonia Plant Saf 21: 130
334. Henstock WH (1986) PlantlOper Prog 5(4): 232
335. Baker DF (1955) Chern Eng Prog 51(9): 399
336. Joly A, Clar R (1968) Nitrogen 52: 30
337. Fabian R, Fiirg W (1975) Linde Rep Sci Tehcnol 22: 9
338. Scholz WH (1982) Linde Rep Sci Technol 34: 3
339. Rowles HC (1974) In: Slack AV, James GR (ed) Ammonia, vol 2. Marcel Dekker, New York,
p 321
340. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 254
341. Taffe P, Joseph J (1978) Chern Age (London) (October 20): 14
342. Savage PR (1978) Chern Eng (NY) 85(25): 68D
343. ECN Eur Chern News (1978) (October 20): 39
344. Rehder R, Stead P(1986) In: FAI Seminar 1985. The Fertiliser Association ofIndia, New Delhi,
TECH II-2/1
345. van Weenen WF, Tielrry J (1980) Nitrogen 127: 38
346. Oil Gas J (1981) 19(18): 270
347. van Weenen WF, Tielrooy J (1980) Proc Fert Soc London 191: 1
348. van Weenen WF, Tielrooy J (1980) Chern Age India 31(12): DEV-2/1
349. Voogd J (1986) In: FAI Seminar 1985. The Fertiliser Association ofIndia, New Delhi, TECH
11-1/1
350. Halstead JM, Haslett AM, Pinto A.(1989) In: FAI Seminar 1988. The Fertiliser Society ofIndia,
New Delhi, SII/2, p 1
351. Nitrogen (1989) 178: 30
352. Royse S (1989) Process Eng (London) 70(3): 36
353. Short H (1989) Chern Eng (NY) 96(7): 41
354. Fert Int (1989) 272 (April): 37
355. Fertilizer Foucs (1989) 6(7): 42
356. Hicks TC, Moss JMS, Pinto A (1989) Paper presented at AIChE Ammonia Saf Symp. San
Francisco, p 1
357. Hicks TC, Pinto A (1989) Fert News (December): 37
358. Corr F, Dropp F, Rudelstorfer E (1979) Hydrocarbon Process 58(3): 19
359. Werner DH, Schlichtharle GA (1980) Ammonia Plant Saf 22: 12
360. Nitrogen (1979) 121: 37
361. Heck JL (1980) Oil Gas J 78(6): 122
362. Allen DW, Yen WH (1973) Chern Eng Prog 69(1): 75; Ammonia Plant Saf (1973) 15: 96
363. Philips JR (1974) In: Slack AV, James GR (ed) Ammonia, vol 2. Marcel Dekker, New York,
p 311
364. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 267
365. Toering W (1985) Ammonia Plant Saf 25: 64; PlantlOper Prog 4(3): 127
366. Chern Metall Eng (1943) (November): 152
367. Karwat E (1964) Erdiil Kohle, Erdgas, Petrochem 17(6): 439
368. Nitrogen (1968) 55: 39
369. Wagener D, Meckel FJ, Laue KH (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp
Int Fert Technol 4th, part II, p 471
370. Blanco RE, Holmes JM, Salmon R, Ullman JW (1967) Chern Eng Prog 63(4): 46
371. Nitrogen (1975) 97: 35
372. Grundt T, Christiansen K (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp Int Conf
Fert Technol, 4th, part I, p 73
373. Grundt T, Christiansen K (1981) Adv Hydrogen Energy 2: 1247
374. Hinrichs H, Niedtzky J (1962) Chern Ing Tech 34(12): 819
375. Paiva CR, Fanti OD (1987) Nitrogen 170: 31
376. Pfeiffer C, Sandler HJ (1955) Pet Refin 34(5): 145
316 Ib Dybkjaer

377. Arnold JH, Dixon WT, (1955) Oil Gas J 54(30): 90

378. Czuppon TA, Lee JM (1989) In: Fertilizer Latin America Int Conf, vol 2. Br Sulphur Corp,
Caracas, p 389
379. Fertilizer Focus (1989) 6(7): 46
380. Bogner F (1958) Chern Ing Tech 30(6): 382
381. Blaskowski HJ (1968) US 3404957; CA (1969) 70: No. 5585
382. Huebler J, Johnson JL, Schora C Jr, Tarman PB, Panos PS (1970) US 3539292; CA (1971) 74:
No. 14671
383. Buskies U, Summers F (1978) Erdol Kohle, Erdgas, Petrochem 31(10): 474
384. Sauchelli V (ed) (1964) Fertilizer nitrogen its chemistry and technology. Reinhold Publishing
Corporation, New York, p 80
385. Hein LB (1952) Chern Eng Prog 48(8): 412
386. Vancini CA (1961) La Sintesi dell'Ammoniaca, Ulrico Hoepli, Milano, p 497
387. Kennedy JL (1967) Oil Gas J 65(46): 105,65(48): 95, 65(1): 72, 65(51): 76, (1968) 66(2): 65, (1968)
66(4): 76
388. Fraschetti F, Filippini U, Ferrara PL (1967) Quad. Pignone 9: 5
389. Begg GAJ (1968) Chern Process Eng (London) 49(1): 58
390. Viti A (1971) Het Ingenieursblad 40(21): 619
391. Ferrara PL, Tesei A (1978) Quad Pignone 25: 131
392. Salvi ani J, Fraschetti F, Ferrara PL, Filippini U, Wickl R, Hile HB, Zech WA, Purcell VP (1977)
In: Slack AV, James GR (ed) Ammonia, vol 2. Marcel Dekker, New York, p 1
393. Nitrogen (1968) 55: 37
394. Labrow S (1968) Chern Process Eng (London) 49(1): 55
395. Forster H (1971) Chern Tech 23(2): 93; (3): 157
396. Plotner W (1972) Chern Tech 24(6): 324
397. Nitrogen (1989) 181: 23
398. Nobles JE, Stover JC (1984) Chern Eng Prog 80(1): 81
399. Caplow SD, Bresler SA (1967) Chern Eng (NY) 74(7): 103
400. Stokes KJ (1979) Chern Eng Prog 75(7): 88
401. Twist DR, Standbridge DW (1968) GB 1134621; CA (1969) 70: No. 59416
402. Uji S, Ikeda M (1981) Hydrocarbon Process 60(7): 94
403. Bakemeier H, Huberich T, Krabetz R, Liebe W, Schunck M (1985) In: Ullman's Encyclopedia
of industrial chemistry, vol 42, 5th ed VCH Verlagsgesellschaft mbH, Weinheim, p 187
404. Quartulli OJ, Fleming JB, Finneran JA (1968) Hydrocarbon Process. 47(11): 153
405. Quartulli OJ, Fleming JB, Finneran JA (1969) Nitrogen 58: 25
406. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p 231
407. Sauchelli V (ed) (1964) Fertilizer nitrogen its chemistry and technology. Reinhold Publishing
Corporation, London, p 81.
408. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 2. Marcel Dekker, New York,
p 326
409. Wilson KC, Grotz BJ, Richez J (1986) In: Proc Nitrogen 86 Amsterdam, Br Sulphur Corp
London, p 63
410. Nitrogen (1973) 82: 34
411. Glover W, Yoars JP (1973) Hydrocarbon Process 52(4): 165
412. Glover WA, Yoars JP (1973) Ammonia Plant Saf 15: 77
413. Process Eng (London) (1983) 64(3): 15
414. Chern Eng (London) (1983) 390: 30
415. Trotter SG, Pinto A, (1987) Fert News (India) (December) 29
416. Livingstone JG, Pinto A (1983) Chern Eng Prog 79(5): 62
417. Nitrogen (1986) 162: 27
418. Fertilizer Focus (1987) 4(10): 40
419. Fertilizer Focus (1987) 4(10): 24
420. Brown FC (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp Int ConfFert Technol,
4th part I, p 39
421. LeBlanc JR Jr, Madhavan S, Porter RE (1978) In: Kirk-Othmer Encyclopedia of chemical
technology, vol 2 3rd ed Wiley, New York, p 495
422. Finneran JA, Quartulli OJ (1977) In: Slack AV, James GR (ed) Ammonia, vol 3. Marcel
Dekker, New York, p 159
 Ammonia Production Processes 317

423. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p324
424. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p 241
425. Fertilizer Focus (1987) 4(10): 36
426. Tops0e HFA, Poulsen HF, Nielsen A (1967) Chern Eng Prog 63(10): 67
427. Nitrogen (1964) 31: 22
428. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p 245
429. Finneran JA, Mayo HC (1967) US 3350170; CA (1968) 68: No. 4558
430. Chern Eng (NY) (1967) 74(24): 112
431. Chakley DE, Turner W (1967) Oil Gas Int 7(10): 49
432. Axelrod LC, Quartulli OJ, Turner W (1966) Europe Oil 5(3): 58
433. Pet Process (1956) 10: 167
434. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p303
435. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 310
436. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 319
437. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 341
438. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 350
439. Spielman M, Baumann GP, Hering B (1968) US 3388968; CA (1968) 69: No. 44970
440. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p234
440. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p 237
442. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 313
443. Mundo K-J, (1972) Chern Anlagen Verfahren 6: 49
444. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
. p 331
445. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p248
446. Shearon WH Jr, Thompson HL (1952) Ind Eng Chern 44(2): 254
447. Thompson HL, Guillaumeron P, Updegraff NC (1952) Chern Eng Prog 48(9): 468
448. Nitrogen (1987) 169: 31
449. Neth N, Puhl H, Liebe W (1982) Chern Eng Prog 78(7): 69
450. Landrum LH (1984) Ammonia Plant Saf 24: 22
451. Vancini CA (1971) Synthesis of ammonia. The Macmillan Press Ltd, London and Basingstoke,
p 122
452. Bendix H, Lenz L (1989) Ammonia Plant Saf 29: 221
453. Ripps DL (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 235
454. Kjaer J (1966) Chern Tech (Berlin) 18(3): 138
455. Almasy GA, Hay 11, Pallai 1M (1966) Br Chern Eng 11(3): 188
456. Almasy G, Hay J, Jedlovszky P, Pallai, I (1966) Int Chern Eng 6(2): 233
457. Rase HF (1977) Chemical reactor design for process plants, vol 2. Wiley, New York, p 61
458. Kjaer J (1985) Comput Chern Eng 9(2): 153
459. Dybkjaer I (1986) In: De Lasa HI (ed) Chemical Reactor Design and Technology, Nasa ASI
Series, Series E: Applied Sciences No. 110. Martinus Nijhoff Publishers, Dordrecht, p 795
460. Dybkjaer I, Gam EA (1985) Ammonia Plant Sat 25: 15
461. Dybkjaer I, Gam EA (1984) CEER Chern Econ Eng Rev 16(9): 29
462. Jarvan JE (1978) Oil Gas J 76(5): 178
463. Kjaer J (1979) Measurement and calculation of temperature and conversion in fixed-bed
catalytic reactors, 2nd ed Haldor Tops0e A/S, Vedbaek, Denmark
464. Hlavacek V (1970) Ind Eng Chern 62(7): 8
318 Ib Dybkjaer

