Progress in Reaction Kinetics and Mechanism. Vol. 37, pp. 411–422.

2012
doi: 10.3184/146867812X13452764677492 1468-6783 # 2012 Science Reviews 2000 Ltd

RESEARCH PAPER

Synthesis of ethylenediamine in a tubular reactor:

experimental and theoretical kinetics
Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian*
Department of Chemical Engineering, Zhejiang University, Hangzhou 310027, P.R. China
*E-mail: supmoney@163.com

ABSTRACT
The mechanism and kinetics of the ammonolysis of 1,2-dichloroethane in the
liquid phase have been investigated using a combination of experimental and
theoretical methods. The experiments was carried out in tubular reactors. The
reaction as carried out at 6 MPa and 160
C, with 5 min of residence time, 60%wt
of ammonia concentration and a reactant molar ratio n(ammonia): n(1,2-
dichloroethane) ¼ 30 : 1 gave the optimum result. The parameters of the reaction
kinetics were obtained using linear regression, and the reaction rate constant at a
low conversion of 1,2-dichloroethane is 1.12 s  1 (160
C, 6 MPa) according to the
calculated kinetics equation. Theoretical calculation was performed using density
functional theory and the reaction profile so calculated gave the dominant
pathway. The results of calculation also indicated a reaction order of
approximately 1 at low conversions of 1,2-dichloroethane with a related rate
constant of 0.51 s  1 (160
C, 6 MPa), which agreed with the experimental results.

KEYWORDS: ammonolysis, 1, 2-dichloroethane, tubular reactor, density

functional theory

1. INTRODUCTION

Ethylenediamine plays an important role in the chemical industry. Among all the
methods used to synthesize ethylenediamine, the ammonolysis of 1,2-dichlor-
oethane is most economical. Thus it is important to perform investigations on this
reaction both experimentally and theoretically. Rahimizadeh et al. [1] have
reported a kinetic study of the ammonolysis of 1,2-dichloroethane, obtaining
the detailed reaction parameters in nonaqueous media. However, the ammonolysis
of 1,2-dichloroethane in an aqueous medium differs much from that in

411
412 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

nonaqueous media as ammonia is partly ionised in water so that the reactions are
more complicated. Thus a study of the reaction parameters of the ammonolysis of
1,2-dichloroethane in water is highly desirable.
The liquid-phase ammonolysis of 1,2-dichloroethane is a nucleophilic
substitution process with ammonia (NH3) acting as a nucleophile. In the
present work, the mechanism of the ammonolysis of 1,2-dichloroethane is
addressed using a combination of experimental and theoretical methods. An
apparent kinetic model is derived from experiments carried out in tubular reactors
without a catalyst, while density functional theory [2 – 5] (DFT) is used to model
the reaction theoretically in the liquid phase. The objectives of this work are (i) to
get a general understanding of the reaction between 1,2-dichloroethane and
aqueous ammonia, (ii) to gain an insight into the reaction pathways, and relative
energy barriers of each step and (iii) to obtain experimental and theoretical kinetic
data, and provide a suitable method for the theoretical study of similar reactions.

2. EXPERIMENTAL AND CALCULATION METHODS

The experiment was performed in a tubular reactor, as shown in Figure 1. The

inner diameter of the tubes was 2 mm and the outer diameter was 6 mm. The
reactions were carried out at high pressure to keep a liquid phase system, and the
volumetric flow rate throughout the reactor could be considered constant for the

Figure 1 Scheme of the tubular reactor, in which V-0 is liquid ammonia cylinder, V-1 is
aqueous ammonia cylinder, V-2 is dichlorethane cylinder, P-1 and P-2 are
dosage pumps, B-1 is back valve, M is static mixer, R-1 is tubular reactor, V-3
is condenser and V-4 is separator. The tubular reactor is made of stainless steel
tubes that are 6 mm in diameter and 2 mm thick. Silicone oil is used as heating
medium. Pressure is controlled through valve B-1.

