V O L U M E 21, NO. 1 2 , D E C E M B E R 1 9 4 9 method for the analysis of such materials is that very small samples (less than 0.1 gram) are required. LITERATURE CITED ( 1 ) Beeghly, H . F., ISD. ENG.CHEM.,AXAL.ED.,14, 137 (1942). (2) Scott, W. W., “Standard Methods of Chemical Analysis,” Vol. I, p. 632, New York, D. Van Nostrand Co., 1939.
1551 (3) Shirley, R. L., and Becker, W. W., IND.ENG.CHEM.,ANAL.ED., 17,473 (1945).
(4) Snell, F. D., and Snell, C. T., “ColorimetricMethods of Analysis,’’ Vol. I, p. 653, New York, D. Van Nostrand Co., 1939. RECEIVED April 1. 1949. Presented before the Divisions of Petroleum and Analytical and Micro Chemistry, Symposium on hlicroohemistry and the Petroleum Industry, a t the 115th Meeting of the AMERICAN CHEMICAL SOCIETY,San Francisco, Calif.
Microdetermination of Chlorine and Bromine JOSEPH GRODSKY, Ortho Research Foundation, Raritan, N. J . A method is described for determining halogen in organic compounds by fusing with potassium and titrating the r h t i n g potassium halide with silver nitrate, using dichlorofluorescein as an indicator. A n attempt was made, with limited success, to determine halide with solid silver iodate by Sendroy’s method.
0
F T H E many reported methods for the determination of halo-
gen in organic compounds, few are suitable for use with very volatile liquids. The Carius method is the simplest and most suitable, but is tedious, as three vieighings of the filter stick are necessary for the first analysis and two wighings are necessary for each succeeding analysis that day (6). Therefore, a volumetric method for determining halogens which could be used for all types of compounds, including volatile liquids, would be of distinct value. The most promising of the methods reported in the literature for decomposing the organic halogen compound was judged to be fusion with an alkali metal in a sealed and evacuated tube. On a macro scale, using this method of decomposition, excellent gravinietric results were obtained by Elving and Ligett in 1942 for all halogens, including fluorine compounds (4). They announced then their intention of extending their work to a semimicro and micro scale, but have not yet done so. .4t about the same time, Burger ( 2 ) used a similar fusion in a nitrogen atmosphere to decompose the organic compound, determining chlorine and bromine by the Volhard method. His prime interest, however, was in the determination of sulfur, and his description of the method for halogen determination is somewhat vague and perfunctory. His procedure for the preparation of the potassium and decomposition of samples also seemed unduly complicated. In view of this, the author undertook to develop a more satisfactory microvolumetric procedure. A review of methods for estimating the resulting halide ion indicated that a trial might profitably be made of the reaction between the ion and solid silver iodate, whereby the silver halide is precipitated and the filtrate analyzed for potassium iodate (5,7 ) . This method was investigated; although known solutions of potassium chloride and potassium bromide gave good results, organic compounds did not. In view of recent favorable reports on the use of dichlorofluoreucein in the titration of halide obtained by decomposition of organic compounds in the Grote combustion apparatus (8), and its reported use in titrating barium chloride in the presence of dispersed barium carbonate ( 1 )--conditions similar t o the author’s-it was decided next to try a direct titration with silver nitrate using that indicator. Early experiments using this method produced low and inconstant results, which were t,raced to the difficulty of washing the reaction tube quantitatively. The use of hot acidified water for washing eliminated that difficulty and good results were then obtained with chloro compounds on a micro scale. However, semimicro samples were necessary to get satisfactory resuks with bromo compounds.