465. Froment GF (1974) Chern Ing Tech. 46(9): 374

466. Jarvan JE (1970) Ber Bunsen Ges Phys Chern 74(2): 141
467. Kjaer J (1972) Computer methods in catalytic reactor calculations. Haldor Tops0e, Vedbaek
Denmark
468. Jarvan JE (1978) Oil Gas J 76(3): 51
469. Hay JJ, Pallai 1M (1963) Br Chern Eng 8(3): 171
470. Bakemeier H, Detzer H, Krabetz R (1965) Chern Ing Tech 37(4): 429
471. Fodor L (1971) Chim Ind Genie Chim 104(8): 1002
472. Singh CPP, Saraf DN (1979) Ind Eng Chern Process Des Dev 18(3): 364
473. Lucas K, Gelbin D (1962) Br Chern Eng 7(5): 336
474. Shah MJ (1967) Ind Eng Chern 59(1): 72
475. Gaines LD (1977) Ind Eng Chern Process Des Dev 16(3): 381
476. Gaines LD (1979) Chern Eng Sci 34: 37
477. Reddy KV, Husain A (1982) Ind Eng Chern Process Des Dev 21(3): 359
478. Shipman LM, Hickman JB (1968) Chern Eng Prog 64(5): 59
479. Siminiceanu I, Petrila C, Pop A (1983) Chern Tech 35(12): 628
480. Baddour RF, Brian PLT, Logeais BA, .Eymery JP (1965) Chern Eng Sci 20: 281
481. Brian PLT, Baddour RF, Eymery JP (1965) Chern Eng Sci 20: 297
482. Almasy GA, Jedlovszky P, Pallai 1M (1967) Br Chern Eng 12(8): 1219
483. Porubszky I, Simonyi E, Ladanyi G (1969) Br Chern Eng 14(4): 495
484. Murase A, Roberts HL, Converse AO (1970j.lnd Eng Chern Process Des Dev 9(4): 503
485. Balzer D, lijin B, Reinig G, Scholz V (1971) Chern Tech 23(9): '513
486. Lewin DR, Lavie R (1984) Ind Chern Eng Symp Ser 87: 393
487. Nielsen A (1953) In: Advances in catalysis, vol V. Academic Press Inc, New York, p 1
488. Nielsen A (1968) An investigation on promoted iron catalysts for the synthesis of ammonia, 3rd
ed. Jul. Gjellerups Forlag, Copenhagen
489. Nielsen A (1981) Catal Rev Sci Eng 23(1,2): 17
490. Zardi U (1982) Hydrocarbon Process 61(8): 129
491. Nitrogen (1972) 75: 33
492. Nitrogen (1982) 140: 30
493. Stre1zoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 23
494. Bakemeier H, Huberich T, Krabetz R, Liebe W, Schunck M (1985) In: Ullman's Ency-
clopedia of industrial chemistry, vol A2. 5th ed. VCH Verlagsgesellschaft mbH, Weinheim,
p 190
495. Worn GA (1962) US 3041150; CA (1963) 58(8): No. 10997
496. Slack AV, Allgood HY, Maune HE (1953) Chern Eng Prog 49(8): 393
497. Vancini CA (1961) La Sintesi dell'Ammoniaca. Ulrico Hoepli, Milano, p 951
498. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 239
499. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 353
500. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 333
501. de Rycker HGEL, Mathieu MFVG (1959) US 2887365; CA (1959) 53: No. 12766
502. Vek V (1988) Paper presented at 7th Int Symp Large Chemical Plants 1988. Bruges, Belgium,
p77
503. Gmelins Handbuch der Anorganischen Chemie (1935) Stickstoff, nitrogen system 4, vol 2.
Verlag Chemie, Berlin, p 405
504. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 235
505. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 249
506. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 232
507. Gmelins Handbuch der Anorganischen Chemie (1935) Stickstoff, nitrogen system 4, vol 2.
Verlag Chemie, Berlin, p 405
508. Pinto A, Ward SA (1980) US 4215099; see GB 1565824 CA (1980) 93: No. 134577
509. Schober R (1962) US 3031274; CA (1961) 55: No. 6962
510. Dohnalek R, Hribal J (1965) NE 6400361; CA (1966) 64: No. 10809
511. Breus YV (1965) SU 170917; CA (1965) 63: No. 9516
512. Christensen A, Rayfiel RD (1958) US 2953371; CA (1959) 53: No. 6707
513. Christensen A (1962) US 3041151; CA (1962) 56: No. 12706
 Ammonia Production Processes 319

514. Christensen A (1962) US 3050377; CA (1962) 57: No. 9623

515. Worn GA (1958) US 2861873; CA (1959) 53: No. 3797
516. Vilceanu M, Bors C (1976) US 3932139; see FR 2141488 CA (1973) 79: No. 94152
517. Hennel W (1964) Br Chern Eng 9(5): 290
518. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 358
519. Takagi T (1968) Jpn Chern Q 4(4): 47
520. Ohsaki K, Zanma J, Watanabe H (1984) DE 3334775; CA (1984) 101: No. 40286
52l. Papillon A, Lesur P, Faury C, Badoual C, Lafleur G (1989) EP 314550 CA (1989) 111:
No 80790
522. Lee JA (1945) Chern Metall Eng (December) 94
523. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 240
524. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 244
525. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 312
526. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 236
527. Eschenbrenner GP, Honigsberg CA (1969) US 3475136; see FR 1528951 CA (1969)
71: 14502
528. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 34
529. Lesur P (1977) Nitrogen 108: 29
530. Kubec J, Saroch V (1974) US 3784362; see FR 2046522 CA (1971) 75: No. 131039
53l. Kubec J, Saroch V (1972) US 3653846; see DE 1927850 CA (1971) 72: No. 68564
532. Eschenbrenner GP, Wagner III GA (1972) Chern. Eng. Prog. 68(1): 62; Ammonia Plant Saf
(1971) 14: 51
533. Axelrod LC, Eagle RS, Fleming JB, Kahn GM, Quartulli OJ (1971) US 3567404; CA (1971) 74:
No. 143654
534. Wett T (1971) Oil Gas J 69(41): 70
535. Chern Eng (NY) (1971) 78(24): 90
536. Quartulli OJ, Wagner II GA, (1978) Hydrocarbon Process 57(12): 115
537. Fawcett R, Smith AW, Westwood D (1972) US 3694169; see DE 2025486 CA (1971) 74:
No. 113691
538. Ridler DE (1972) Ammonia Plant Saf 14: 57
539. Mehta DD, Miller EJ (1972) US 3663179; see DE 2119127 CA (1972) 76: No. 26712
540. Nielsen A (1969) Chim Ind (Milan) 51(10): 1052
54l. Vek V (1973) Chern Ing Tech 45(9, 10): 608
542. Vek V (1977) Ind Eng Chern Process Des Dev 16(3): 412
543. Pet Int (1974) 14(11): 72
544. Hansen HJ (1968) US 3372988; CA (1965) 63: No. 14703
545. Chern Week (1966) 98(9): 97
546. Nielsen A (1972) Ammonia Plant Saf 14: 46
547. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 345
548. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York,
p 354
549. Hinrichs H, Lehner H, Wagner A, Pesl J, Niedetzky J, Faschinger F (1973) US 3754078, see FR
2089755 CA (1972) 77: No. 93392
550. Kohn HB, Friedman G (1975) US 3918918; CA (1976) 84: No.166677
55l. Zardi U (1983) US 4372920; see DE 3026199 CA (1981) 94: No. 123623
552. Zardi U, Pagani G (1988) EP 287765; CA (1989) 110: No. 40965
553. Zardi U, Commandini E, Gallazzi C (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur
Corp Int Fert Technol, 4th part I, p 173
554. Fauser G (1959) US 2898183; CA (1959) 53: No. 19489
555. Vancini CA (1971) Synthesis of ammonia. Macmillan London p 233
556. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 27
557. Hinrichs H, Niedtzky J (1962) Chern Ing Tech 34: 88
558. Graeve HW (1981) Chern Eng Prog 77(10): 54
559. Graeve HW (1981) Ammonia Plant Saf 23: 78
560. Ind Eng Chern (1962) 54(5): 11
320 Ib Dybkjaer

561. Nitrogen (1962) 17: 23

562. Nitrogen (1966) 42: 40
563. Nitrogen (1973) 81: 29
564. Hinrichs H, Lehner H (1972) Chern Anlagen Verfahren 6: 55
565. Oesterreichische Stickstoffwerke A-G (1966) US 3440021; CA (1977) 66: No. 12354
566. Forster F (1980) Chern Eng 87(9): 62
567. Strelzoff S (1981) Technology and manufacture of ammonia. Wiley, New York, p 26
568. Slack AV (1977) In: Slack AV, James GR (eds) Ammonia, vol 3. Marcel Dekker, New York, p 315
569. Hennel W, Sobolewski C, Bartoski Z (1975) US 3892535; see DE 2323678 CA (1974) 80:
No. 147234
570. Handman SE, Leblanc JR Jr (1983) Chern Eng Prog 79(5): 56
571. Peterson R, Finello R, Denavit G (1985) EP 134260; CA (1985) 102: No. 187456
572. Wright LE, Pickfor AE (1973) US 3721532; CA (1973) 79: No. 33234
573. Wright LE, Pickford AE (1974) US 3851046: CA (1975) 82: No. 142248
574. Grotz BJ (1988) US 4744966; CA (1988) 109: No. 57548
575. Grotz BJ (1988) EP 268469; CA (1988) 109: No. 57554
576. Wilson KC, Grotz BJ (1990) J Tech Dev 14: 54
577. Ward SA (1979) US 4180543; see DE 2815856 CA (1979) 90: No. 123588
578. LeBlanc JR, Peterson RB (1981) US 4298589; CA (1982) 96: No. 37843
579. Nitrogen (1976) 101: 42
580. Topsoe HFA, Gam EA (1980) US 4181701; see DE 2710247 CA (1978) 88: No. 52614
581. Haldor Topsoe A/S (1981) CA 1112847; see DE 2710247 CA (1978) 88: No. 52614
582. Zardi U, Commandini E (1983) US 4405562; see DE 3146778 CA (1983) 98: No. 56154
583. Zardi U (1987) EP 222069; CA (1987) 107: No. 98608
584. Weicken R (1986) Chern Ind (Berlin) 8: 674
585. Uhde Engineering News (1987) 1-87: 1
586. Hakmann R, Niibel W, Weicken R, Lehner H (1986) Ammonia Plant Saf 26: 44
587. Eagle RA, Patel VA (1980) US 4230669; see EP 7743 CA (1980) 93: No. 28739
588. Becker CL, LeBlanc JR (1980) JS 4230680; see EP 7743 CA (1980) 93: No. 28739
589. Pagani G (1978) US 4101281; see DE 2631898 CA (1977) 86: No. 123879
590. Nitrogen (1987) 165: 27
591. Shannahan N (1989) Hydrocarbon Process 68(1): 60
592. Peterson RB (1984) US 4482523; CA (1985) 102: 26997
593. Nitrogen (1986) 162: 35
594. Howerton RG, Noe SA (1989) Ammonia Plant Saf 29: 157
595. Dybkjaer I (1989) In: Fertilizer Latin America International Conference, vol 1. Br Sulphur
Corp, Caracas, p 77
596. Novacek D, Newland RL, Horsey JF (1990) Ammonia Plant Saf 30: 161
597. Fert News (1987) (December): 19
598. Brown FC, Zardi U, Pangani G (1988) In: Nitrogen 88 Br Sulphur's 12th Int ConfGeneva, p 159
599. Nitrogen (1987) 169: 33
600. Comandini E, Zardi U (1989) In: Fertilizer Latin America Int Conf vol 1. Br Sulphur Corp,
Caracas, p 113
601. Sutherland L, Wallace B (1989) Ammonia Plant Saf 29: 167
602. Pagani G, Zardi U (1989) In: FAI Seminar 1988. The Fertiliser Society ofIndia, New Delhi,
SII/I-1/9
603. Newland RL, Pierce JD, Borzik DM (1988) Ammonia Plant Saf 28: 53
604. Axelrod LC, O'Hare TE (1964) In: Sauchelli V (ed) Fertilizer nitrogen its chemistry and
technology. Reinhold Publishing Corporation, New York, p 78
605. Hinrichs H (1962) Ind Chern (January): 7
606. Hinrichs H (1963) Ind Chern (October): 519
607. Bogner F, Karsch D (1965) Nitrogen 37: 21
608. Lachmann H, Fromm D (1988) Ammonia Plant Saf 28: 91
609. Prescott GR, Podhorsky M, Blommaert P (1988) Ammonia Plant Saf 28: 135
610. Podhorsky M, Prescott GR, Blommaert P, Grisola L (1989) Ammonia Plant Saf 29: 253
611. Stahl H (1982) Proc Fert Soc London 206: 61
612. Weber H, Graeve HW, Herbort H-J, Marsch H-D (1984) Chern Ing Tech 56(5): 356
613. Johnson DL, O'Connor TM (1987) Ammonia Plant Saf 27: 27
614. Cocchi V, Mezzedimi V (1983) Quad Pignone 36: 17
615. Nitrogen (1988) 176: 25
 Ammonia Production Processes 321

