A N A L Y T I C A L CHEMISTRY
1192 be helpful in correlating and integrating these facts into a coherent body of useful information.
C. F. Gray for the photomicrograph of the unground catalyst, m d G. H. Walden for many helpful suggestions.
ACKNOWLEDGMENT
LITER4TURE CITED
The writer takes pleasure in expressing his appreciation to
R. N. Watts for supplying the ammonia synthesis catalyst in various particle size ranges, F. M. Long for chemical analyses 011 these samples, J. S.McIlhenny for particle size determinations,
(1) Michei, -1.. An!!. c h i ~ i . 8, , 3li (1937). iZ) ~ ~ lP, l L-, ~ p.~ Bur, , )1ines, T ~Paper ~ 490 ~ (1931), , (3) Wyckoff. R. TI-. G., and ('littenden, E. D., J . A m . Chew,. SOC.,47. 286 (1925).
s,,
R E ~ ~ I V E ,November ,
22, 1948
Determination of Water in Hydrogen Chloride By Means
of Karl Fischer Reagent
E K V E S l C. \IILBEKGEK', KARL UHKIG, I I 4 R R Y C. BECKER, The Texas Company, Beacon, S. Y .
4x1
H$KRY LEVIN
A rapid titrimetric procedure employing the Karl Fischer reagent has been developed for the determination of small amounts of water in gaseous hydrogen chloride. The standard deviation in the results obtained with this method is 0.5 mg. of water for samples of known composition. Hydrocarbon gases in the Cq range, which may be present in the hydrogen chloride of modern petroleum refinery operations, do not interfere; mercaptans (thiols) and hydrogen sulfide, which are oxidized by the reagent, constitute a source of error, but a correction may be applied if the amount of each is known.
T
HE utility of the Fischer reagent, originally used for the de-
termination of water in sulfur dioxide-hydrocarbon mixtures (S), has since been extended to the determination of water in various substances and of water liberated in certain reactions. Such applications include the determination of n ater in practically all kinds of materials. I t has also been used successfully in determining water in liquid sulfur dioxide (Z), alkylation sulfuric acid (41, and hydrogen fluoride ( 7 ) . Inasmuch as hydrogen chloride is acidic and gaseous in nature, a pyridine-methanol solvent is employed in the collection of the sample; pyridine serves to bind the gas and methanol to dissolve the resulting pyridinium hydrochloride. Sufficient pyridine must be present to bind all the hydrogen chloride introduced, for in the presence of uncombined hydrogen chloride the action of Fischer reagent is unreliable. In his original work, Fischer titrated the samples directly with the reagent and used as the end point the first permanent appearance of a brown iodine color. Although this direct titration is satisfactory, the end point is rather difficult to detect, so several other titration procedures have been tried. Back-titration of an excess of reagent with a water-in-methanol solution has been employed by several investigators who used potentiometric (1), dead-stop ( 8 ) ,and cathode ray magic-eye ( 5 )devices to detect the end point. However, excellent results 'can be obtained by a simple back-titration to an end point detected visually as a color change from red-brown through gold-brown to a final brassy yellon-. REAGENTS
Water in Methanol. Because this solutio11 is not standardized, it is prepared simply by adding approximately 5 nil. of water to 2 liters of C.P. methanol. Sample Solvent. The solvent consists of 4 parts by volume of dry pyridine and 3.5 parts of redistilled methanol. Fischer Reagent. Gaseous sulfur dioxide, led through a drying tube containing Drierite, is passed into 226 grams of pyridine in a small bottle until 55 grams have been added. The pyridine I
Present addres-, Standard Oil Cotnpany of Ohio, Cleveland, Ohio.
