0.2 scale division.

As each division tions of the cryoscopic constant which LITERATURE CITED

0
is equivalent to approximately 0.005 C. were made over a period of 10 months
(see below), this corresponds to a stand- on various lots of benzene. The value (1) Batson, F. M., Kraus, C. A., J. Am.
Chem. Soc. 56, 2017 (1934).
ard deviation of 1 X 10_3° C., which of 1.4% so obtained, and based upon
(2) Beck, A., J. Sci. Instr. 33, 16 (1956).
is the same as the value reported by 23 degrees of freedom, is not signifi- (3) Daniels, F., Mathews, J. H., Wil-
McMullan and Corbett {10). cantly different from the estimate of liams, J. W., Bender, R., Alberty,
If only the first six solutions of Table 0.8% based upon 5 degrees of freedom. R. S., “Experimental Physical
I are considered, the data up to 0.06 The data presented here are for an Chemistry,” 5th ed., p. 68, Mc-
Graw-Hill, New York, 1956.
molal can be represented by the linear ideal solute-solvent system and were (4) Giguére, P. A., Secco, E. A., Can. J.
equation obtained solely to illustrate the per- Chem. 32, 550 (1954).
1.046 X ”3 Ad formance characteristics of the equip- (5) Glasgow, A. R., Jr., Ross, G., J.
m =
(3) Research Natl. Bur. Standards 57,
ment. When cryoscopic measurements
with an estimated standard deviation 137(1956).
are to be made on nonideal systems, ap-
(6) Glasgow, A. R., Jr., Streiff, A. J.,
of 0.4 scale division or 2 X 10~3° C. propriate extrapolations of the data Rossini, F. ~D.,Ibid., 35, 355(1945).
It is from the slope of Equation 3 (mo- must be employed to correct for the (7) Herington, E. F. G., Handley, R.,
lality per scale division) and the cryo- variation with composition of the ap- J. Chem. Soc. 1950, 199.
(8) Herington, E. F. G., Handley, R.,
scopic constant reported by Kraus and parent molecular weight of the solute. J. Sci. Instr. 25, 434 (1948).
coworkers {1, 9) for benzene (5.07° C. (9) Kraus, C. A., Vingee, R. A., J. Am.
per molality) that the estimate of ADVANTAGES OF AUTOMATIC CRYOSCOPY
Chem. Soc. 56, 511 (1934).
0.005° C. per scale division was obtained (10) McMullan, R. K., Corbett, J. D.,
J. Chem. Educ. 33, 313 (1956).
for this work. The following advantages charac- (11) Mikhkelson, V. Ya., J. Anal. Chem.
In practice the cryoscopic constant for terize the apparatus described in this U.S.S.R., 9, 21 (1954).
each new bottle of benzene was cal- report. (12) Müller, R. H., Stolten, H. J., Anal.
culated from the freezing point de- Chem. 25, 1103 (1953).
Once the solute and solvent have been (13) Newman, M. S., Kuivila, H. G.,
pression (expressed in scale divisions) Garrett, A. B., J. Am. Chem. Soc.
measured for an approximately 0.05 weighed into the cryoscopic cell and the
cell has been assembled in the freezing 67, 704 (1945).
molal solution of triphenylmethane in (14) Richards, L. A., Campbell, R. B.,
bath, no further operator attention is Soil Sci. 65, 429 (1948).
benzene.
required. (15) Spooner, D. C., J. Sci. Instr. 29, 96
A 0.05 molal solution of triphenyl- A permanent record of the cooling (1952).
methane in benzene has a freezing point curve is automatically obtained. (16) Stull, D. R., Ind. Eng. Chem., Anal.
about 50 scale divisions below that of The extrapolations required for the Ed. 18, 234 (1946).
the pure solvent (Equation 3). At this determination of the freezing points of (17) Stull, D. R., Rev. Sci. Instr. 16,
solutions are more easily made from a 318 (1945).
concentration the estimated standard
continuous curve than from one plotted (18) Witschonke, C. R., Anal. Chem.
deviation, or error, of 0.4 division 24, 350 (1952).
corresponds to a relative error of 0.8%.
manually from discrete temperature (19) Zeffert, B. M., Hormats, S., Ibid.,
readings. 21, 1420 (1949).
Assuming a negligible error in the weights
of solvent and solute used for a single (20) Zeffert, B. M., Witherspoon, R. R.,
ACKNOWLEDGMENT Ibid., 28, 1701 (1956).
determination, the estimated relative (21) Zemany, P. D., Ibid., 24, 348 (1952).
error for the determination of the cryo- The author gratefully acknowledges
scopic constant from the ratio of the assistance of Harley W. Middleton,
Marie DeVito, and Joyce H. Northrop Received for review May 9, 1957.
molality to freezing point depression
Accepted November 23, 1957. Division
_

should also be 0.8%. This was con- in the construction and testing of the of Analytical Chemistry, 131st meeting,
firmed by sets of replicate determina- equipment. ACS, Miami, Fla., April 1957.