616. ECN Eur Chern News (1966) (August 19): 34

617. Silberring L (1985) Nitrogen 157: 33
618. Vancini CA (1961) La Sintesi dell'Ammoniaca. Ulrico Hoepli, Milano, p 769
619. Nitrogen (1962) 16: 48
620. Steels for hydrogen service at elevated temperatures and Pressures in petroleum refineries and
petrochemical plants. API publication 941, (1990) 4th ed, American Petroleum Institute
621. Bonner WA (1977) Hydrocarbon Process 56(5): 165
622. Kobrin G, Kopecki ES (1978) Chern Eng (NY) 85(28): 115
623. Prescott GR, Badger FW (1980) Ammonia Plant Saf 22: 146
624. Prescott GR (1982) Plant/Oper Prog 1(2): 94
625. Prescott GR, Blommaert P, Grisola L (1986) Ammonia Plant Saf 26: 228
626. Marsch HD (1982) Plant/Oper Prog 1(3): 152
627. Sathe SY, O'Connor T (1988) Ammonia Plant Saf 28: 115
628. van Grieken CA (1989) Ammonia Plant Saf 29: 145
629. Zeis LA (1975) Ammonia Plant Saf 17: 56
630. Ueda S, Onishi H, Okubo M, Takegawa T (1978) Ammonia Plant Saf 20: 98
631. van Grieken CA (1979) Ammonia Plant Saf 21: 18
632. Phillips BR (1979) Ammonia Plant Saf 21: 21
633. Orbons HG, Huurdemann TL (1985) Ammonia Plant Saf. 25: 89; Plant/Oper Prog (1985) 4(1):
49
634. Cermak J, Burex Z (1967) Br Chern Eng 12(9): 1372
635. Bresler SA (1977) US 4056603; CA (1978) 88: No. 25032
636. Lipman M (1979) Hydrocarbon Process 58(1): 153
637. Bogart MJP (1981) In: More AI (ed) Fert nitrogen: Proc Br Sulphur Corp Int Fert Technol, 4th
part I, p 141
638. Bogart MJP (1982) Plant/Oper Prog 1(3): 147
639. Pagani G, Zardi U (1972) Hydrocarbon Process 51(7): 106
640. Lavie R (1985) Chern Eng Sci 40(11): 2019
641. Lavie R (1987) Plant/Oper Prog 6(2): 118
642. Lavie R (1985) US 4537760; see FR 2511666 CA (1983) 98: No. 218143
643. Brubaker DW, Kammermeyer K (1954) Ind Eng Chern 46(4): 733
644. Laciak DV, Pez GP (1988) US 4758250; CA (1988) 109: No. 152435
645. Pez GP, Laciak DV (1988) US 4762535; CA (1988) 109: No. 131707
646, Otto W (1972) Chern Tech 24(8): 489
647. Parrish RW (1985) Ammonia Plant Saf 25: 162
648. Wang SI, Nicholas DM, Dimartino SP (1984) Oil Gas J 82(6): 111
649. Merlo M, Saviano F (1985) Fert Ind 63
650. Isalski WH (1984) Nitrogen 152: 100
651. Oil Gas J (1979) 77(10): 182
652. Banks R (1977) Chern Eng (NY) (October 10): 90
653. Isalski WH (1979) Chern Age (London) (November 30): 19
654. Haslam A, Brook P, Isalski H (1976) Hydrocarbon Process 55(1): 103
655. Nitrogen (1976) 102: 35
656. Haslam A, Isalski WH (1975) Ammonia Plant Saf 17: 80
657. Harmon R (1981) In: More AI (ed) Fert Nitrogen: Proc Br Sulphur Corp Int Fert Technol, 4th
part I, p 113
658. Finn A (1988) Nitrogen 175: 25
659. Fabian R, Tillmann D (1986) Linde Ber Tech Wiss 59: 6
660. Harmon R, Isalski WH (1981) Ammonia Plant Saf 23: 39
661. Banks R (1978) Ammonia Plant Saf. 20: 79
662. Combs CA (1981) Ammonia Plant Saf 23: 32
663. Berry RI (1979) Chern Eng (NY) 86(15): 62
664. Overmann L, Staude E (1987) Erd61 Kohle, Erdgas, Petrochem Brenns Chern 40(10): 427
665. Sculz G, Michele H, Werner U (1982) Chern Ing Tech 54(4): 351
666. Fritzsche AK, Narayan RA (1987) CEER Chern Econ Eng Rev 19(1,2,3): 19
667. Knieriem H Jr (1980) Hydrocarbon Process 59: 65
668. Rosenzweig MD (1981) Chern Eng (NY) 88(24): 62
669. Nitrogen (1982) 136: 29
670. Chae YC, Legendre GS, van JM Gelder (1981) In: More AI (ed) Fert Nitrogen: Proc Br Sulphur
Corp Int Fert Technol, 4th part I, p 457
322 Ib Dybkjaer

611. MacLean DI, Prince CE, Chae YC (1980) Ammonia Plant Saf 22: I
672. Schendel RL, Mariz CL, Mak JY (1983) Hydrocarbon Process 62(8): 58
673. Chern Eng (London) (1979) 345: 395
674. Sheridan III JJ, Eisenberg FG, Greskovich EJ, Sandrock GD, Huston EL (1983) J Less
Common Met 89: 447
675. Santangelo JG, Chen GY (1983) CHEMTECH 13(10): 621
676. Duff BS (1955) Chern Eng Prog 51(1): 12-J
677. Mundo K-J (1973) Chern Eng Tech 45(lOa): 632
678. Strelzoff S (1974) Hydrocarbon Processes 53(10): 133
679. Chern Eng (NY) (1974) 81(23): 52
680. Buividas LJ, Finneran JA, Quartulli OJ (1974) Chern Eng Prog 70(10): 21
681. Rothman SN, Frank ME (1975) Ammonia Plant Saf 17: 19
682. Hess M (1976) Hydrocarbon Process 55(11): 97
683. Netzer D, Moe J (1977) Chern Eng (NY) 84(23): 129
684. Waitzman DA (1978) Chern Eng (NY) 85(3): 69
685. Czuppon TA, Buividas LJ (1979) Hydrocarbon Process 58(9): 197
686. Nichols D, Blouin GM (1979) CHEMTECH 9(9): 512
687. Buividas LJ (1981) Chern Eng Prog 77(5): 44; Ammonia Plant Saf (1981) 23: 67
688. Axelrod L (1981) Catal Rev Sci Eng 23(1,2): 53
689. Omori T (1982) CEER Chern Econ Eng Rev 14(9): 7
690. Staege H (1982) Tech Mitt Krupp Werksber 40(1): 1
691. Stratton A, Teper M (1984) The economics of producing ammonia and hydrogen. lEA Coal
Research, London, p 1
692. Holroyd R (1967) Chern Ind (August 5): 1310
693. Bogner H (1965) Nitrogen 34: 32
694. Axelrod L, Daze RE, Wickham HP (1968) Chern Eng Prog 64(7): 17
695. Quartulli OJ, Turner W (1972) Nitrogen 80: 28; (1973) 81: 32
696. Nitrogen (1976) 100: 77
697. Ennis R, Lesur PF (1977) Hydrocarbon Process 56(12): 121
698. Arkley KI, Pinto A (1987) In: FAI Seminar 1986. The Fertiliser Society ofIndia, New Delhi,
p SI/2
699. Pinto A, Trotter SG (1987) Proc Fert Soc London 260
700. Arkley KI, Halstead JM, Pinto A (1988) Ammonia Plant Saf 28: 58
701. Dybkjaer I (1984) In: IFA Technical Conf, Paris, 13-1
702. Denbigh KG (1956) Chern Eng Sci 6(1): 1
703. Riekert L (1974) Chern Eng Sci 29: 1613
704. Gaggioli RA, Petit PJ (1977) CHEMTECH 7(7): 496
705. Gaggioli RA (ed) (1980) Thermodynamics second law analysis ACS Symp Ser 122
706. Sussman MV (1980) Chern Eng Prog 76(1): 37
707. Christiansen LJ (1987) In: CEF 87 Proc XVIII Congo EFCE, Sicily, P 217
708. Vaclavek V, Duy Phi D, Kleinova N (1987) In: CEF 87 Proc XVIII Congr EFCE, Sicily, p 721
709. Silberring L (1971) Chern Ing Tech 43(12): 711
710. Pinto A, Rogerson PL (1977) Chern Eng Prog 73(7): 95
711. Cremer H (1980) In: Gaggioli RA (ed) Thermodynamics second law analysis ACS Symp Ser 122:
111
712. Nimmo NM (1967) Chern Eng (London) 210: CE 156
713. Tops0e H (1966) In: World Power Conf. Tokyo, paper 135 IlIA
714. Finneran JA, Mayo HC, Multharp RH, Smith RB (1969) US 3441393; CA (1969)71: No. 23310
715. Silberring L (1969) Chern Process Eng Heat Transfer Surv (August) Suppl: 74
716. Slack JB (1972) Chern Eng (NY) 79(2): 107
717. Renker W (1972) Chern Tech 24(10): 596
718. Silberring L (1979) Nitrogen 120: 35
719. van Campagne WL (1981) Hydrocarbon Process 60(8): 117; (1982) 61(7): 145; (1983) 67(7): 57
720. Tasrif A, Madsen J (1989) In: Proc Fertilizer Asia Conf and Exhibition, Manila 1989, Br
Sulphur Corp, pISS
721. Quartulli OJ (1975) Hydrocarbon Process 54(10): 94
722. Buividas LJ (1980) Proc Environ Symp, p 365
723. Bignon M (1970) In: Proc Fert Soc London 117: 98
724. Hinchley P (1979) Chern Eng (NY) 86(17): 120
725. Sault RA (1968) Chern Eng Prog 64(3): 57
 Ammonia Production Processes 323

726. Camps JA, Gentry JF (1987) Ammonia Plant Saf 27: 36

727. Badame PJ (1989) Ammonia Plant Saf29: 179
728. Barnes KB (1949) Oil Gas J (August 4): 40
729. Chern Eng (NY) (1951) 58(8): 174
730. Reidel JC (1953) Oil Gas J 51(36): 82
731. Resen FL (1954) Oil Gas J 53(32): 120
732. Reidel JC (1954) Oil Gas J 52(44): 86
733. Pet Refin (1954) 33: (11) 168/170
734. Reidel JC (1956) Oil Gas J 54(47): 106
735. Reynolds W (1965) Proc Fert Soc London 89: 1
736. Guccione E (1965) Chern Eng (NY) 72(24): 124
737. Topsoe H (1965) Chim Ind 94(1): 1
738. Mayo HC, Finneran JA (1968) Oil Gas J 66(12): 78
739. Bakemeier H, Huberich T, Krabetz R, Liebe W, Schunk M (1985) In: Ullman's Encyclopedia of
industrial Chemistry, vol A2, 5th ed. VCH Verlagsgesellschaft mbH, Weinheim, p 204
740. Brown FC (1976) Nitrogen 100: 65
741. Mundo KJ (1978) Erdal Kohle, Erdgas, Petrochem Brenns Chern 31(2): 74
742. Nitrogen (1970) 65: 32
743. Fangel PB (1970) Nitrogen 67: 35
744. Daigre III LC, Nieman GR (1974) Ammonia Plant Saf 16: 45
745. Tijssen P (1977) Ammonia Plant Saf 19: 155
746. Weems S, Ball DH, Griffin DE (1979) Chern Eng Prog 75(5): 64
747. Yost CC, Curtis CR, Ryskamp CJ (1980) Ammonia Plant Saf 22: 200
748. Collier G, Voelkers JD, Griffin DE (1980) Ammonia Plant Saf 22: 206
749. Yazhari F (1982) Hydrocarbon Process 61(5): 187
750. Huurdeman TL (1989) Ammonia Plant Sat 29: 234
751. Madhavan S (1984) Plant/Oper Prog 3(1): 14
752. Solomon SM (1985) Ammonia Plant Saf 25: 129
753. Moore SC, Piper TM, Chen CC (1986) Ammonia Plant Saf 26: 56
754. Pathe DC (1986) Oil Gas J 84(14): 49
755. Nitrogen (1987) 169: 35
756. Jarvan JE (1987) Fert Ind Annu Rev India 10: 8
757. Fert Int (1988) 267: 25
758.- Mani S, Shoor SK, Pedersen HS (1989) Ammonia Plant Saf 29: 244
759. Christensen KG, Gosnell JH, Grotz BJ (1989) In: Fertilizer Latin American Int Conf, Br
Sulphur Corp, Caracas p 99; Nitrogen (1989) 181: 31
760. Setoyama I, Wada E, Funakoshi Y (1974) Ammonia Plant Saf 21: 82
761. Grotz BJ, Gosnell JH, Grisola L (1985) Ammonia Plant Saf 25: 38
762. Grisolia L, Johnstone, WD, Grotz BJ, Crowley MR (1988) In: Nitrogen 88 Br Sulphur's 12th Int
Conf Geneva, p 11
763. Bizzotto V (1986) In: FAI Seminar 1986, The Fertiliser Society ofIndia, New Delhi p S 1/1-1
764. Pinto A (1981) US 4298588; see JP 54060298 CA (1979) 91: No. 142758
765. Pinto A, Johnson BH (1987) US 4695442; see EP 157480 CA (1985) 103: No 217689
766. LeBlanc JR (1985) Ammonia Plant Saf 25: 29
767. LeBlanc JR (1984) Hydrocarbon Process 63(7): 69
768. LeBlanc JR (1985) Energy Prog 5(1): 4
769. Becker CL (1979) US 4153673; CA (1979) 91: No. 23422
770. Rudbeck P (1982) J Tech Dev 7(4): 24
771. Bajpai T (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf Geneva, p 39
772. Hariharan TS (1987) J Tech Dev 12(2): 42
773. Sahore SR, Krishnan TS (1989) Fert News (December): 15
774. Dybkjaer I (1990) Fertilizer Industry Annu Rev XIII: 42A
775. Kalthoff P (1989) Fert News (December): 21
776. Ruziska PA, Song CC, Wilkinson RA, Unrch W (1985) Ammonia Plant Saf25: 22; Plant/Oper
Prog (1985) 2: 79
777. Ruziska PA, Dranse P, Song CC (1986) In: Nitrogen 86 Amsterdam, Br Sulphur Corp, London,
p 45
778. Bogart MJP (1978) Hydrocarbon Process 57(4): 145
779. Banquy DL (1981) US 4296085; see EP 32096 CA (1981) 95: No. 189496
780. Brook PH (1983) In: Fertilizer 83, Br Sulphur Corp 7th Int Conf London
324 Ib Dybkjaer