should previously be dried over potassium hydroxide pellets, decanted, and distilled. The addition is interrupted several t,imes to cool the contents of the bottle in an ice bath. dfter the addition of sulfur dioxide is complete, the resulting solution is poured into a solution of 72.5 grams of resublimed iodine in 2 liters of freshly redistilled methanol. The reagent is allowed to stand a t least 24 hours before use, and then should have an initial strength of approximately 1.5 nig. of water per ml. of reagent. Standardization of Reagent. .Ipproximately 25 ml. of solvent are run into a flask (dried in an oven and cooled in a desiccator), and then a small excess of Fischer reagent is added to react with the water in the solvent. This excess of reagent is titrated with the water-in-methanol solution, and then a measured volume of Fischer reagent (10 to 25 ml.) is added to the dry solvent and tiack-titrated with the water-in-methanol solution. The yellow end point is somewhat sluggish and must be approached slowly. The color changes gradually from a red-brown to a gold-brown and then, upon the addition of 0.2 to 0.3 ml. of water-in-methanol reagent, t,o a brassv yellow color which is the end point. This color can be roughly approximated by a solution containing 70 mg. of potassium dichromate in 250 ml. of distilled water. Because moist air must he excluded from the flask a t all times, the tip of the buret should be fitted with a rubber stopper or groundglass joint to close the flask during the titration. A capillary tube of small diameter in the stopper will serve as a vent and still prevent diffusion of moisture into the flask. The solution may be stirred conveniently by means of a magnetic stirrer, and in this case the glass-encased agitator should be dried with the flask in an oven. The value of R, which is the volume ratio between the Fischer reagent and the water-in-methanol solution, is then calculated. The strength of the Fischer reagent (factor F ) is determined by adding to the resulting anhydrous solution in the flask approximately 20 nig. of water weighed accurately from a weighing bottle. -in excess of 5 to 10 ml. of Fischer reagent is added to the flask and back-titrated with the water-in-methanol solution to the same yellow end point previous1 observed. The actual volume of Fischer reagent consumed by t i e water is calculated by utilizing ratio R. The value of F is obtained by dividing the weight of water taken (mg.) by the volume of reagent (ml.) consumed. hs the reagent gradually decreases in strength, even when protected from atmospheric moisture, it should be standardized every day. PROCEDURE
A 75-1nl. portion of solvent in a carefully dried flask is made anhydrous by the addition of an excess of Fischer reagent and
lected in as many minutes. When the flow of gas is interrupted the ground-glass connection joining the delivery tube to S must he quickly disconnected to prevent the liquid in the flask from backing up. The flask and delivery t’ube, with the ends stoppered, are reweighed to determine the amount of sample collected. The glass delivery tube is removed and the Hask is immediately closed with a ground-glass stopper. .\fter the flask is shaken until the gas fumes are dissolved, the excess Fischer reagent is back-titrated with water-in-methanol solution. The same apparatus is used as for the initial hack-titration of the excess Fischer reagent in the sample solvent, and the same end point is detected. This decreases the possibility of errors i n judgment on the true end point.
W
I -1
8
9 0
5 3 a
eI
1,2,3, a 4 -NEOPRENE
NEEDLE VALVE
x,x’ ’ GATE VALVE
CALCULATIONS
the per cent water found
=
(T-
- U X R)Fand ( V - U X R)F
wt. of sample (grams) X 10
‘rahle I. Blank Determinations on Hydrogen Chloride and Hydrogen Chloride Plus Hydrocarbon Gas Hydrogen Chloride Sample H10 found Grsms MQ.
.I\-. ”
Hydrogen Chloride Plus Hydrocarbon Gasa Sample H20 found Grams NO.
0 .8
BLEEDER LINE
w-
SCREW CLAMP
-
Ileterniination of water in sample. Let V = volume (ml.) of Fischer reagent added to anhydrous solvent prior to collection of sample U = volume (ml.) of water-in-methanol solution used in backtitration of unconsumed reagent after sample has been collected R = volume ratio of Fischer reagent to the water-in-methanol solution F = factor of the Fischer reagent in mg. of water per ml. of reagent =
S i
2
~r
c
reage;t, which cannot be discharged even upon titrating with a large excess of water-in-methanol solution. The volume of solvent employed is sufficient to bind 18 grams of hydrogen chloride.