www.prkm.co.uk
 Synthesis of ethylenediamine in a tubular reactor 413

pressure drop of the fluid though the reactor was low (less than 0.05 MPa after
3 min of residence time). The velocity of the fluid was fixed at a high level (0.2 m
s  1) to ensure that the Reynolds number was larger than 4000 in order to keep
turbulent flow in the tube. When calculating Reynolds numbers, the average fluid
density and viscosity were used. The density and viscosity were estimated as
1.16103 kg m  3 and 1610  4 Pa s respectively. Tubes of different lengths were
used to perform reactions with different residence times. Ethylenediamine was
analysed using GC (Agilent 1790) quantitatively, while 1,2-dichloroethane was
analysed by titrating the chloride anion using the Mohr method [6]. All
concentrations were measured under steady-state operations. The concentrations
of product and by-product were measured at the same residence time. The
influence of operation conditions on the reaction was investigated. When
investigating the influences of temperature and pressure, the residence time was
fixed at 5 min, so that thermodynamic equilibrium was approached. The apparent
kinetics were investigated using the following approach.
Two consecutive reactions can be represented by:
k1
A þ D ?B ð1Þ
k2
A þ B ?C ð2Þ

If the concentration of D is constant, the kinetic equations can be established as:

dCA
¼ k1 CAa  k2 CAb CBg ð3Þ
dt
dCB
¼ k1 CAa  k2 CAb CBg ð4Þ
dt
dCC
¼ k2 CAb CBg ð5Þ
dt

where t represents the residence time. Then the following equation can be
deduced:

ln dCC ydt ¼ ln k2 þ b ln CA þ g ln CB ð6Þ

The reaction rate constant is related to the activation energy by

Ea
ln k ¼ þC ð7Þ
RT

from which we can get the activation energy using linear regression.

www.prkm.co.uk
414 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

The reaction between 1,2-dichloroethane and ammonia consists of two

elementary reactions, the aminations of 1,2-dichloroethane and 2-chloroethyla-
mine respectively. However, the concentration of chloroethylamine is always too
small to detect, so the two elementary reactions can be regarded as one when
investigating the apparent reaction kinetics. The concentration of by-products is
high enough to be detected when the conversion of 1,2-dichloroethane is higher
than 20%. Thus the possible reactions in the liquid phase can be shown as:
k1
R þ Nu ?P ð8Þ

k2
R þ P ?B ð9Þ

where R represents 1,2-dichloroethane, Nu represents ammonia, P represents

ethylenediamine, and B represents the by-product. When the conversion of 1,2-
dichloroethane is lower than 20%, there is a trace of B in the system, then the
apparent kinetics equations can be assumed as:
d½P
¼ k1 ½Rn1 ½Nun2  k2 rn3 ½Pn4 ð10Þ
dt

d½B
¼ k2 ½Rn3 ½Pn4 ð11Þ
dt
where t represents the residence time. The apparent kinetic parameters can be
obtained using linear regression with the above equations.
In order to compare with the results of theoretical calculation, the initial
apparent kinetic parameters were measured. Taking into account that [Nu] is
constant and [B] is approximately zero at low conversion of 1,2-dichloroethane,
the initial apparent kinetic equation can be established as:
d½P
¼ k1 ½Rn1 ð12Þ
dt
of which a derivative is:
d½P
ln ¼ ln k þ n1 ln½R ð13Þ
dt
With the experiment data, k and n1 can be obtained using the linear regression
method. The residence time is constrained to be less than 100 s to keep a low
conversion of 1,2-dichloroethane.

www.prkm.co.uk
 Synthesis of ethylenediamine in a tubular reactor 415

Calculation

All calculations were performed using GAUSSIAN 03 software [7]. The hybrid
DFT theory was applied to this problem with the B3LYP method, which uses
Becke’s three-parameter hybrid functional [8], together with the non-local
correlation provided by the LYP expression [9,10], and VWN functional III
[11] for local correlation. The polarised continuum (overlapping spheres) model
[12 – 14] (PCM), which is based on a description of the solvent as a macroscopic
continuum medium, was employed for the calculations of the reactions in the
liquid phase with water as solvent. The geometries of the reactants and products
were fully optimised until a stationary point on the potential surface was found.
The geometries of the transition states were searched and reoptimised. Subsequent
calculations, such as the frequency calculation and the intrinsic reaction coordi-
nates [15,16] (IRC) calculation were performed to confirm the transition states as
the saddle points of the reactions. The frequency calculations on the optimised
geometries were also used for the zero point energy (ZPE) corrections to calculate
the thermodynamic properties.
The thermodynamic properties were calculated for a pressure of 6 MPa and