nil. Banquy D (1984) Ammonia Plant Saf 24: 8

782. Brown FC (1978) Paper presented at Process Dev Tech Symp 1978, The Institution of
Chemical Engineers, p 1
783. Winter CL (1984) GB 2126208; CA (1984) 101: No. 9578
784. Brown FC (1982) ECN Eur Chern News Process Rev (November 15): 8
785. Nitrogen (1983) 141: 36
786. Brown FC, Topham C (1990) In: Proceedings European Conf on Energy Efficient Production
of Fertilizers, Bristol UK, p 1
787. Orphanides P, Polychronides C (1984) In: IFA Technical Conf, Paris, p 15-1
788. Taffe P (1978) Chern Age (London) (December 15): 8
789. Ghosal SR, Karkun K (1982) Fert News (December): 47
790. Karkun K, Ghosal SR (1985) IN 156125; CA (1986) 104: No. 170928
791. Chatterjee B (1980) Fert News (December): 19
792. Thompson DC (1981) In: More AI (ed) Fert Nitrogen: Proc Br Sulphur Corp Int Conf Fert
Technol, 4th part II, p 545
793. Madhavan S, Landry B (1988) Ammonia Plant Saf 28: 1
794. Becker CL (1981) In: More AI (ed) Fert Nitrogen: Proc Br Sulphur Corp Int Conf Fert Technol
4th, part II, p 537
795. Low G (1984) Nitrogen 147 Suppl: 1
796. Nielsen A, Hansen JB, Houken J, Gam EA (1982) Plant/Oper Prog 1(3): 186
797. Graeve HW (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf Geneva, p 77
798. Wang SI, Patel NM (1984) Plant/Oper Prog 3(2): 101; Ammonia Plant Saf (1984) 24: 1
799. Dark AM, Stallworthy EA (1985) Nitrogen 153: 25
800. Singh D (1986) Process Plant Eng (Jan/March): 73
801. LeBlanc JR, Moore DO, Schneider III RV (1982) Oil Gas J 80(38): 115
802. LeBlanc JR (1986) Hydrocarbon Process 65(8): 39
803. Darajat R, LeBlanc JR (1988) In: Nitrogen 88 Br Sulphur's 12th Int Conf, Geneva, p 29
804. MacLean DL, Prince CE, Chae YC (1980) Ammonia Plant Saf 22: 1
805. Howerton RG (1980) Ammonia Plant Saf 22: 9
806. Patel NM, Wang SI, Kittelstad KJ (1989) Ammonia Plant Saf 29: 33
807. Reidel JC (1954) Oil Gas J 53(13): 78
808. Skinner GF, Kowal WM (1983) US 4409196; CA (1984) 100: No. 9464
809. Skinner GF (1981) In: More AI led) Fert nitrogen: Proc Br Sulphur Corp Int Conf Fert
Technol, 4th, part II, p 491
810. Gent CW, Ward SA (1980) Chern Eng (London) 353: 85
811. Partridge LJ (1980) Chern Eng (London) 353: 88
812. Tippmer K (1986) Erdal Kohle, Erdgas, Petrochem Brenns Chern 39(8): 361
813. Brown F (1977) Hydrocarbon Process 56(11): 361
814. Nurse TW (1982) In: Proc 5th lnt Symp Large Chern Plants, Antwerpen, p 131
815. Nitrogen (1986) 160: 39
816. Hargreaves HG, Kirk RJ (1980) Inst Chern Eng Symp Ser 62: sl
817. Stewart RM, Marqueen TJ (1978) Coal Process Technol4: 201
818. CEER Chern Econ Eng Rev (1980) 12(6-7): 33
819. Dyekjaer Hansen E (1980) Chern Age India 31(12) : DEV 3/1
820. Keens D (1982) lnst Chern Eng Symp Ser 74: 55
821. Nitrogen (1966) 43: 26
822. Oil Gas J (1966) 64(33): 48
823. Nitrogen (1970) 64: 17
824. Hsu H, Jaeschke WC (1967) US 3349126; CA (1968) 68: No. 12509
825. Quartulli OJ, Axelrod LC, Randall R (1971) Het Ingenieursblad 40(21): 642
826. Lagana V, Zardi U (1977) Proc Fert Soc London 167: 1
827. Lagana V (1978) Chern Eng (NY) 85(1): 37
828. Farinola G, Lagana V (1979) Hydrocarbon Process 58(9): 202
829. Van Dijk CP (1989) US 4869887; CA (1990) 112: No. 39261
830. Hale CC, Lichtenberg WH (1988) Ammonia Plant saf 28: 76
831. Hale CC (1988) In: Nitrogen 88 Br Sulphur's 12th lnt Conf, Geneva, p 161
832. Reed JD (1979) In: Slack AV, James GR (eds) Ammonia, vol 4. Marcel Dekker, New York, p 1
833. Bakemeier H, Huberich T, Krabetz R, Liebe W, Schunck M (1985) In: Ullman's Encyclopedia
of industrial chemistry, vol A2, 5th ed. VCH Verlagsgesellschaft mbH, Weinheim, p 221
834. Hale CC (1974) Ammonia Plant Saf 16: 23
 Ammonia Production Processes 325

835. Hale CC (1979) Ammonia Plant Saf 21: 61

836. Hale CC (1979) Nitrogen 119: 30
837. Hale CC (1984) Plant/Oper Prog 3(3): 147
838. Hale CC (1984) Ammonia Plant Saf 24: 181
839. Hale CC (1984) Nitrogen 150: 27
840. Swaminathan BV, Jain BK (1986) Fert News (May): 35
841. Safety recommendations for the large scale storage of anhydrous ammonia (1985) IFA,
International Fertilizer Industry Association Ltd, Paris
842. Hale CC, Lichtenberg WH (1980) Ammonia Plant Saf 22: 35
843. Parmar AS (1986) Fert News (May): 65
844. Huberich T (1982) Plant/Oper Prog 1(2): 117
845. Suri IK, Bhola RK (1986) Fert News (May): 52
846. Phelps EH (1972) Ammonia Plant Saf 14: 109
847. Arup H (1977) Ammonia Plant Saf 19: 73
848. Lunde L (1984) Ammonia Plant Saf 24: 154
849. Lunde L, Nyborg R (1987) Ammonia Plant Saf 27: 49; Plant/Oper Prog (1987) 6(1): 11
850. Stephens JD, Vidalin F (1988) Ammonia Plant Saf 28: 9
851. Olsen EA (1969) Ammonia Plant Saf 11: 46
852. Byrne JR, Moir FE, Williams RD (1989) Ammonia Plant Saf 29: 122
853. Nielsen A (1971) Ammonia Plant Saf 13: 103
854. Blanken JM (1984) Ammonia Plant Saf 24: 140
855. KA van Grieken (1976) Ammonia Plant Saf 18: 15
856. Cracknell A (1980) Ammonia Plant Saf 22: 63
857. Brown RS (1982) Plant/Oper Prog 1(2): 97
858. Guth DC, Clark DA (1985) Ammonia Plant Saf 25: 60
859. Guth DC, Clark DA (1985) Plant/Oper Prog 4(1): 16
860. Comeau ET (1972) Ammonia Plant Saf 14: 69
861. Hendriks NA (1979) Ammonia Plant Saf 21: 69
862. Aarts 11, Morrison DM (1981) Ammonia Plant Saf 23: 124
863. Al-Abdulally F, Al-Shuwaib S, Gupta BL (1987) Ammonia Plant Saf 27: 153
864. Comeau ET, Weber ML (1977) Ammonia Plant Saf 19: 63
865, Vick KA, Withaus JB, Mayo HC (1980) Ammonia Plant Saf 22: 54
866. Martinez OA, Madhavan S, Kellett DJ (1987) Ammonia Plant Saf 27: 162
861: Code of Practice for Large Scale Storage of Fully Refrigerated Anhydrous Ammonia in the UK
(1975). Chemical Industries Association, London
868. Shah JM (1982) Plant/Oper Prog 1(2): 90
869. Thompson JR, Carnegie RN (1989) Ammonia Plant Saf29: 116; Plant/Oper Prog (1989) 8(2): 92
870. Hale CC, Josefson AJ, Mattick DE (1985) Ammonia Plant Saf 25: 172
871. Hale CC (1982) Plant/Oper Prog 1(2): 107
872. Arunachalam PN (1986) Fert News (May): 57
873. Dayasagan I, (1986) Fert News (May): 60
874. Briggs PP, Richards III JM, Fiesinger EG (1986) Ammonia Plant Saf 26: 89
875. Prasek F (1988) Ammonia Plant Saf 28: 66
876. Development and Transfer of Technology Series, No. 13 (1980). Fertilizer Manual. United
Nations, New York, p 49
877. Dybkjrer I (1991) In: Nitrogen 91 Preprints, Br Sulphur Corp, London, p 59; Indian Fertilizer
Scene Annual p 1
878. Wilhelm SM (1991) Ammonia Plant Saf 31: 20
879. Orieux A, Barthel Y (1992) Hydrocarbon Techn Int 3: 113
880. Nitrogen (1992) 195: 22
881. Storgaard L (1991) In: Nitrogen 91 Preprints, Br Sulphur Corp, London, p 97
882. Cromarty BJ (1991) In: Nitrogen 91 Preprints, Br Sulphur Corp, London, p 115
883. Hansen J-H, Bak Storgaard L, Pedersen PS (1992) Ammonia Plant Saf 32: 52
884. Wanger ES, Froment GF (1992) Hydrocarbon Process (July): 69
885. Salot WJ (1991) Ammonia Plant Saf 31: 121
886. Cromarty BJ (1992) Ammonia Plant Saf 32: 197
887. Cromarty BJ, Crewdson BJ (1991) Nitrogen 191: 30
888. Vannby R, Madsen SEL Winter (1992) Ammonia Plant Saf 32: 122
889. Elkins KJ, Jeffery IC, Kitchen D, Pinto A (1991) In: Nitrogen 91 Preprints, Br Sulphur Corp,
London, p 83
326 Ib Dybkjaer

890. Schneider RV III: (1991) FAI Seminar 1990, The Fertiliser Association ofIndia, New Delhi, p
SII-l
891. Shiris PJ, Cassata JR, Mandelik BG, van Dijk CP (1984) US 4479925
892. Degand P, Schurmans JP, Julemont V (1992) Ammonia Plant Saf 32: 129
893. Quintana ME, Skinner GF (1991) FAI Seminar 1990. The Fertiliser Association ofIndia, New
Delhi, p SII-2
894. Kumar R, Srinivasan V (1991) Indian Fertilizer Annual Scene p 17
895. Hauser N, Bayens CA (1992) Ammonia Plant Saf 32: 268
896. Nitrogen (1990) 186: 21
897. Roos H, Wanjek H, Sprague M (1990) Ammonia Plant Saf 30: 187
898. Kitchen D, Henson WGS, Madsen JK (1990) Ammonia Plant Saf 30: 105
899. Carstensen JH, B0gild-Hansen J, Pedersen PS (1991) Ammonia Plant Saf 31: 113
900. Bartoo RK, Gemborys TM, Wolf CW (1991) In: Nitrogen 91 Preprints. Br Sulphur Corp
London, p 127
901. Nitrogen (1992) 198: 26
902. Song CC, Wohlgeschalfen KR, Verduijn WD, Qadir A (1992) Ammonia Plant Saf 32: 187
903. Shiyong Z (1991) In: Nitrogen 91 Preprints, Br Sulphur Corp London, p 141
904. Shah VA, Huurdeman TL (1990) Ammmonia Plant Saf 30: 216
905. Hefner W, Herion C, Meissner H, Appl M (1991) FAI Seminar, 1990. The Fertiliser Association
of India, New Delhi, p SII-3
906. Meissner H, Wammes W, Hefner W (1991) In: Nitrogen 91 Preprints. Br Sulphur Corp
London, p 151
907. Nitrogen (1992) 197: 18
908. S0gaard-Andersen P, Hansen 0 (1992) Ammonia Plant Saf 32: 177
909. Rail W (1991) Ammonia Plant Saf 31: 247
910. Rasmussen CV (1991) Nitrogen 194: 30
911. Peterson RB, Finello R, Denavit GA (1984) US 4452760; CA (1985) 102: No. 187456t
912. Grotz BJ (1986) WO Patent Application 86/06058; CA (1987) 106: No. 69646q
913. Clayton KA, Shannahan N, Wallace B (1991) Ammonia Plant Saf 31: 212
914. Campbell JD, Rawlinson RL, WIlson KC (1992) Ammonia Plant Saf 32: 1
915. Huurdeman TL, Schrijen HC (1992) Ammonia Plant Saf 32: 10
916. Prescott GR (1992) Ammonia Plant Saf 32: 217
917. Heuser A (1992) Ammonia Plant Saf 32: 243
918. Wagner GH, Heuser A, Heinke G (1992) Ammonia Plant Saf 32: 252
919. Timbres DH, van Moorsel WH, Johnson DL (1990) Ammonia Plant Saf 30: 200
920. Madsen J (1991) Ammonia Plant Saf 31: 227
921. Patel NM, Kittelstad KJ (1990) Ammonia Plant Saf 30: 208
922. Allen RL Jr, Moser GA (1992) Ammonia Plant Saf 32: 170
923. Grossmann G, Dejaeger J (1992) Ammonia Plant Saf 32: 164
924. Kitchen D, Pinto A (1991) FAI Seminar 1990. The Fertiliser Association ofIndia, New Delhi, p
SIII-l
925. Hicks TC, Pinto A, Moss MJS (1990) Ammonia Plant Saf 31: 219
926. Kitchen D, Pinto A (1991) Ammonia Plant Saf 31: 152
927. Armitage PM, Elkins KJ, Kitchen D, Pinto A (1992) Ammonia Plant Saf 32: 111
928. LeBlanc JR, Shires PJ (1992) Hydrocarbon Technology Int 3: 141
929. Dybkjrer I (1991) FAI Seminar 1990. The fertiliser Assoication ofIndia, New Delhi, p S III-2
930. Hakmann R (1991) FAI Seminar 1990. The Fertiliser Assoication ofIndia, New Delhi, p SIll 3
931. Jungr M (1991) Nitrogen 191: 42
932. Sharma SP, Bhaduri AK (1992) FAI Seminar 1991. The Fertiliser Assoication of India, New
Delhi, p SII/l-l
933. Tsujimoto M, Bendix H, Lenz L, Johannes D (1990) Ammonia Plant Saf 30: 167
934. Tusjimoto M, Johannes D (1991) In: Nitrogen 91 Preprints. Br Sulphur Corp London, p 69
935. Brown F (1992) Finds, First Quarter, p 32
936. Hale CC, Lichtenberg WH (1990) Ammonia Plant Saf 30: 225
937. Lunde L, Nyborg R (1990) Ammonia Plant Saf 30: 60
938. Lunde L, Nyborg R (1991) Proc No. 307. The Fertiliser Society, London, p 1
939. Hewerdine S (1991) Proc No. 308. The Fertiliser Society, London, p 1
940. Appl M, Fassler K, Fromm D, Gebhard H, Porte H (1990) Ammonia Plant Saf. 30: 22
941. Selva RA, Heuser AH, (1990) Ammonia Plant Saf 30: 39
942. Burke BG, Moore DE (1990) Ammonia Plant Saf 30: 91
 Ammonia Production Prucesses 327