Then the milligrams of water found
‘ I ’
CONNECTION
0.9
0 . 5 t o 2.0 grains of C4 hydrocarbon gas added to each sample.
EXPERIMENTAL RESGLTS
Hydrogen chloride taken from commercial cylinders of compressed gas was not satisfactory for preparing standards because of the variation in water content of the gas in different cylinders.
The hydrogen chloride could be dried satisfactorily by passing it through two scrubbers filled with sulfuric acid, but moreconsistent results were obtained on hydrogen chloride generated from ammonium chloride and concentrated sulfuric acid. The data in Table I show an average blank of 0.8 mg. of water, but it seems probable that the gas was anhydrous and that small amounts of water were picked up during the collection of the sample and its titration. This is substantiated by the data, which do not indicate any correlation between sample size and blank. It appears that traces of water weie introduced when hydrocarbon gases were added to the hydrogen chloride. The data in Tables I, 11, and I11 were obtained on the generated hydrogen chloride. The results in Table I1 are for the analysis of samples containing known added amounts of water. This was accomplished by passing the gas through a “humidifiei” containing a weighed amount of water which was vaporized with the aid of a microburner. For these analyses, the average deviation is 0.3 mg. and the standard deviation is 0.5 mg. The presence of hydrocarbon gases in the C4 range [n-hutane,
Table 111. knalysis of Samples of Hydrogen Chloride Containing Added Water and Hydrocarbon Gas K a t e r Added
Water Found
Error
MQ.
M Q .a
.Mg.
7.6 16.2 26.0 31.1 38.4 66.1
7.8
16.2 26.1 30.9 38.5 65.5 .%\.erayedeviation Standard deviation
0 Corrected for blank of 0.9 m g . grams of C I hydrocarbon.
0.2 0.0 0.1 0.2 0.1 0.6 0.2 0.2
Each sample contained from 0.5 to 2 . 0
1194
ANALYTICAL CHEMISTRY
isobutane (2-methylpropane), isobutene (2-methylpropene), 2butene, and butadiene] was found to cause no difficulty. This is indicated by the data in Table 111, which show an average deviation of 0.2 mg. and a standard deviation of 0.2 mg. The solvent action of the solution a t the low temperature ( - 15' to -20" C.) prevented the escape of the hydrocarbon gases. If lighter hydrocarbons were present, provision would have to be made to meter the exhaust gas for inclusion in the sample weight. The time required to conduct the complete determination is 30 to 40 minutes. This time does not include preliminary standardization of the reagent or drying of flasks and delivery tube. A note of caution should be added on two points. During the time that the sample is being collected, the flask must be kept cold because high results have been obtained occasionally when the solution became warm. This may have been the result of reaction between the hydrogen chloride and methanol (6). Because hydrogen chloride is such a hygroscopic material, every precaution should be used to exclude atmospheric moisture during the analysis. Although the samples of gas from petroleum refinery operations, for which this method was intended, will, in all probability, contain nothing but hydrogen chloride, small amounts of hydrocarbon gases, and minute amounts of water, a limited investigation of the effect of some sulfur compounds was made. As would be expected, there was no reaction with sulfur dioxide in the absence of water. However, hydrogen sulfide and methyl and ethyl mercaptans (methanethiol and ethanethiol) were oxidized in the absence of moisture. I t is well known that thp following equations H p S
2RSH
+ I p --+ S + 2HI
+ I p +RS - SR + 2HI
apply in the reaction of hydrogen sulfide and mercaptans with aqueous iodine solutions. As the data in Table IV show, the same stoichiometric relationship was observed when the mercaptan and hydrogen sulfide were oxidized with the Fischer reagent. As the exact composition of the Fischer reagent was not known in terms of available iodine, the amount consumed is expressed in
Table IV. Wt. of Sulfur Compound
Relationship of Ethyl Mercaptan and Hydrogen Sulfide with Fischer Reagent
Af ff .