160 C. The rate constants of the reactions were also calculated. As there are
continuous interactions between each molecule and its neighbours, even the
simplest physical transport processes cannot be treated accurately in liquids. There
is thus no satisfactory general theory of solution kinetics. However, the rates can
be predicted using gas-phase theories in most cases when the influences of
neighbours on the reactants are relatively small, making it reasonable to consider
for reactions two-molecule collisions only. The activated-complex theory of
chemical kinetics is used to predict the rate constants from the properties of
molecules [17], which is given by:
6¼ 6¼
kB T Q6¼ DE0 k T DG0
k¼k NA e RT ¼ k B e RT ð14Þ
h Qr1 Qr2 h

where kB , h, and NA are the Boltzmann, Planck, and Avogadro constants

respectively, k is the transmission coefficient which is assumed to be 1, Q6¼ is
the single-molecule partition function per unit volume for the transition state, Qri
is the single-molecule partition function for reactant i, DE06¼ is the difference
between the ground-state energies of the transition state and the reactants, DG0 is
the Gibbs free energy change between the transition state and reactants. In our
work, the solvent effect was considered in the calculation of geometries and

www.prkm.co.uk
416 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

energies of reactants, intermediates, products and transition states. The main effect
of water on the reactants is on the concentration of ions, so we believe that the
above equation can be used in this system.

3. RESULTS AND DISCUSSION

The reaction conditions were firstly optimised using parallel experiments. The
influence of ammonia concentration is shown in Table 1. The conversion of 1,2-
dichloroethane increases with ammonia concentration. Pressure also has a great
effect on the reaction. The conversion increases with system pressure. However,
the conversion remains nearly the same when the pressure is higher than 6 MPa.
The results are shown in Table 2. Figure 2 shows the influences of residence time
and temperature, from which it can be concluded that reactions at higher
temperature reach equilibrium faster, and a lower temperature favours the yield
of ethylenediamine.
The reagents used in this work, 1,2-dichloroethane and aqueous ammonia,
form a heterogeneous system at room temperature and normal pressure, so mass
transfer is inhibited under these conditions. Only when the reagents are in
supercritical states is the system homogeneous. However, the critical temperatures
of 1,2-dichloroethane and water are too high. To enhance mass transfer, the
practical method is to maintain turbulent flow in the reactor. A higher temperature

Table 1 Influences of ammonia concentrationa

Concentration Conversion Selectivity
of ammonia of dichlorethane of ethylenediamine
(%) (%) (%)
20 76.5 51.7
30 84.4 51.5
40 98.6 52.1
60 98.7 52.3
a
6 MPa, 160
C, n(ammonia) : n(1,2-dichloroethane) ¼ 30 : 1, 5 min of residence time.

Table 2 Influences of system pressurea

Pressure (MPa) 2 6 9
Selectivity of ethylenediamine (%) 50.3 52.3 52.2
Conversion of dichlorethane (%) 61.0 98.7 98.8
a
Reaction condition: 160
C, ammonia concentration 60%,
n(ammonia) : n(1, 2-dichloroethane) ¼ 30 : 1, 5 min of residence time.

www.prkm.co.uk
 Synthesis of ethylenediamine in a tubular reactor 417

130
C
160
C
190
C

Figure 2 Influence of residence time and temperature, at ammonia concentration of

60%wt and n(ammonia) : n(1,2-dichloroethane) ¼ 30 : 1.

favours the reaction rate but decreases the yield of ethylenediamine, while a lower
temperature increases the equipment cost. A solution is to use a multi-section
tubular reactor with different temperatures in different sections, the temperature of
the forepart is high to obtain a fast reaction rate while the temperature of the rear
part is low to increase the yield.
The apparent kinetic parameters calculated using our experimental data are
listed in Table 3. The average calculated values of n1 , n3 and n4 are 0.963, 0.941
and 0.032 respectively. From the rate constants at different temperatures, the
activation energy of ammonolysis of 1,2-dichloroethane can be calculated to be
53.8 kJ mol  1, i.e. the rate at different temperature can be calculated from
Eqn (15)

ln k1 ¼ 53775yRT þ 14:288 ð15Þ

Table 3 Apparent kinetic parameters calculated using experimental data

Temperature k1 n1 k2 , n3 n4
(
C) (s  1) (L mol  1 s  1)
130 0.168 0.968 0.037 0.935 0.030
160 0.544 0.962 0.179 0.947 0.032
190 1.340 0.959 0.696 0.940 0.033

www.prkm.co.uk
418 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

yield of ethylene diamine

Time (s)
Figure 3 Kinetic profile for the yield of ethylenediamine at 6 MPa and 160
C.