943. Conley MJ, Angelsen S, WIlliams D (1991) Ammonia Plant Saf. 31: 159
944. Thompson JH (1990) Ammonia Plant Saf 30: 241
945. Tilton IN, Squire RH, Saffle CS, Atkins CR (1992) Ammonia Plant Saf 32: 63
946. Wiltzen RCA (1991) Ammonia Plant Saf 31: 28
947. Squire RH (1991) Ammonia Plant Saf 31: 131
948. Ali SB, Smallwood RE (1991) Ammonia Plant Saf 31: 142
949. Appl M (1992) Nitrogen 199: 46; (1992) 200: 27; (1992) 202: 44
950. Tomasi L (1992) Nitrogen 199: 35
951. Elkins KJ, Gow AJ, Kitchen D, Pinto A (1992) Proc No. 319. The Fertiliser Society, London, p
1
952. Grotz BJ, Grisolia L (1992) Nitrogen 199: 39
953. Nitrogen (1989) 180: 20
954. Nitrogen (189) 182: 25
955. Karthikeyan S, Neelakantan PS (1991) IFA-FADINAP Regional Fertilizer Conf for Asia and
the Pacific. New Delhi, December 1991, p 3
956. Simanjuntak R, Eko NB (1992) Asiafab 1: 18
957. Asiafab (1992) 1: 15
958. Grotz BJ, Stupin WJ (1990) Fertilizer Latin American International Conf (British Sulphur).
Caracas, Venezuela (October 1990) p 1
959. Dybkjrer I. (1990) IFA Technical Conference, October 1-5, 1990. Venice, Italy, p 1
960. LeBlanc JR (1986) CEER Chern Econ Eng Rev 18(5): 22
961. Rostrup-Nielsen JR, Nielsen PEH, S0rensen NK, Carstensen JH (1993) Ammonia Plant Saf. 33:
184
962. Mohri T, Takemura K, Shibasaki T (1993) Ammonia Plant Saf 33: 86
963. Verduijn WD (1993) Ammonia Plant Saf 33: 165
964. March HD, Thiagarajan N (1993) Ammonia Plant Saf 33: 108
965. Deshmukh D, Raagaard S, Chawes L, Olesen L (1993) Ammonia Plant Saf 33: 278
966. Dybkjrer I (1992) IFA Technical Conference, October 5-8, 1992; The Hague, The Netherlands,
pi
967. Dybkjrer I (1992) IFA-FADINAP Regional Conference for Asia and the Pacific, Bali,
Indonesia, November 30-December 2, 1992, p 1
968. Lee JM, Cialkowski EJ (1993) Ammonia Plant Saf 33: 53
Chapter 7
Ammonia Storage and Transportation-Safety

Anders Nielsen
Haldor Tops0e A/S Copenhagen, Denmark

Contents

7.1 General Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330

7.2 Toxicity and Contact Injuries . . . . . . . . . . . . . . . . . . . . . . . . 330

7.3 Explosion and Fire Hazards . . . . . . . . . . . . . . . . . . . . . . . . . 332

7.4 Safety in Plant Storage and User Storage . . . . . . . . . . . . . . . . . 332

7.5 Ammonia Transportation and Related Accidents. . . . . . . . . . . . . 334

_7.5.1 Pipelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
7.5.2 Tankers and Barges . . . . . . . . . . . . . . . . . . . . . . . . . . 335
7.5.3 By Rail. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
7.5.4 By Road . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336

7.6 Accidents with Fixed Tanks . . . . . . . . . . . . . . . . . . . . . . . . . 336

7.7 Behavior of Ammonia Spills . . . . . . . . . . . . . . . . . . . . . . . . . . 338

7.8 Ammonia Spill Studies . . . . . . . . . ; . . . . . . . . . . . . . . . . . . 340

7.9 Ammonia Storage Facilities . . . . . . . . . . . . . . . . . . . . . . . . . 342

7.10 Ammonia Transportation Systems. . . . . . . . . . . . . . . . . . . . . 343

7.11 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344

Nielsen (Ed.), Ammonia

© Springer-Verlag Berlin Heidelberg 1995
330 A. Nielsen

7.1 General Aspects

This section deals with certain physical and chemical data, which have particu-
lar reference to the safety in storage, transportation and use of ammonia.
At atmospheric temperature and pressure, ammonia is a colorless gas with a
sharp and pungent odor. Ammonia also exists as a colorless liquid. Its vapor
pressure at - 33.6 DC is 1 atm, at + 4.7 DC it is 5 atm, at 20 DC it is 8 atm, and at
25.7 DC it is 10 atm. At 50.1 DC, the vapor pressure is 20 atm. Ammonia vapor at
the boiling point of - 33 DC will have a vapor density of approximately 70% of
the density of ambient air. However, ammonia and air can, under certain
conditions, form mixtures that are denser than ambient air [1], as a result ofthe
mixture being at a lower temperature caused by the evaporation of cold
ammonia. An ammonia cloud formed by an accidental release of ammonia may
contain a mist ofliquid ammonia and its density may be greater than that of air.
For the density of an ammonia cloud containing liquid ammonia as it is diluted
with dry air, see [2].
Several literature sources list different values for the flammability limits of
ammonia mixtures with air, but 16-27% ammonia by volume appears typical.
The flammability limits of ammonia in oxygen are 15 to 79% [3]. Limits of
flammability for various oxygen/nitrogen ammonia mixtures of various temper-
atures as well as the flammability characteristics of ammonia/water vapor/air
mixtures are given in [4]. Flammability limit measurements were carried out at
three temperatures with a supporting atmosphere containing from 21 % oxygen
(air) down to 11 % oxygen. All experiments were carried out at atmospheric
pressure. The data show that the flammability limits widen as temperature
increases. At 400 DC the flammability limits have widened to 11 to 37% in air
[4].
The high solubility ofNH 3 in H 2 0 and the high heat of mixing must be fully
considered in the estimation of its behavior in accidents.

7.2 Toxicity and Contact Injuries

The threshold of perception of ammonia varies with the individual and may also
depend on atmospheric conditions. A lower limit of perception in the range of
0.4 to 2 wt ppm is given in [5]. In this report it is also concluded that levels of
ammonia at the perception threshold may cause changes in the biopotentials of
the brain.
Plant surveys [6] report concentrations of 9 to 45 ppm for various areas.
Such ammonia concentrations initially are irritating to the eyes or nose, but the
 Ammonia Storage and Transportation-Safety 331

exposed persons rather quickly became accustomed to the ammonia. Another

report [7] lists recommendations as to the limits of ammonia concentrations in
air to which the public may be exposed for short periods of time as follows:
10 min 100 ppm
30 min 75 ppm
60 min 50ppm
The threshold limit value of the American Conference of Governmental In-
dustrial Hygienists is 25 ppm (or 18 mg/m 3 ) [8].
In the USA, the above Time Weighted Average Threshold Value has been
supplemented with a Threshold Limit Value-Short Term Exposure of35 ppm in
air (or 27 mg/m 3 ). The Short Term Exposure refers to an average of 15 minutes.
For symptomatology at various exposure limits, see [8, 9]. Medical manage-
ment of anhydrous ammonia emergencies is thoroughly discussed in [9],
distinguishing between the patient mildly affected, those moderately affected,
and those with severe ocular lesions, skin lesions, and gastro-intestinallesions.
The prehospital phase, hospital care, long-term complications and accident
follow-up are discussed. Based on [8-10], the effects of high concentrations of
ammonia on health may roughly be indicated as follows. There are obviously
individual differences:
400 ppm: Throat irritation.
700 ppm: Eye injury.
1700 ppm: Coughing appears and labored breathing may be present
for hours. Some patients may suffer momentary inability to
breath.
2500 to 4500 ppm: May be fatal after short exposure.
5000 ppm and higher: Death results from respiratory arrest.
According to [9], the most frequent sources of exposure are: (1) the spraying of
liquid ammonia, (2) being caught in a cloud of ammonia vapor, and (3) by
ingestion. Six less common means of exposure are also listed in [9].
Both ammonia spray and liquid are freezing cold and cause frostbite.
Temperatures ofliquid ammonia down to - 50°C may be reached as a result of
rapid evaporation of ammonia released at - 33 °C [11]. Lessenger [9] lists
decontamination as the first treatment of victims, pointing out, however, that
clothing may be frozen to the skin, in which case much water must be
administered before clothing can be removed. In the prehospital phase, eye
injuries are flushed with water for at least 15 minutes. Eyelids must be held open
while a constant stream of water flushes out the ammonia [9].
The metabolism and pharmacokinetics of ammonia as well as the health
effects of chronic, subchronic and acute exposure to ammonia is discussed in
[81]. It is concluded that due to its chemical properties and interactions within
the respiratory tract ammonia does not have the potential to cause chronic
damage from long term, low level exposure.
332 A. Nielsen

7.3 Explosion and Fire Hazards

Accidents in which fire and/or explosion of ammonia-air mixtures have been

reported to cause the death or injury of personnel are known [77, 78], but are
not very frequent. Such accidents have occurred in cold storage warehouses in
which ammonia was released in an enclosed space and ignition was caused
accidentally. One reason that explosions and fire accidents have not been
frequent is that upon release of ammonia from atmospheric or pressurized
tanks, rapid evaporation of ammonia cools down the liquid and thereby reduces
the vapor pressure of ammonia over the liquid. As noted above, even moment-
ary fatal concentrations of ammonia are much lower than the lower explosion
limit in air. Also the fairly high ignition temperature and the low normal flame
velocity of an ammonia/air mixture (0.07 to 0.08 m/s) are important [12]. For
protection against fire and explosion in ammonia storage or transportation
vessels see [13].
Another explosion hazard associated with ammonia is related to the reac-
tion of ammonia with mercury in the presence of trace amounts of water leading
to the formation of an explosive compound [14]. This kind of explosion has
been encountered when mercury manometers, after service for several years in
an atmosphere of ammonia, explode at the moment of cleaning [15]."ln this
connection it should be mentioned that natural gas from a number of sources
contain elemental mercury in significant quantity [82].