Iodine Consumed %-iff.H20
72.7 629.7 615.0
10.28 84.5 83.1
6.4 6.3 6.3
3.7 3.6 3.1
Sulfur Compound Mole
1320 Mole
Molar Ratio S Compound/HnO
Ethyl Mercaptan 0.0006 0,0012 0.0101 0.0047 0,0099 0.0046 Hydrogen Sulfide 0.00019 0.00020 0.00018 0.00020 0.00018 0.00017
2/1,00
2/0,93 2/0.93 1/1,05 l/l,ll 1/0.95
terms of the equivalent amount of water. On this basis, each mole of mercaptan consumes an amount of Fischer reagent approximately equivalent to 0.5 mole of water, and each mole of hydrogen sulfide is approximately equivalent to 1 mole of water. On this basis it will be possible to apply a correction for the interference of these sulfur compounds when the amounts of hydrogen sulfide and mercaptan have been determined. LITERATURE CITED
Almy, E. G., Griffin, W.C., and Wilcox, C. S., ISD. E m . CHEM., AXAL.ED.,12,392 (1940). DiCaprio, B. R., AN.AL.CHEY.,19, 1010 (1947). Fischer, K., Angew. Chem., 48, 394 (1935).
Goff, W.H., Palmer. W.S.,and Huhndorff, R. F., ANAL.('HEM., 20, 344 (1948) McKinney, C . D., and Hall, R. T..ISD.EXG.CHEM., ANAL.ED., 15, 460 (1943). Mitchell, J . , J r . , and Smith, D. M . , "Aquametry," p. 239, Xew York, Interscience Publishers, 1948. Universal Oil Products Laboratory. "Test Methods for Petroleum a n d Its Products," Method A-166-43, 1913. Wernirnont, G., and Hopkinson, F. J., IXD.ESG. CHEM.,.INAL. ED.,15, 272 (1943). RECEIVEDOctober 2, 1946. Presented before the Division of Analytical and hlicro Chemistry a t the 110th Meeting of the ANERICAK CHEMICAL SOCIETY, Chicago, Ill.
Determination of Unsaturation in Dehydrogenated Dichloroethylbenzene By Use of Mercuric Acetate ROLAND P. MARQUARDT AND E. N. LUCE, The Dow Chemical Company,Midland, Mich.
W
H E X dichlorostyrene was first considered for use in the rubber and plastic industry, no suitable analytical procedure was available for ascertaining the purity of this monomer. Owing to the presence of two chlorine atoms on the benzene ring, the addition of chemical reagents to the unsaturated side rhain proceeas with considerable difficulty. Thus, the usual methods of analysis for unsaturation \yere found to be inadequate. The popular bromate-bromide method (Koppeschaar, 5)and the bromination procedure using bronllne in cdi bon tetrachloride (RlcIlhiney, 4 ) did not give quantitative results. Low results were likewise obtained by the methods of Wijs (f4),using iodine chloride in acetic acid; Hanus (I),using iodine bromide in acetic acid; and Hub1 (21, using mercuric chloride and iodine in methanol.
A procedure previously disclosed by the authors (6) for the determination of the unsaturation in styrene and styrene derivatives by use of an aqueous l,.l-dioxane solution of mercuric acetate, in which the amount of mercury adding to the double bond is determined by direct titration with standard ammonium thiocyanate, gives low results with dichlorostyrene. However, it was found that mercuric acetate in methanol solution would, with moderate warming, add much more easily to carbon-carbon double bonds, and this led to the development of the following method for the quantitative estimation of the unsaturation in dichlorostyrene. It differs from the previous mercuric acetate method in that the acetic acid produced by the addition reaction is titrated instead of the mercury that chemically combines with the styrene derivative.