Drying and Decomposition of Sodium Carbonate

ARTHUR E. NEWKIRK and IFIGENIA ALIFERIS
General Electric Research Laboratory, Schenectady, N. Y.

Thermobalance curves are given for primary standard, and the lower limit sodium carbonate at temperatures close
the drying and decomposition of sodium for the initiation of the reaction of so- to its melting point, 851 ° C.
carbonate in the temperature range dium carbonate and silica are the same Richards and Hoover {8) found that
from 25° to 1040° C. using different —i.e., 300° C. This fact is not generally sodium carbonate heated under carbon
crucible materials and atmospheres. recognized. Directions for drying so- dioxide for a long time just below the
The reaction of sodium carbonate and dium carbonate usually specify porce- fusion point lost on fusion only about
silica resulted in a weight loss at tem- lain or platinum containers, but glass 0.003% in weight. Duval {2) reported
peratures as low as 500° C. It is weighing bottles are also used. In this that sodium carbonate, when heated in a
recommended that sodium carbonate laboratory, satisfactory results have thermobalance, is stable in the range
for analytical use be dried in platinum been obtained by drying in glass at from 100° to 840° C. Motzfeldt (7),
to prevent errors due to its reaction 250° C. {1). The recommended tem- who made a careful study of this
with silica and silicates. perature range from 250° to 300° C. is decomposition in a platinum cell, con-
the result of many studies (6), but the cluded that there is no chance for

It is more than a coincidence that the

accepted maximum temperature for
drying sodium carbonate for use as a
upper temperature limit is surprisingly
low in view of the several reliable
reports of the thermal stability of
significant decomposition at any tem-
peratures below 800° C. at least. He
attributes previous discordant results to

982 · ANALYTICAL CHEMISTRY

 reaction of the sodium carbonate with due to water in the sodium carbonate of sodium carbonate with the porcelain
the moisture of the air (or the water as pointed out by Duval (0). In runs or alumina crucible. When the heating
content of the initial sodium bicarbon- 8 through 11 the sodium carbonate is carried out for a long period of time,
ate) to form sodium hydroxide and was dried by heating in weighing bottles the weight loss is approximately quanti-
carbon dioxide. A recent study by of borosilicate glass for 16 hours at tative (Table I).
Easterbrook (3) with the aid of a 350° C., and hence showed no initial When the sodium carbonate is heated
thermobalance showed that sodium water loss. in platinum or gold, the weight loss is
carbonate prepared by heating sodium The large losses in runs 1 through 4, much less rapid and is probably due to
sesquicarbonate reached constant starting in the neighborhood of 800° C., the decomposition of sodium carbonate
weight at 210° C. in 2 hours; there was are chiefly attributed to the reaction to form sodium oxide and carbon
no further change in weight after 19 dioxide as proposed by Motzfeldt (7)
hours at 270° C. or an additional hour at (runs 5, 6, 7, and 11). As would be
340° C. Easterbrook concluded that expected, the presence of a positive
the slightly high titer of the product Table I. Weight Losses in the Reaction stream of gas to flush out the carbon
(maximum calculated purity, 100.023%) of Sodium Carbonate with Porcelain dioxide resulted in a faster rate of
resulted from the formation of oxide and Silica weight loss (run 6 vs. run 5). However,
or hydroxide produced by reaction with when water was present in the gas
Caled. Obsd.
water during the decomposition of the Run Loss, Loss, stream, the observed rate of weight loss
sesquicarbonate. No. Gram Gram was less (run 7 vs. run 6).
The authors studied the reaction 1 0.2076 0.1927 A sample of the sodium carbonate
between sodium carbonate and other 2 0.2076 0.2028 dried at 350° C. showed no significant
substances at elevated temperatures, 9 0.2076 0.2083 change in weight on further heating 12
and heated sodium carbonate and 10 0.2076 0.2087 hours at 600° C. and 4 hours at 650° C.
sodium carbonate plus silica in a Cheve- in a platinum crucible in air.
nard thermobalance using a number of The decomposition of sodium carbon-
’
different crucible materials and atmos- ate at temperatures up to 1040° C.
pheres. The results and additional is completely suppressed by an atmos-
data from the literature suggest that phere of carbon dioxide (run 8).
another factor causing the discordant The reaction of sodium carbonate
results may be reaction of sodium with coarse silica sand occurs rapidly at
carbonate with glass or porcelain used 800° to 850° C. (run 9); the reaction
as a container. temperature is lowered by grinding the
silica (run 10). In this latter run, the
first evidence of reaction as shown by
APPARATUS AND MATERIALS
weight loss is at about 500° C. The
weight loss in these runs wras quantita-
This work was done with a pen- tively equivalent per formula weight to
recording Chevenard thermobalance, the carbonate present (Table I), there
whose construction and performance
have been described (9). Experiments being an excess of silica or silicate in
each case.
in air were performed with the balance
as received, but experiments in con-
trolled atmospheres were performed DISCUSSION
using a quartz liner tube for the furnace
10
'---^ :
j