7.4 Safety in Plant Storage and User Storage

Storage facilities for ammonia are discussed in this volume in the chapter by
Dybkjaer. Here we will discuss storage facilities with reference to associated
safety aspects. Three principally different types of ammonia storage tanks are
used in producing plants and distribution terminals. Pressurized (unrefrigerated)
storage tanks operate at ambient temperature and are typically designed for a
pressure of 18.25 bar. Storage spheres are refrigerated typically by a single-stage
refrigeration compressor, operated at temperatures from - 1 to + 2°C, and
are designed for pressure of 3.8 to 5.15 bar. The fully-refrigerated storage tank is
serviced by two-stage refrigeration, operated at a temperature of - 33°C, and is
typically designed for a pressure of 1.117 bar [16]. It is important to also design
the tank for a certain underpressure. A capacity of pressurized storage tanks up
to 270 t, refrigerated spheres from 450 to 2750 t, and fully-refrigerated storage
tanks up to 45000 t is reported in [16]. Agricultural-use tanks range from less
than one ton to a few tons in size, the larger tanks being used as applicator tanks
as well as nurse tanks for smaller tanks.
 Ammonia Storage and Transportation-Safety 333

Liquid ammonia has a high thermal coefficient of expansion, and rules and
regulations exist in various countries governing the maximum filling of station-
ary and transportation tanks. A guide for developing a training program for
anhydrous ammonia workers has been issued by the U.S. National Institute for
Occupational Safety and Health, Cincinnati [18].
A further reference to actual design, practice and experience of refrigerated
ammonia storage in North America is reported in [16], while an ammonia
storage terminal safety program discusses, among other subjects, safety training
and accident prevention [13].
While the many thousands of small tanks used in distribution and local
storage present their own safety problems, of which examples will be listed
below, very large storage facilities represent a potential for very large ammonia
spills. One effort to reduce the potential hazard is to use a double-walled tank or
to protect the tank by dikes, retention pits or concrete walls. Dikes, and
particularly concrete walls, will limit the surface of the pool of liquid ammonia
which could be formed in the case of the failure of a large storage tank. In the
case of a tank surrounded by a dike, an initial high rate of vaporization would be
controlled primarily by heat transfer from the ground and therefore the initial
flash depends upon the condition of the ground within the dike [19].
The initial vaporization can create toxic and fire hazards a considerable
distance downwind from the tank for a period of time, which is determined
largely by atmospheric conditions. Subsequent steady state vaporization pre-
sents a smaller fire hazard and a more moderate toxicity hazard, and is
controlled by the surface area of the spill and atmospheric conditions.
Stress corrosion cracking has been a serious problem in Europe and also in
the U.S. Early observations came from Denmark [20]. Quenched and tempered
steels appear particularly vulnerable. Inhibition with a minimum of 0.2 % water
is used to reduce the risk of cracking. Oxygen (air) plays an important role in the
corrosion. Since stress corrosion cracking is the result of chemical processes, it is
much less likely and would occur much slower in atmospheric storage tanks
containing liquid ammonia at - 33°C [20]. According to T. Hallan [94] crack
initiation is very infrequent at - 33°C and crack growth rate is approximately
three times slower at - 33 °C than at 18°C. Cf. also Nyborg, Lunde and Conley
[95] concerning ammonia storage vessel life prediction and Nyborg and Lunde
concerning measures for reducing stress, corrosion cracking [96].
For a long time it was believed in the industry that fully-refrigerated
ammonia storage tanks would not suffer stress corrosion cracking. In 1987 [80]
in the U.K., a fully-refrigerated ammonia storage tank was decommissioned for
inspection after almost ten years of service. In the inspection, stress corrosion
cracking was found. The water content of ammonia storaged in this tank had
been typically 0.02 wt %, which had been considered sufficient to inhibit stress
corrosion cracking. The oxygen content of the liquid or vapor had not been
determined, and there was a possibility, particularly during the commissioning
ofthe tank, that some oxygen could have been present and played a role in crack
initiation. The structural integrity of the tank was analyzed and verified by
334 A. Nielsen

fracture mechanics [83]. Certain precautions were taken during the recommis-
sioning of the tank.
A new disclosure of stress corrosion cracking in a fully-refrigerated ammonia
storage tank was reported a year later [89]. Two identical tanks were inspected
of which one showed a significantly higher level of attack by SCC than the other
which could not be correlated with differences in the oxygen or water content of
the ammonia.
A detailed study of stress corrosion cracking has been carried out at
"Institutt for Energiteknikk" Kjeller, Norway financed by ammonia producers
and safety authorities. The experimental results and conclusions as to the
influence of tank environment, electrochemical potential, temperature, steel
properties and types of welding electrodes are reported and recommendations
given [92]. Application of cathodic protection in a large atmospheric ammonia
storage tank is described in [97].
A description of leakage in the floor of an ammonia storage tank (a
particularly critical point of damage) and the repair operations are given in [21].

7.5 Ammonia Transportation and Related Accidents

Liquid ammonia is transported in tankers, barges, by railcars and by various

means of road transportation. A newer way is the use of pipelines for ammonia
transportation.

7.5.1 Pipelines

There are three major ammonia pipelines; two in the U.S. and one in the USSR.
One of the American pipelines (MidAmerica Pipeline System) transports am-
monia from the Texas Panhandle to points in Kansas, Nebraska and Iowa. Its
total peak capacity delivered to a number of points is 8000 t/day [22]. The other
U.S. pipeline, the Gulf Central Pipeline, transports anhydrous ammonia from
major producers along the Texas and Louisiana Gulf Coast to points in Iowa,
Illinois, Nebraska, Indiana and Missouri [23]. Further information on the
maintenance and repair of the MidAmerica Pipeline System is given in [24]. The
total weight of ammonia in the MidAmerica Pipeline System, when full, is
approximately 20,000 tons. Lock valves are 10 miles apart so it would be
possible that 400 tons could get out between lock valves [22]. It is known from
[12] that in one accident 700 tons ofliquid ammonia leaked from the pipeline in
the USSR. In the USSR pipeline accident, the aerosol cloud of ammonia covered
a forested territory about 40 km 2 in area; however, all residents in the region
were evacuated in time. Pipeline operations in the U.S. are governed by the
national Gas Pipeline Safety Act of 1968, as amended in 1979 [17].
 Ammonia Storage and Transportation-Safety 335

Beside the main distribution lines, a number of smaller pipelines between

ammonia plants and adjacent chemical plants and loading facilities exist. A
rupture of such a small pipeline resulting in the release of 50 t of ammonia has
been described [46].

7.5.2 Tankers and Barges

Large quantities of ammonia are transported by tankers and barges. In 1991 a

total of about 10 million tons of anhydrous ammonia was transported by ocean
going vessels (IF A Ammonia Statistics, yearly issue). According to a Battelle
report [17] U.S. Coast Guard regulations limit the capacity of tankers to 25000
tons and of barges to 2500 tons. The Safety Features and Loading Rates of
Refrigerated Barges is described in [25]. There have been a number of accidents
during loading and unloading of vessels. The ammonia loading hoses or lines
appear a central part of the facilities. A tanker's loading line ruptured and
180 tons of ammonia leaked out and a cloud of ammonia gas moved over the
vessel. After the ammonia cloud had dispersed, 2 men of the 7-man crew were
found dead and the other 5 were rescued [26]. Another accident, occurred
during loading of a large vessel [27]. No fatalities occurred in this accident.
However, a dockman received first and second degree burns.

7.5.3 By Rail

Ammonia is extensively transported by rail car. The type oftank cars used in the
U.S. for transportation of anhydrous ammonia as well as their safety aspects,
inspections and safety record are discussed in [28]. Here also two accidents
involving ammonia tank cars are dealt with. The detailed description of the
accident which occurred in 1976 at Glen Ellyn, Illinois, is described in [29,49].
This accident was due to a derailment as two trains passed each other. The tank
head of an ammonia car was punctured by a coupler of an adjacent car. One of
the conclusions of the analysis is that if the tank car had been provided with
head shields, the tank head would not have been punctured during the derail-
ment. Another accident in which a railroad tank car carrying anhydrous
ammonia suffered a puncture is described in [30]. This tank car contained about
75 t of anhydrous ammonia at a pressure of 4 to 5 atm. The handling of the
railroad derailments involving anhydrous ammonia and other hazardous ma-
terials is discussed in [31]. Computer-aided ammonia rail car loading is
described in [32].
A first-hand account of the efforts of an emergency response team at the
Pensacola ammonia accident, which was due to a train derailment, is given in
[48]. A special feature in some ammonia tank car accidents is the appearance of
336 A. Nielsen

delayed tank car failures. This is reviewed in [50]. In this article, delayed tank
car ruptures which occurred in accidents at Cummings, Iowa and Crestview,
Fla. are discussed. A detailed survey of damages from ammonia spills has been
given by Markham [79].
For transportation accidents by railroad, see also [47]. This report gives a
survey of damage during 177 anhydrous ammonia highway incidents and 570
rail incidents in the U.S. for the years 1971 to 1982. It lists the identification
number for anhydrous ammonia as UNlO05 and the required labeling as a
poisonous gas.

7.5.4 By Road

Various road vehicles are used to transport ammonia. The U.S. Department of
Transportation has limited the use of certain high stress steels only for ammonia
which has a minimum content of water of 0.2 wt %, or a purity at least 99.995%
[33]. A more recent reference [17] mentions that when ammonia is shipped in
containers constructed of quenched and tempered steel it must contain a
minimum of 0.2 wt % water. Various countries specify maximum percentage
filling of tanks to provide a cushion. The European Council of Chemical
Manufacturer's Federation (CEFIC) Zurich has issued a safety card for road
transportation of anhydrous ammonia, containing certain information and
recommendations which have been used in some European countries in connec-
tion with local recommendations and lists of laws and rules relevant to
transportation and storage of anhydrous ammonia [34].
Two road vehicle accidents are particularly well known. In one, a semi-
trailer truck ruptured suddenly in the yard of a factory at Lievin (France). The
accident and the analysis of the cause of the accident is reported in [35]. The
almost instantaneous escape of 19 tons of ammonia caused severe injuries to 20
persons, 5 of whom died. Another serious accident occurred in Houston, Texas,
when a tractor semi-trailer tank transporting approximately 27 m 3 of anhyd-
rous ammonia struck and then penetrated a bridge rail on a ramp connecting
two highways. The tractor and trailer left the ramp and fell on to a highway. The
anhydrous ammonia was released and 78 persons were hospitalized. Six persons
died as a result of the accident. Since the accident happened on a highway, most
of the victims were motorists [36, 37].

7.6 Accidents with Fixed Tanks

The quantities of ammonia that may be released in the manufacturing facilities

are typically smaller than quantities that may be released from storage or
 Ammonia Storage and Transportation-Safety 337

transportation tanks. However, under unfavorable conditions even a small

quantity may present a danger to life.
One fatal accident is described in [17]. A cylinder used in servicing air
conditioning equipment and containing only 5.7 kg of ammonia was being
transported in the cargo space of a van-truck. The cylinder ruptured as the truck
was moving at approximately 60 mph. The driver succeeded in stopping the
truck and opening the door, after which he fell out and died, either at the scene
or on the way to the hospital.
For pressurized storage tanks one concern is stress corrosion cracking.
Stress corrosion cracking in pressurized storage tanks was first observed in
Denmark in 1964 [38] when. a new 230 m 3 cylindrical tank started to leak
through a long crack. A detailed review of stress corrosion of steels in anhydrous
ammonia service and laboratory investigations into its cause and prevention is
presented in [39, 40, 92].
Other accidents with pressurized storage vessels have occurred as a result
of brittle fractures. In one accident in South Africa, a 50 t pressurized storage
tank failed, releasing 38 t of anhydrous ammonia and killing 18 people.
Analysis of the tank after the accident showed that the failure was a brittle
fracture and it was felt that reasidual stresses from an earlier weld repair of some
minor cracks may have contributed to the cause of the failure. After the welding,
the tank had not been subsequently stress-relieved. A hydraulic pressure testing
of the vessel would have contributed to the strain-aging process of the metal
[41].
The large atmospheric storage tanks represent a significant potential hazard
due to the large inventories. Stress corrosion cracking occurs much slower in
these tanks because of the low temperature, - 33°C, and pressure, and because
they are newer and more information on the selection of steels for tanks and
other related factors were available when they were built. The kind of incidents
that have been reported are associated with settling of the ground below the
tanks, loss of heating below, and improper welding techniques in the floor plate
or the cylindrical part of the tank [42, 43, 44]. Cf also earlier discussion of stress
corrosion found during tank inspections.
A serious accident occurred in a Lithuanian storage facility in 1989. The tank
in question was a single-wall tank protected by a concrete wall. The cause of the
failure, which was very dramatic with the tank opening up on one side at the
bottom seam and part of the tank being dislodged from its foundation, smashing
through the concrete wall and being displaced 40 m, is not entirely clear [86].
There is a possibility that the cause of the accident may have been the so-called
"roll-over". This may be caused by a quantity of warm ammonia being fed into
the bottom of a storage tank, partly filled with cold ammonia. The phenomena
associated with the feeding of warm ammonia into a fully-refrigerated ammonia
storage tank are discussed in [87].
An instance of an ammonia spill has been reported when a refrigerated
ammonia storage tank was filled from a barge and the high level alarm and
shutdown system failed to operate [17]. Also, instances of creating a vacuum in
338 A. Nielsen

the tank, due to instrument failure as a result of freezing up, have been reported
[45].
A general survey of refrigerated ammonia storage in North America, treating
design practices and experience, tank locations relative to neighborhoods, and
various technical questions, is given in [16].