Another attempt was then made to develop the iodate procedure, using hot acidified washes for the reaction tube, and with this modification somewhat better results were obtained. HOSever, although bromo compounds gave acceptahle values, the results with chloro compounds were still not very satisfactory. APPARATUS AND REAGENTS
Reaction tubes were 25-em. lengths of 10- to 11-mm. Pyrex tubing, on which test tube ends were blown. Tubes can be reused twice before they become too short for use. A gas-fired Carius furnace was used and an additional Bunsen burner was found necessary to attain the proper temperature. The funnel was of the Hirsch type with a fritted 20-mm. disk M porosity. The disk surface was protected by a Whatman,So. 540 2.1-em. filter paper disk. After about six to eight filtrations, the paper was lifted out and the funnel washed by reverse suction with about 200 ml. of water, after which it was again ready for use. When necessary, it could be cleaned in warm dichromate cleaning solution overnight. Phosphoric acid was made by diluting 100 ml. of 85% reagent to 1liter. Dichlorofluorescein (8), 10 mg. dissolved in 100 nil. of 6070 ethanol to which 2.5 ml. of 0.01 N sodium hydroxide were added. Silver nitrate, 0.01 iL‘ mas used as primary standard. Potassium chloride, 0.01 N , was standardized against silver nitrate. Sodium thiosulfate, 0.04 N , was standardized against potassium dichromate. All other reagents were best commercial grade. EXPERIMENTAL
Preliminary Experiments on Determination of Inorganic Halide by Silver Iodate Method. Destpoy several slivers of potassium in a 50-ml. beaker with ethanol and ad a measured volume of 0.01 N potassium chloride or potassium bromide to each solution. Acidify the resulting solution with phosphoric acid to an alizarin red end point and add one drop in excess. Then dilute to about 25 ml. and concentrate on the hot plate to the proper volume. Add a spatula tip (about 50 nig.) of silver iodate,, heat tke mixture with occasional stirring for about 2 minutes, then allow it to cool, filter, and titrate as below. Procedure for Decomposition of Organic Compounds. Keigh the liquid sample (4 to 8 mg. for micro and 10 to 20 mg. for semimicro) in a weighing capillary, the handle of \yhich is a glass rod 20 to 25 mm. long, and place the capillary end down in the reaction tube. 1J7eighthe solid sample with a long-handled weighing stick and place as far down in the reaction tube as possible. Cut under ether about three shavings of potassium each about 1 CU. mm., dry rapidly with a clean tissue, and add to the microsample. Use two to five times as much for semimicrosamples or in the presence of nitrogen or sulfur. Heat with rotation a t a point about 3 em. below the open end of the tube and allow glass to gather there, then draw to a thick-walled capillary. When cool, connect the open end to a water aspirator protected by a safety
ANALYTICAL CHEMISTRY
1552 trap and when the tube is evacuated, seal off a t the capillary. Shake the tube so as t o break the inner capillary containing the Liquid sample, and place in a furnace preheated to 400” C., allowing it to remain for 15 minutes. S o precautions are necessary in handling these tubes and they can safely be removed from the hot furnace. When the tube is cool, wash the outside and open it by cutting just below the shoulder; add 1 ml. of ethanol to destroy excess potassium and wash the sides down with hot water. Form a lip on the tube by heating the open end and pressing with forceps. Procedure for Argentimetric Titration. Transfer the solution through the sintered-glass funnel into a 125-m1. Erlenmeyer Bask containing 5 ml. of 0.01 N potassium chloride. Wash‘the reaction tube four times with hot water and add a drop of 1 to 3 nitric acid to each washing. Add several nlundum boiling stones to the solution in the Erlenmeyer and bring to a boil on an electric hot plate a t low heat. Add about 5 mg. of barium carbonate to the hot solution, and acidify if it does not prove acid. Continue boiling for about 10 minutes. This removes the hydrocyanic acid formed in the presence of nitrogen. In the presence of sulfur, boil a t least 2 hours with replacement of evaporated water to assure quantitative removal of the hydrogen sulfide, then carefully add exccss barium carbonate until an undissolved residue of about 50 mg. remains. Add 1ml. of dichlorofluorescein indicator and 10 ml. of acetone to the cooled solution and titrate with 0.01 .V silver nitrate, using side lighting and dimming overhead lights to increase sharpness of the end point ( 3 ) . Procedure for Silver Iodate Method and Iodometric Titration. Transfer the solution in the reaction tube to a 50-ml. beaker containing a drop of alizarin red indicator and a small stirring rod. Wash the reaction tube four times with hot water, adding a drop of phosphoric acid solution to each washing unless the indicator in the beaker turns ycllow, in which case use only one more drop. Keutralize the solution in the beaker to a yellow color with phosphoric acid and add one drop in excess. Concentrate on a hot plate to the proper volume (1 mg. of chlorine or 2.4 mg. of bromine should be in about 10 ml. of solution). Add about 50 mg. of solid silver iodate and allow the hot solution to remain on the hot plate for 2 minutes, with occasional stirring. Allow to cool and filter through a sintered-glass funnel into a 125-ml. glassstoppered Erlenmeyer. Wash the beaker five times with small portions of 75% ethanol. Dilute the solution in the Erlenmeyer with distilled water to about 75 ml. Add 0.5 gram of solid potassium iodide and 5 ml. of 6 N sulfuric acid. Allow to stand for 2 minutes and titrate with 0.04 N sodium thiosulfate to a starch end point.
Table 111. Precision and Accuracy of Argentimetdc 3licrodetermination of Chloro Compounds No. of Detns.