fitted with a loose cap at the bottom and

a tube for the introduction of gas at the ; |
The above results show that sodium
top. The porcelain crucibles were L Í '-i™' carbonate-silica mixtures will lose
Coors regular laboratory grade. The weight at temperatures as low as
alumina crucible was Triangle RR 500° C. Howarth, Maskill, and Turner
grade supplied by Morgan, England. 1040
(4, 5) have shown that the reaction be-
The gold crucible was spun from pure C"
'

400
' ’

! Í-
2Ó0 6Ó0
tween these compounds can occur even
gold sheet, and the platinum crucibles TEMP. ‘C, TIME HR.

at 300° C. Unfortunately, reports of

were of regular laboratory grade.
The sodium carbonate (General Figure 1. Thermobalance runs with decomposition of sodium carbonate
Chemical Co., anhydrous reagent grade) sodium carbonate" below 800° C. do not specify the ma-
was stored over Drierite (W. A. Ham- Atmos- terial of the vessel: Waldbauer, Mc-
mond Drierite Co., Yellow Springs, Ohio) Curve Crucible phere Sample Cann, and Tuleen (11) report decom-
except during weighing of the samples. Porcelain Air position at 482° C., and Smith and
Na2C03
Nitrogen from the laboratory supply
1

2 Porcelain Air Na2C03fc Croad (10) report decomposition above

was dried by passage through a trap 3 Porcelain Dry N2C Na2CO* 300° C. but neither mentions the nature
packed with glass wool and cooled in 4 Alumina Air Na2COs of the container. It is likely that they,
liquid nitrogen. It was humidified by 5 Platinum Air Na2C03
and others, were observing a chemical
passage through water in a sintered- 6 Platinum Dry N2e Na2COg
glass bubbler at 25° C. The carbon 7 Platinum Wet N2c Na2COj reaction with the container rather than
dioxide (Matheson Co., “bone dry” 8 Platinum C02c Na2COs the thermal decomposition of sodium
9 Platinum Dry N2C Na2COj " Si02 carbonate.
grade) was used as received. 10 Platinum Dry N2c Na2C08 + Si02d The authors suggest that sodium
11 Gold Dry N2c Na2C03e
carbonate for analytical use be dried by
EXPERIMENTAL RESULTS “Heating rate 300° C. per hour, except runs
heating in dry air or carbon dioxide
9 and 10.
6
Crucible covered. using platinum or other inert material
The experimental thermobalance Gas flow rate 250 ml. per minute.
c
as a container. Under these conditions,
d
curves are reproduced in Figure 1. Heating rate 300° C. per hour to 520° C, any temperature in the range of 250°
The explanation for the shape of these then 50° C. per hour.
e
Maximum temperature 922° C., but sample
to at least 700° C. should be satisfac-
curves seems straightforward. The cooled and held 1 hour at 915° C. after reach- tory. For work of the highest accuracy
small loss below 100° C. is probably ing 922° C. it may be necessary to just fuse the

VOL. 30, NO. 5, MAY 1 958 · 983

material in an atmosphere of carbon (3) Easterbrook, W. C., Analyst 82, 383 (8) Richards, T. W., Hoover, C. R.,
dioxide as proposed by Richards and (1957). J. Am. Chem. Soc. 37, 95 (1915).
Hoover (8). (4) Howarth, J. T., Maskill, W., Turner, (9) Simons, E. W., Newkirk, A. E.,
W. E. S., J. Soc. Glass Technol. Aliferis, Ifigenia, Anal. Chem.
17, 25T (1933). 29,48(1957).
LITERATURE CITED
(5) Howarth, J. T., Turner, W. E. S., (10) Smith, G. F., Croad, G. F., Ind.
Ibid., 14, 402T (1930). Eng. Chem., Anal. Ed. 9, 141
(6) Kolthoff, I. M., Stenger, V. A., (1937).
(1) Balis, E. W., Bronk, L. B., Lieb- “Volumetric Anatysis,” Vol. II, (11) Waldbauer, L., McCann, D. C.,
hafsky, . A., Pfeiffer, H. G., p. 80, Interscience, New York, Tuleen, L. F., Ibid., 6, 336 (1934).
Anal. Chem. 27, 1173 (1955). 1947.
(2) Duval, C., Anal. Chim. Acta 13, (7) Motzfeldt, K., J. Phys. Chem. 59, Received for review August 9, 1957.
32 (1955). 139 (1955). Accepted December 28, 1957.