However, the rate constants so calculated are average values and to compare them
with the theoretical results, the initial reaction rate constant must be calculated.
Our investigation on the initial apparent kinetics was performed within 60
seconds of the residence time. The rate constant calculated using the linear
regression method with Eqn (13) is 1.40 s  0.964, and n1 is 0.964, so the apparent
reaction order can be considered as 1 at low conversions of 1,2-dichloroethane.
Then the data were refitted with the reaction order fixed as 1, giving a rate
constant of 1.12 s  1.

Theoretical results

Nucleophilic substitution causes an inversion of configuration [18,19], so the

product of the first step is cis-chloroethylamine which is unstable and prone to be
transformed to trans-chloroethylamine. Similarly, cis-ethylenediamine will be
transformed to trans-ethylenediamine. There are thus two reaction pathways for
the second step, as shown in Figure 4. The geometries of the reactants,
intermediates and products were optimised using the B3LYP=6-31 þ þ G(d,p)
method. Then the transition state for every elemental reaction was searched on the
potential surface using the B3LYP=6-31G(d) method the larger basis set (6-
31 þ þ G(d,p)) was employed in the subsequent optimisation of the transition
state geometry. The effect of solvent on the geometries is considered by
employing the PCM.

www.prkm.co.uk
 Synthesis of ethylenediamine in a tubular reactor 419

Figure 4 Two possible reaction pathways that have been investigated in the study.

The transition states of all the steps for the two pathways have been located
respectively and the reaction barriers determined. Table 4 shows the Gibbs free
energy barrier for every reaction of the two pathways, as well as the rate constant.
The energetic profiles for the two pathways are shown as Figure 5. All the Gibbs
free energies calculated have been corrected with ZPE.
The reaction between 1,2-dichloroethane and ammonia has a barrier of
22.62 kcal mol  1, it is indispensible for both of the two pathways. The calculated
barrier of this step is higher than any other elementary reaction in the two
pathways, which means the reaction between 1,2-dichloroethane and ammonia is
the rate limiting step for both pathways. In the subsequent steps, the barrier for the
pathway TS1 – TS2 – TS3 – TS4 is 15.85 kcal mol  1, and the rate constant for the
related reaction is 8.9916104 s  1. While for the pathway TS1 – TS5, the barrier

Table 4 Gibbs free energy barriers to the transition states formed along the two pathways
and rate constant for each step respectively (Superscript mark for each TS refers to the
reverse step)
Transition Gibbs Rate Transition Gibbs Rate
state free energy constanta state free energy constanta
barriers barriers
(kcal mol  1) (kcal mol  1)
TS1 22.62 33.99 TS1 0 9.782 1.0376108
TS2 11.85 9.41136106 TS2 0 13.68 1.1166106
TS3 15.85 8.9916104 TS3 0 17.10 2.0846104
TS4 11.09 2.2696107 TS4 0 12.30 5.5546106
TS5 19.72 9.9386102 TS5 0 19.41 1.4276103
a
The unit of rate constants for TS1=TS 0 and TS3=TS3 0 is L mol  1 s  1, for TS2=TS2 0 and
TS4=TS4 0 is s  1.

www.prkm.co.uk
420 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

Figure 5 Energetic profiles for the two pathways.

is 19.72 kcal mol  1, the rate constant for the related reaction is 9.9386102 s  1.
The significant difference in reaction rate between the ‘‘sub-limiting’’ steps of the
two pathways indicates that the reaction tends to process along the pathway TS1 –
TS2 – TS3 – TS4. Although the aminations of cis-chloroethylamine and trans-
chloroethylamine proceed by the same mechanism, a significant discrepancy in
their rates is observed. The probable reason is that the cis-configuration greatly
increases the energy of the transition state. Subsequent discussion on the kinetics
is based on the above approach.
The elementary reactions for the dominant pathway are:
k1
R þ Nu 
/?
  cM þ X ð16Þ
0
k1

k2
cM 
/?
  tM ð17Þ
0
k2

k3
tM þ Nu 
/?
  cP þ X ð18Þ
0k3

k4
cP 
/?
 tP ð19Þ
k4

www.prkm.co.uk
 Synthesis of ethylenediamine in a tubular reactor 421

where R represents 1,2-dichloroethane, X represents chloride anion, Nu represents

ammonia, cM represents cis-chloroethylamine, tM represents trans-chloroethyla-
mine, cP represents cis-ethylenediamine and tP represents trans-ethylenediamine.
It is clear from Table 4 that k1 is smaller than the other rate constants by two or
three orders, and the forward rate constants (k2 , k3 , k4 ) are larger than the reverse
rate constants (k2 0 , k3 0 , k4 0 ). Therefore, the concentrations of cM and tM should be
always very small so that they are undetectable, as found in our experiments.
When the conversion of 1,2-dichloroethane is low, the observed rate equation is:

rtP ¼ k½R ð20Þ

where k can be estimated by k1 [Nu]. With the concentration of nucleophilic

reagent used (approximately 0.015 mol m  3), the rate constant k can be obtained.
At 160
C and 6 MPa, k is 0.51 s  1, i.e. is consistent with the experimental data
(1.12 s  1).
In order to obtain the observed kinetic parameters, several approximations
have been included, but the relative error of the calculated rate constant is still
within one order. Also the calculated results also give an explanation for the low
concentration of chloroethylamine throughout the reaction. Therefore, the kinetic
model used in this study is suitable for this reaction at low conversions of the
reactants. The comparison between the theoretical and experimental results also
indicates that the PCM model is suitable for this and similar systems, probably
due to the low strength of the interaction between the solute and solvent. Also, the
theories employed can be also suitable for the study of similar reactions. If more
accurate prediction of apparent kinetic parameters is required, a more sophisti-
cated mass transfer model needs to be established.

4. CONCLUSIONS

The reaction between 1,2-dichloroethane and ammonia has been investigated both
theoretically and experimentally. The reaction pathways have been probed using
DFT theory, with the transition state located for each elementary reaction, and the
dominant pathway has been determined. The kinetic model is compiled to
estimate the observed rate constant. Also, the theoretical result is consistent
with the experimental data, with a relative difference of one order. More
investigation is needed both theoretically and experimentally to gain a deeper

www.prkm.co.uk
422 Xinzhi Chen, Shaodong Zhou, Haijiang Zhang and Chao Qian

insight into the reaction kinetics. The investigative method used in this article can
be applied to the prediction of other reactions in similar systems which are
difficult to track experimentally.

5. ACKNOWLEDGEMENT

We acknowledge the funding support by a grant from the National Natural

Science Foundation of China (No. 20776127), the National Key Technology
R&D Program (No. 2007BAI34B07).

REFERENCES
[1] Rahimizadeh, M. and Rajabzadeh, G. (1995) Iran. J. Chem. Chem. Eng., 14, 55.
[2] Hohenberg, P. and Kohn, W. (1964) Phys. Rev., 136, B864.
[3] Kohn, W. and Sham, L. (1965) J. Phys. Rev., 140, A1133.
[4] Salahub, Ed. D.R. and Zerner, M.C. (1989) The challenge of d and f electrons. American
Chemical Society, Washington, D.C.
[5] Parr, R.G. and Yang, W. (1989) Density-functional theory of atoms and molecules.
Oxford Univ. Press, Oxford.
[6] Jiao, L.W. (2006) Physical testing and chemical analysis part B: Chemical analsis. 42,
219.
[7] Frisch, M.J. et al. (2003) GAUSSIAN 03, Revision D.01 SMP, Gaussian Inc.,
Pittsburgh, PA.
[8] Becke, A.D. (1993) J. Chem. Phys., 98, 5648.
[9] Lee, C., Yang, W. and Parr, R.G. (1988) Phys. Rev., B, 37, 785.
[10] Miehlich, B., Savin, A., Stoll, H. and Preuss, H. (1989) Chem. Phys. Lett., 157, 200.
[11] Vosko, S.H., Wilk, L. and Nusair, M. (1980) Can. J. Phys., 58, 1200.
[12] Miertus, S., Scrocco, E. and Tomasi, J. (1981) Chem. Phys., 55, 117.
[13] Miertus, S. and Tomasi, J. (1982) Chem. Phys., 65, 239.
[14] Cossi, M., Barone, V., Cammi, R. and Tomasi, J. (1996) Chem. Phys. Lett., 255, 327.
[15] Gonzalez, C. and Schlegel, H.B. (1989) J. Chem. Phys., 90, 2154.
[16] Gonzalez, C. and Schlegel, H.B. (1990) J. Phys. Chem., 94, 5523.
[17] Berry, R.S., Rice, S.A. and Ross, J. (1980) Physical chemistry: physical and chemical
kinetics, John Wiley & Sons, New York.
[18] Ingold, C.K. and Taher, N.A. (1940) J. Chem. Soc., 979.
[19] Ingold, C.K. (1969) Structure and mechanism in organic chemistry. Cornell University
Press, Ithaca.

www.prkm.co.uk