7.7 Behavior of Ammonia Spills

Upon release into the atmosphere of an ammonia spill, a portion of it will flash
off as a vapor. This portion is the initial flash followed by a period of
evaporation until all of the ammonia has evaporated, been dissolved in water or
otherwise disposed of. Some of the fundamentals of the behavior of ammonia
released are discussed in [1,51].
There is a significant difference in the release of ammonia from a refrigerated
tank operating slightly above atmospheric pressure, where the amount of the
initial flash is only a few tenths of a percent of the total, and the release from a
pressurized storage system under a pressure of about 9 atm and 24 cC, in which
case approximately 20% of the spilled ammonia would be flashed. Figures are
given showing the amount of flash depending upon the initial pressure and
temperature. Figures are also shown about the behavior of small and large pools
regarding temperature development and evaporation rates at certain wind and
temperature conditions. The importance of meteorological conditions is pointed
out. There is a difference between the situation at night with an overcast
condition, and the situation on a hot sunny day, where as much as 80% of the
radiation may be absorbed and re-radiation is negligible due to the low
temperature of the ammonia pool. In addition, the question of vapor dispersion
in dealt with [51].
In [1], the various ways of releasing ammonia are discussed, and spills on
land and on water are discussed in a qualitative manner. The· question of liquid
ammonia jets, their direction, the fraction of ammonia remaining airborne and
falling to the ground, the formation of dense air ammonia mixtures, and the
formation of fogs are discussed in [1].
A simple model has been established [52] for the release of anhydrous
ammonia from pressurized containers. The flash off and the entrainment of air in
ammonia vapor is discussed. It is pointed out that the density of the mixture
depends on the airborne fraction of the liquid, the ambient temperature, the
relative humidity and the mass of air entrained in a given sized ammonia release.
Based on the model, the calculation of hazard ranges is discussed and two
instances of releases from pressurized containers, that are fairly well described,
are used for comparison. It is concluded that the mixture of ammonia and air
resulting from a certain release from a pressurized tank is likely to be denser
than air. A detailed account of the Houston, Texas anhydrous ammonia release
 Ammonia Storage and Transportation-Safety 339

following the crash of a tank semi-trailer in a highway accident is given in [37].

The report shows a map of the area and also photographs of the accident site
and the cloud at 1, 2, 3 and 4 minutes after the crash.
A review of the production of dense gas mixtures from ammonia releases is
given in [2]. The paper concentrates on issues in the modeling of accidental
releases. It lists a number of major ammonia releases and discusses some
differences between spillage of refrigerated liquid and release through smaller or
larger holes from pressurized containers. It points out that in the spillage of
refrigerated liquid, the boiling is likely to be gentle and without a large amount
ofliquid ammonia ejected into the air, except possibly for the very early stages of
such spill.
The importance of containing a spill of refrigerated ammonia within a space
with a fairly small liquid surface area so as to restrict heat transfer and boiling
has been pointed out earlier. In the spillage of refrigerated ammonia, the initial
pool will not remain at - 33°C. Buoyancy of ammonia vapor will cause strong
convection currents leading to evaporation and cooling down of the pool to
- 50°C or possibly lower. Atmospheric conditions are very important in
relation to spills of refrigerated as well as hot ammonia.
In the case of one rail car failure, it was pointed out that a ground fog was
reported, the wind was calm, and a temperature inversion had occurred [53]. In
the Houston semi-trailer accident in which about 16 t of ammonia was released,
witnesses reported that the white ammonia vapor cloud initially reached a
height of about 30 m before being carried by the wind for a distance of
approximately 800 m. Within 3 minutes of release the maximum width of the
vapor cloud over the ground was about 300 m and after 5 minutes, when most of
the liquid ammonia had boiled off, the cloud had dispersed [36]. The import-
ance of wind direction and strength cannot be overemphasized. It is evident in
the accounts from many accidents.
A specific question relates to the use of water on ammonia releases. If a
stream of liquid water is directed into a liquid pool of anhydrous ammonia,
there is an exothermic reaction and a large release of ammonia vapor results. On
the other hand, water is a main tool to use to absorb and control ammonia
vapors. Fog nozzles can be used to contain a release of ammonia and prevent
the spreading of a cloud downwind. Rapid and wide rotation of fog nozzles
increases the efficiency of ammonia absorption. When using the right proced-
ures, as demonstrated in many ammonia vapor control exercises, a "capture"
zone can be created downwind to prevent the cloud from spreading [54].
In the Pensacola accident, 50% ofthe contents of one ammonia rail car were
released within 10 minutes, while the contents of another ammonia rail car were
slowly vaporized over 12 hours. A lethal cloud engulfed a populated area within
minutes, and before the residents could be evacuated. They were later escorted
on foot for distances from 60 to 300 m through the ammonia cloud, until out of
the affected area. One of the lessons learned in this accident and also in the
Crete, Nebraska accident, in which 3 anhydrous ammonia tank cars were struck
by a passing train and one tank ruptured, is that people who stay in their homes
340 A. Nielsen

closing doors and windows survive, and that they can help themselves by
stuffing towels in openings to make the house airtight [28, 55]. (The chance of
survival in a house will depend on the time the house is engulfed in an ammonia
cloud, the conditions of the cloud and the tightness of the house). In the
Pensacola accident the cloud travelled almost 15 miles in a period of an hour
before dissipating [55]. Fire fighters directed a hose into the liquid ammonia
stream and backed this up with a water fog to knock down the vapors which at
the time had reached about 40 m upwind from the spill [48]. A two-phase cloud
that formed during release from a pressure tank is discussed in [56] as well as
the atmospheric dispersion of ammonia.
It is reported that much of the information concerning ammonia aerosol
clouds comes from investigators at the scene of the accidents rather than from
controlled experiments. Risk models for the prediction of risk distances during
release of toxic industrial chemicals, including ammonia, are discussed in [57].
Two sizes ofreleases are considered: one 15 to 25 t and one 100 to 150 t. The risk
distances are given for two wind velocities.

7.8 Ammonia Spill Studies

A number of companies or groups of companies have carried out studies of

ammonia spills. A number of French companies and organizations carried out a
significant number of spill tests on land. A total of 14 spills were carried out
involving quantities of ammonia between 80 kg and 1 t. Some liquid spills were
vertically ejected into the air, others were parallel to the ground; also gaseous
emissions were studied. Several spills of 100 to 300 kg ofliquid ammonia, ejected
vertically over 1 to 3 minutes from a 50 mm tubing bent upwards, were carried
out. The atmospheric conditions are given. The wind velocity was from 0 to
30 m/s. It is found in all ejections that a cloud was formed right from the exit
from the pipe. The liquid was ejected under a pressure of at least 6 atm and
apparently a very large fraction appeared as microdroplets which formed an
aerosol that was very stable due to the low wind velocity and the low
temperature of the cloud. Measuring devices were installed at three distances
from the emission point and there are several photographs of the clouds.
Ammonia concentrations in the 1 to 2 % range were measured at 40 m from the
ejection point. The cloud went up to approximately 20 m and then fell back
towards the ground where it stretched out and could be followed for about
100 m until it disappeared in a forest area. In some tests the analyzer furthest
away from the emission test was 175 m distant. In several of the tests maximum
concentrations in excess of 1000 ppm were observed at that distance [58].
A report on ammonia spills at Frenchman's Flat has been given [59,60].
Four tests were carried out with ammonia quantities of 25 to 60 m 3
(10-41 tons). Test number four was particularly interesting since it involved
 Ammonia Storage and Transportation-Safety 341

41 tons of ammonia and was conducted under the most stable atmospheric
conditions. An extensive instrumentation system was set up to measure width
and height of the ammonia cloud as it moved in the wind direction, particularly
at 100 m and 800 m from the spill point. Further measurements were carried out
by portable ground level stations at 1.4, 2.8 and 5.5 km downwind. Preliminary
results of maximum ammonia concentrations measured in test four were as
follows: 100 m 6.5%, 800 m 2.1 % and 2800 m 0.5%. It is noteworthy that a fatal
concentration of 0.5% ammonia was registered at a distance of 2800 m.
Reference [93] includes a simulation of Test No. 1.
A study on the predicted hazards from ammonia spills on and under water
has been reported [61-63]. An important aspect is the partition between
ammonia dissolving in water and that being released to the atmosphere. This
partition function is higher, close to 0.9, when liquid ammonia is released
underwater, and lower, depending on release conditions and atmospheric
conditions, when a large quantity is released instantaneously on the water
surface where liquid ammonia will boil to produce saturated ammonia vapor,
and the reaction between ammonia and water also contributes to the vapor
formation. Another phenomenon is the aerosol formation as fine drops of
ammonia are thrown into the air. Partition function ratios of approximately
0.56 have been found for such releases. It must be pointed out that this is an
extrapolation from smaller spills of approximately 10 and 200 kg of ammonia,
the first in a swimming pool and the second on the surface of a lake. Models
have been established. The situation for release on a river such as from a barge
carrying liquid ammonia are studied in detail. Two cases treated are a 200t
surface release and a 3000 t surface spill. It is pointed out that a liquid ammonia
barge that sinks in water may eventually release liquid or gaseous ammonia. The
partition between ammonia picked up by water and ammonia released to the
atmosphere depends on many factors; an important one for underwater releases
is the ratio of the water depth to the pipe diameter of release. The model predicts
that if the depth is on the order of 20 outlet pipe diameters, the partition ratio
may be about 0.9, and it may approach 1.0 for greater depths. One of the
conclusions of the report is that vapor clouds formed from spills on water,
although containing a fraction of aerosols, will have buoyant behavior depend-
ing on wind velocity.
The model has been used to describe the simulation of a number of
situations of transport of hazardous cargo by ship [64]. One of the situations
covered is a fictitious anhydrous ammonia casualty at Louisville, Kentucky. The
greatest potential hazards in ammonia spills, on or below the surface of water,
are obviously those associated with barge transport on rivers and in narrow
coastal waters, and those associated with tanker transport in narrow streams
and harbors.
342 A. Nielsen

7.9 Ammonia Storage Facilities

The largest ammonia storage facilities are located as part of ammonia producing
plants, or at large distribution centers or terminals. A large number of smaller
storage tanks are typically operated by ammonia distribution companies and
thousands of small tanks are used in distribution and local storage. In Denmark
where the use of liquid ammonia as a direct application fertilizer is wide spread,
there are on the order of 10000 small tanks [20].
A large amount of ammonia is stored in the transportation system in vessels,
barges, tank cars and in ammonia pipelines. According to [65] the Gulf Central
Pipeline holds about 70000 tons of ammonia.
For transfer of - 33°C ammonia, e.g., from an ocean going tanker, into a
pressure or semi-refrigerated storage installation, an ammonia heating install-
ation is required. The actual temperature and pressure in refrigerated storage
tanks depends upon the location of the storage, whether at sea level or at a
higher elevation.
The pressurized storage tanks are typically made for capacities of up to
270 tons, but are found in all sizes down to a few cubic meters in size and even
smaller pressurized containers are widely used in industry down to sizes of one
to a few kg capacities.
For maintaining safety around ammonia installations rules, and regulations
exist in the various countries and even the smallest containers are subject to
rigid and detailed specifications [17]. A detailed survey of refrigerated ammonia
storage tanks in the U.S. and Canada, with information on location and design
features of such tanks, is given by Hale [16]. A HAZOP study of a fully-
refrigerated storage installation is described in [90]. Cf. also [91] on topics of
risk analysis and emergency management.
Two underground ammonia facilities are described. One storage cavern of a
capacity of 20000 tons of ammonia has been operated by Dupont, and Norsk
Hydro, is operating underground storage in Norway of approximately 50000
tons in size [66].
Modern ammonia plants typically in the size of 1000-2000 tid capacity, in
most cases include one or more atmospheric storage tanks as part of the facility.
Details of the design and construction of a high~integrity ammonia storage tank
is given in [84]. Comeau [67] discussed precautions for ammonia storage tanks
and particularly the design of surrounding protection dikes. The function of
such dikes is to keep evaporation to a minimum in the event of a leak and to
protect the tanks against external physical damage. Many consider the best
protection of an atmospheric storage tank to be a high wall of prestressed
concrete around the tank which would further limit evaporation if the tank fails.
One of the technical and safety aspects of storing anhydrous ammonia in
atmospheric tanks is to protect the soil below the tank from freezing. This
problem and experience with tank foundation heaters are discussed by Comeau
and Weber [43]. Special problems arise in the shutdown and startup of
 Ammonia Storage and Transportation-Safety 343

refrigerated atmospheric ammonia storage tanks [68,69]. One has to be

particularly careful in certain situations not to create a vacuum inside the tank.
An inspection outline for retrofitting refrigerated ammonia storage for improved
safety and economy is given in [70].
The high integrity ammonia storage tank described in [84] is characterized
by double-integrity steel walls and a double-integrity bottom. The bottom is
supported on an elevated foundation above grade, and the tank top is provided
with a suspended deck. Such a construction is obviously safer than a single-wall
atmospheric pressure storage tank, supported on the soil with only a layer of
insulation, a layer of sand and gravel and with an electrical heating system
below the insulation to protect against freezing of the soil. However, in a risk
assessment, not only the tank itself but the complete installation should be
considered [88].
As a further safety dimension an ammonia detection system in an area
around the storage facility may be used. Such a system would give indications of
any leaks from the ammonia storage and record even a few ppm of ammonia in
the atmosphere [85]. The knowledge learned on stress corrosion cracking,
construction, safe operation, and maintenance has contributed greatly to the
safety of atmospheric ammonia storage systems.