Compound Chlorobenzene p-Chlorophenoxyacetic acid p-Chlorothymol 2,4-Dinitrochlorobenzene Benzyl isothiourea hydrochloride
Recovery of Halide by Silver Iodate Procedure
Halide Chloride
Bromide
Taken
Recovered
Deviation
Mg.
.lip.
0.8865 1.084 1.241 1.418 1.596 1.773 1.950
0.8910 1.070 1.243 1.408 1.584 1,783 1.941
Parts/1000 +5 +6 +2 -7 -8
1.998 2.398 2.797 3.197 3.596 3.996
2.016 2.408 2.794 3,207 3.593 3.990
-6
-5 f 9
+4 -1
+3 -1
-2
Table 11. Precision and Accuracy of Microdetermination of Bromo Compounds by Iodate Method Compound o-Bromobenaoic acid Benzalacetophenone dibromide
No. of Detns.
Theory %
Found, Av. %
4
39.76
39.72
Preoiaion % *O 16
4
43.42
43.39
==0.18
Found, Av. %
Precisioo
4
31.50
31.44
tO.06
7 I
19.00 19.20
19.00 19.25
*O 12 r0.13
5
17.60
17.52
*0.09
3
17.49
17.55
r0.11
%
Table IV. Precision and Accuracy of Argentimetric Semimicrodetermination of Bromo Compounds No. of Detna.
Compound o-Bromobenzoic acid Benzalscetophenone dibromide p-Nitrobromobenzene Bromobenzene
Found, Av. %
Precision
% 4
39.76
39.81
r0.10
4 3 3
43.42 39.56 50.90
43.53 39.47 50.74
r0.13 *O.W r0.16
Theory
%
The argentimetric titration worked well on a micro scale with compounds containing chlorine alone or in combination with dtrogen and sulfur (Table 111), but it did not work with bromo compounds. With these compounds, however, good results were obtained on a semimicro scale (10- to 20-mg. samples) and using 0.015 N silver nitrate (Table IV). CalCUlatiOnS.
% halide
=
.&RGENTIMETRIC
ml.
x
normality X equivalent weight X 100 weight of sample (mg.)
IODATE %bromide =
Table I.
Theory %
ml. X normality X 79.9 X 100 6 X weight of sample (mg.)
Equivalent weight for chlorine = 35.5 Equivalent weight for bromine = 79.9 CONCLUSION
Decomposition of halo-organic compounds with potassium in a sealed, evacuated tube on a micro scale was found practical. The chloride formed could be determined, after proper treatment, by titrating with silver nitrate using dichlorofluorescein indicator. The bromide formed could be determined on a semimicro scale by a similar argentimetric titration or, if no other acid-forming elements were present, on a micro scale by treatment to convert the potassium halide to potassium iodate and determining as usual. These methods have been used in the laboratory of the OrthoResearch Foundation for the past few months and have proved convenient and reliable. ACKNOWLEDGMENT
RESULTS
Table I shows the results obtained by the iodate procedure on Jolutions of potassium chloride and potassium bromide of known concentration. When the same procedure was tried with decomposition products, the results for chlorine were not satisfactory but bromo compounds gave good results on samples not containing nitrogen (Table 11). S o sulfur-containing compounds were tried.
The author wishes to thank John Cozzolino for his work and suggestions in the preparation of this paper. LITERATURE CITED
(1) Belcher, R., and Godbert, A. L., “Semimicro Quantitative Organic Analysis,” p. 107, New York, Longmans, Green and Co., 1945. (2) Btlrger, K., Angew. Chem., 54, 479-81 (1941); Die Cfiemie, 55 245-7 (1942).
V O L U M E 2 1 , NO. 1 2 , D E C E M B E R 1 9 4 9 43) Bullock, B., and Kirk, P. L., IND.ENG.CHEM.,A X A L E . D . , 7, 178 (1935).
Elving, P. J., and Ligett, W. B., Zbid., 14, 449 (1942). Grangaud, R., Bull. 8oc. chim., 10,236-8 (1943). 6 ) Niederl, J. B., and Niederl, V., “Organic Quantitative Microanalysis,” 2nd ed., pp. 157-9, New York, John Wilw 8~ Sons.
4) 5)
1942.
1553 (7) Sendroy, J., Jr., J . Biol. Chem., 120, 335 (1937). E N G .CHEM.,ANALED., (8) Sundberg, 0. E . , and Royer, G. L., IND. IS, 719-23 (1946). RECEIVED April 1 , 1949. Presented before the Division of Petroleum Chemistry, Symposium on Microchemistry in the Petroleum Industry. at the 115th Meeting of the AMERICAN CHEMICAL S O C I ~ T YSsn , Franaisco Calif.