Titrimetric Determination of Fluorine

Particularly in Aluminum Fluoride
LAWRENCE V. HAFF
General Chemical Division, Allied Chemical & Dye Corp., Morristown, N. J.

C. P. BUTLER and J. D. BISSO

General Chemical Division, Allied Chemical & Dye Corp., Port Chicago, Calif.

Discordant results for total fluorine minum fluoride. Recoveries were re- in reported assays. It quickly became
were reported from different labo- producible and about 99% complete. apparent that recovery of fluorine was
ratories analyzing identical samples of never complete by either method.
aluminum fluoride. Serious losses of LOSSES IN CONVENTIONAL PROCEDURES While the relative percentage of fluorine
fluorine occur in distilling and titrating lost would not be important in micro-
large amounts of fluorine by conven- For a number of years two procedures analysis, it assumed serious proportions
tional methods. Recoveries are es- for determining relatively large amounts in assay of a material containing some
pecially low and erratic when boro- of fluorides have been used throughout 50% of fluorine. Because some of
silicate glassware is employed. De- this company. As applied to aluminum these errors will affect almost all
tails of a rapid and accurate pro- fluoride, both involve fusion of the macrodeterminations of fluorine, some
cedure for pyrohydrolytic assay sample with sodium carbonate, followed details of this work are presented.
of aluminum fluoride were worked by acidification and steam distillation Standard Samples. No standard
out. Routine analyses of aluminum as described by Willard and Winter fluoride sample was available to the
fluoride are greatly expedited by (22). The analysis may be concluded authors at the time this investigation
the procedure, which, however, re- by Armstrong’s procedure (1, 2), as was initiated. Cryolite has been sug-
quires a standard sample of alumi- modified by Rowley and Churchill gested as a standard (7), but its purity
num fluoride. In assaying the stand- (20). Practically, this titration is is problematical. Sodium fluoride, used
ard sample, appropriate correction limited to determination of small by Hoskins and Ferris (8), Kimball
must be made for the losses incurred amounts of fluorine, and the small ali- and Tufts (9), Reynolds and Hill
in the distillation and subsequent con- quot which must be titrated involves (17), and Matuszak and Brown (12),
centration of the fluorine. The num- serious magnification of small errors. is of uncertain assay. Kimball and
ber of replicate samples required to Alternatively the fluorine in the dis- Tufts (9) reported fluorspar unsatis-
establish the correction and the mag- tillate may be converted quantitatively factory as a standard and suggested
nitude of the correction can be greatly to fluosilicate, precipitated as potassium analyzed lead chlorofluoride. The pres-
reduced by avoiding use of borosili- fluosilicate, filtered off, and titrated. ence of large amounts of either calcium
cate apparatus. This procedure is essentially a macro- or lead in samples intended to simulate
method and all of the fluoride distilled sodium carbonate fusions of aluminum
from a 0.2-gram sample may be titrated fluoride wTas considered undesirable.

A study of conventional volumetric

methods for determining fluorine
in aluminum fluorine indicates that
conveniently. Unlike the preceding
titration, this procedure is insensitive to
the interferences of small amounts of
In this work, the exact weight of
fluorine involved in each analysis was
determined by titrating about 0.2 gram
recoveries of large amounts of fluorine phosphates and sulfates and the end of 48% hydrofluoric acid to the phenol-
from the distillation procedure are point is familiar to most analysts. Ac- phthalein end point with standard O.lzV
only about 98% complete and further cordingly it was designated the standard sodium hydroxide solution. The titra-
losses are incurred when glassware con- procedure for the analysis of aluminum tions were performed in platinum dishes
taining boron is used. Under stand- fluoride. As rather lengthy evapora- and concluded at the boiling point to
ardized conditions recoveries are repro- tion and filtration procedures are in- eliminate interference by fluosilicic acid
in the hydrofluoric acid or dissolved
ducible. volved, manufacturing locations were silica in the standard base.
Analysis by a pyrohydrolytic method allowed the option of using the thorium The titrated solution, after cooling, is
of samples standardized by conven- nitrate titration on all but critical transferred to the still, acidified, and
tional methods established that pyro- samples. distilled. The distillate is then analyzed
hydrolysis offers a rapid, accurate, and An investigation of the procedures was by one or both of the above methods to
convenient procedure for assaying alu- initiated when discrepancies appeared determine the weight of fluorine re-

984 · ANALYTICAL CHEMISTRY