7.10 Ammonia Transportation Systems

Much of the ammonia manufactured in the world is transported for various

distances. Plants located in the areas with abundant natural gas and low
population density may transport close to 100% of the ammonia production or
40-60% if combined with some capacity to make finished fertilizers. In the U.S.
approximately 25% of the ammonia produced or imported is subjected to
transportation.
Ocean transportation of ammonia is by ammonia tankers; river and coastal
transportation by barges; and overland transportation by pipelines, rail or road
tankers. Ammonia tankers may be designed as semi-refrigerated or fully re-
frigerated vessels [25]. Semi-refrigerated carriers typically contain up to
15000 m 3 while fully refrigerated tankers have capacities of up to 46500 tons
[65]. Ammonia barges may have two cylindrical fully refrigerated tanks [25].
Transportation of ammonia by pipeline is discussed in [65], [72]. In the U.S. the
MidAmerica Pipeline System and the Gulf Central Pipeline transport ammonia
from high production areas in Texas and Oklahoma, and Texas and Louisiana
respectively to heavy agricultural-use areas in the Midwest. A small ammonia
pipeline connects an ammonia plant in Northeastern France with a fertilizer
industry in Western Germany. The longest ammonia pipeline is found in the
USSR. Its total length is 2424 km connecting Togliatti, Gordlovka and
344 A. Nielsen

Grigor'evskii. The USSR pipeline has a throughput of 2.5 million t/year and
crosses several major rivers.
According to [65] ammonia transportation by pipelines requires the am-
monia to be heated up, at least, to 2°C, which in most cases means that it must
be warmed at the supply terminal and cooled again to - 33°C at the receiving
terminal. For transportation of ammonia, particularly by pipeline, the PVT
properties of ammonia with a small content of water are important. Such data
are found in [73]. The Interstate Commerce Commission requires that all
anhydrous ammonia transported interstate by pipeline must contain a min-
imum of 0.2 wt % of water [73]. The water is added as a corrosion inhibitor and
will, according to experience gained by a survey of ammonia storage spheres,
limit the risk of stress corrosion cracking in surfaces exposed to ammonia liquid
[74, 75]. It is also stated in [75] that contamination with air is the primary cause
of stress corrosion cracking in ammonia; and [75] also discusses the phenomena
that the oxygen content is higher in the gas phase of the tank while water is
contained predominantly in the liquid, and thus a fairly complex picture for
stress corrosion cracking is developed which is important for transportation of
ammonia in any tank. In [17] is given a survey of the various codes and
regulations, covering storage and handling of ammonia in the USA as regulated
through OSHA. It is emphasized that additional municipal, country and state
regulations exist in many locations. It is pointed out that shipping containers
constructed of quenched and tempered steel must contain, at least, 0.2 wt %
water, which is also a requirement for ammonia transported by pipeline [65].
Rail cars can typically hold 26 tons of ammonia, while larger jumbo cars
may hold up to 80 tons and road tankers are, in the U.S., limited to 25 tons of
ammonia. One of the minimum requirements in the U.S. is that containers such
as road and rail tankers used for ammonia shipments must not be filled to more
than a certain percentage by volume to provide a cushion for thermal expansion
should its temperature increase [17]. The tanks are typically designed for 15.5
bar (vapor pressure of ammonia at 43°C) and provided with pressure relief
devices [17]. Detailed description of a tanker loading station for liquid am-
monia is given in [76]. The facility described has the capacity to handle the
loading of 12 tankers per day and to unload 15 tankers per day. Details of a
computer-aided ammonia rail car station are given in [32] including a descrip-
tion of the computer hardware and software used in the facility.

7.11 References

1. Blanken JM (1980) Ammonia Plant Saf 22: 25

2. Griffiths RF, Kaiser GD (1982) J Hazard Mater 6: 197
3. Charp C (1966) Agric Anydrous Ammonia Technol Use Proc Symp, St. Louis, MO, p.21
4. Harris GFP, MacDermott PE (1977) Inst Chern Eng Symp Ser No 49:29
5. Nuttonson MY (1972) PB-209478, 1
 Ammonia Storage and Transportation-Safety 345

6. National Institute for Occupational Safety and Health (1974) Rockville, MD, PB-246669, p.l
7. National Research Council (1972) Committee on toxicology, Washington, DC, PB-244336, P 1
8. Legters L (1980) AD-A094501 p 1
9. Lessenger JE (1985) Plant Oper Progr 4: 20
10. Barber JC (1978) Ammonia Plant Saf 20: 5
11. Morgan GO, Reed JD (1965) Ammonia Plant Saf 7: 38
12. Zakaznov VF, Kursheva LA, Upadyshev KL (1980) Khim Prom (Moscow) 12: 361; Soviet Chern
Ind (1980) 12: 728
13. Hale CC, Lichtenberg WH (1980) Ammonia Plant Saf 22: 35
14. Michels AMJF, Dumoulin EM, Gerver JH, (1957) Rec Trav Chim 76: 5
15. Leleu J (1976) Inst NatI Rech Secur Paris Cah Notes Doc No. 1024-8476,427
16. Hale CC (1984) Ammonia Plant Saf 24: 181
17. Brenchley DL, Athey GF, Bomelburg HJ (1981) PNL-4006: 1
18. Karches GJ, Froehlich PA, Bicknell RJ (1978) PB 80-189475: 1
19. Husa WH, Bulkley WL (1965) Ammonia Plant Saf7:41
20. Arup H (1977) Ammonia Plant Saf 19:73
21. Lichtenberg WH (1972) Ammonia Plant Saf 14: 24
22. Rohleder GV (1969) Ammonia Plant Saf 11 : 35
23. Inkofer WA (1969) Ammonia Plant Saf 11 :40
24. Ludddeke DE (1975) Ammonia Plant Saf 17:99
25. Briley GG (1967) Ammonia Plant Saf 9: 10
26. Hakansson R (1977) Ammonia Plant Saf 19: 119
27. Caserta LV (1972) Ammonia Plant Saf 14:31
28. Heller FJ (1981) Ammonia Plant Saf 23: 132
29. National Transportation Safety Board (1976) Washington, DC PB 267939, P 1
30. Cato GA. Dobbs WF (1971) Ammonia Plant Saf 13: 1
31. O'Driscoll JJ (1974) Abtr Papers 78th Nat Meeting AIChE Salt Lake City Paper 35D, p 1
32. Arseneaux AA (1985) Ammonia Plant Saf 25: 150
33. Olsen EA (1969) Ammonia Plant Saf 11 :46
34. Conseil Europeen des Federations de l'Industrie Chimique (1979) Eur Council Chern ManufFed
Safety Card-CEFIC TEC-T-I-Rev 3
35. Medard L (1970) Ammonia Plant Saf 12: 17
36. National Transportation Safety Board (1977) Washington, DC PB-268251, p 1
37. Nfitional Transportation Safety Board (1979) Washington, DC PB-80-144942, p 1
38. Nielsen A (1971) Ammonia Plant Saf 13: 103
39. Hutchings J, Sanderson G, Davies DGS, Davies MAP (1972) Ammonia Plant Saf 14: 102
40. Phelps EH (1972) Ammonia Plant Saf 14: 109
41. Lonsdale H (1975) Ammonia Plant Saf 17: 126
42. Lichtenberg WH (1977) Ammonia Plant Saf 19: 59
43. Comeau ET, Weber ML (1977) Ammonia Plant Saf 19:63
44. Esrig MI, Ahmad S, Mayo HC (1975) Ammonia Plant Saf 17: 93
45. Winegar BW (1980) Ammonia Plant Saf 22: 226
46. Sterling MB (1977) Ammonia Plant Saf 19: 77
47. Transportation Research Board (1983) Washington, DC PB-84-143635, P 1
48. Stueben WJ, Ball WL (1979) Ammonia Plant Saf 21 : 76
49. Day BF (1978) Ammonia Plant Saf 20: 30
50. Eiber RJ (1981) Ammonia Plant Saf 23: 146
51. Ball WL (1970) Ammonia Plant Saf 12: 1
52. Kaiser CD, Walker BC (1978) Atoms Environ 12: 2289
53. National Transportation Safety Board (1969) Washington, DC, PB-198790, p 1
54. Greiner ML (1984) Ammonia Plant Saf 24: 109
55. National Transportation Safety Board (1978) Washington DC, PB-28325, p 1
56. Kansa EJ, Ermak DL, Chan ST, Rodean HC (1983) UCRL-88649-Rev 2, p 1
57. Andersson JO, Broxvall A, Karlsson E, Karlsson N, Nyren K, Rejnus L, Winter S (1983) FOA-C-
40183-C2
58. Resplandy A (1969) Chim Ind Genie Chim 102: 691
59. Goldwire HC (1986) Chern Eng Prog 82:35
60. Goldwire HC et al. (1985) UCID-20562
61. Raj PK, Hagopian JH, Kale1kar AS (1974) CG-D-74-74, AD-779400, pI
62. Raj PK, Hagopian JH, Kale1kiar AS, Cece J (1975) Ammonia Plant Saf 17: 102
346 A. Nielsen

63. Raj PK, Reid RC (1978) Environ Sci Technol12: 1422

64. National Research Council Panel on Response to Casualties Involving Ship-Borne Hazardous
Cargoes (1979) AD-A075203, p 1
65. Hignett TP (1979) Transportation and Storage of Ammonia. Fertilizer Industry Round Table,
Washington, DC
66. Sherman JL (1970) Ammonia Plant Saf 12:8
67. Comeau ET (1972) Ammonia Plant Saf 14:69
68. Bell HV (1982) Chem Eng Prog 78: 74
69. Vick KA, Witthaus JB, Mayo WC (1980) Ammonia Plant Saf 22: 54
70. Hale CC, Josephson AJ, Mattick DE (1985) Ammonia Plant Saf 25: 172
71. Schrader RF (1979) Nitrogen No. 117,26
72. Kharlamov VV, Tsybal YM (1983) Zh Vses Khim Ob-va 28: 103
73. Inkofer W, Wilson GM, Adams JE (1971) Ammonia Plant Saf 13:67
74. Blanken JM (1984) Ammonia Plant Saf 24: 140
75. Lunde L(1984) Ammonia Plant Saf 24: 154
76. Schlichtharle G, Huberich T (1983) Plant Oper Progr 2: 165
77. Klem TJ, Saunders JC (1986) Ammonia Plant Saf 26: 145
78. McRae MH (1987) PlantjOper Progr 6: 17
79. Markham RS (1987) Ammonia Plant Saf 27: 136
80. Byrne JR, Moir FE, Williams RD (1989) Ammonia Plant Saf 29: 122
81. Ryer-Powder JE (1991) Ammonia Plant Saf 31: 93
82. Mark Wilhelm S (1991) Ammonia Plant Saf 31: 20
83. Selva RA, Heuser AH (1990) Ammonia Plant Saf 30: 39
84. Thompson JR, Carnegie RN (1989) Ammonia Plant Saf 29: 116
85. Thompson JR, Sekula LJ, Whitson PE (1992) Ammonia Plant Saf 32: 77
86. Andersson BO (1991) Ammonia Plant Saf 31 : 5
87. Tilton IN, Squire RH, SafJIe CS, Atkins CR (1992) Ammonia Plant Saf 32: 63
88. Squire RH (1991) Ammonia Plant Saf 31: 131
89. Appl M, Fassler K, Fromm D, Gebhard H, Port! H (1990) Ammonia Plant Saf 30:22
90. Lawley HG, Shepherd JS (1987) Fertilizer Focus May: 57
91. Ham JM, Gansevoort J (1992) Ammonia Plant Saf 32: 86
92. Lunde L, Nyborg R (1991) Proc No. 307, The Fertilizer Society, London, p 1
93. Statharas JC, Bartzis JG, Venetsanos A, Wurtz J (1993) Process Safety Prog 12: 118
94. Hallan T (1994) Ammonia Plant Saf 34 to appear
95. Nyborg R, Lunde LR, Conley MJ (1991) Mater Perform 30: 61
96. Nyborg R, Lunde L (1995) Ammonia Plant Saf 35 to appear
97. Bickel W, Fassler K, Geis B, Guns L, Reininghaus J, Walter M (1995) Ammonia Plant Saf 35 to
appear
 Spri nger-Verlag
and the Environment

We at Springer-Verlag firmly believe that an

international science publisher has a special
obligation to the environment,'and our corpo-
rate policies consistently reflect this conviction.

We also expect our busi-

ness partners - paper mills, printers, packag-
ing manufacturers, etc. - to commit themselves
to using environmentally friendly materials and
production processes.

The paper in this book is made from

low- or no-chlorine pulp and is acid free, in
conformance with international standards for
paper permanency.