Determination of Small Amounts of Water F. &I. ROBERTS AND HARRY LEVIN Beacon Laboratories, The Texas Company, Beacon, N . Y . A m e t h o d , employing azeotropic distillation a n d subsequent d e t e r m i n a t i o n of the water in the distillate b y titration w i t h K a r l Fischer reagent, has been applied to samples of oils, greases, deposits, additive concentrates, etc., containing as little as 0.0002570 of water. The d e t e r m i n a t i o n is m a d e in a special distillation system protected f r o m atmospheric moisture, a f t e r the system has been dried by partial distillation of the azeotrope f o r m e r used in the analysis itself. The m e t h o d is accurate t o 0.3 mg. of water.
THE
reported chemical methods employing the Karl Fischer reagent for determining very small amounts of water in petroleum fractions or other liquids involve a direct determination of the water in the sample with the Fischer reagent. Aepli and MeCarter (I) detcrmined water in liquid petroleum fractions by titrating the sample directly in a large flask protected from atmospheric moisture. The method has disadvantages in that a large amount of sample is required, which makes direct titrations unwieldy, and the mixture in the titrating flask exists in two phases, necessitating the extraction of the water with Fischer reagent. Snyder and Clark (IO) also applied the direct titration procedure to petroleum fractions, using a solvent to effect a single phase in the titration vessel. This method also mvolves titration of large volumes of liquid, and it is necessary to prepare and keep uantities of dry solvent. Weaver and Riley 71,) have developed a method for determining water in gases by measuring the change in electrical conductance of a hygroscopic film. They have done a small amount of work on liquid samples; however, the amounts of water reported are not very low. Evans and Davenport (5) have described a manometric procedure for small amounts of water in insulating oils. Their method is limited to oils, and would not be applicable to greases or semisolid materials such as engine deposits. Benning, Ebert, and Irwin (4) applied infrared spectroscopy for water in Freons, using the strong absorption band of water a t 2.67 microns, but state that compounds containing hydrogen will also absorb a t this wave length. This rules out the application of the method to a wide range of organic liquids. Obviously, the A.S.T.M. distillation method ( 3 ) is not suitable for very small amounts of water, for direct visual volume measurement of the water is required. If water contents are very low, it is possible that the entire amount contained in the sample will remain dissolved in the distillate. During the course of this investigation, a paper by Suter (11) described a method for water in inorganic alkaline materials. Water was separated from the sample by distillation with xylene and determined in the distillate with Karl Fischer reagent. This procedure, although suitable for high water contents, does not allow the precision required for very small amounts of water; the blank determinations are greater than the total water often determined by the present method. I t was found in the present investigation that ground-glass joints, as found in Suter’s apparatus, could not be tolerated where small amounts of water were being determined. Opening th8 system to atmospheric moisture, as is done in the reported method, cannot be tolerated wherp amounts of water are low. The present method employs azeotropic distillation and determination of the water in the distillate with Karl Fischer reagent. The necessity for keeping a dry solvent, or correcting for the blank on the solvent, is eliminated. The determination LB made in a system completely protected from atmospheric water. The use of large samples is necessary; however, the
final titration is more convenient because the water is con. centrated into a comparatively small volume of liquid. Even in the case of dark samples the distillate is water-white; it is therefore possible to titrate the water without the aid of potentiometric devices for determining the end point. Snyder and Clark (IO)have reported a number of substances that interfere with the Karl Fischer reagent. These substance8 will interfere in the present method only if they are volatile under the conditions of the test. This method was developed primarily for new and used p e t r e leum products including greases, although by the choice of a proper azeotrope former, it may be applied to other substances The method has been applied over the past year to other organic liquids-for example, additive concentrates, and trichloroethyl. ene-and to engine deposits and sludges. REAGENTS
Solvent or Azeotrope Former. In the present work, benzene wab used for all samples except greases. Pyridine waa used for greases Karl Fischer Reagent. The reagent may be prepared (9, 19) or purchased. It should be equivalent to 1.5 to 2 mg. of water pel milliliter of reagent. Water-in-Methanol Solution. This solution should contain 1.5 to 2 mg. of water per milliliter. Because commercial anhydrow methanol sometimes contains as much as 1 mg. of water pei milliliter, its water content should be estimated and adjusted to thr desired concentration. APPARATUS
The apparatus, shown assembled in Figure 1, consists of a dis. tilling flask, a receiver and titration flask, and two 25ml. automatic burets, the tips of which pass through 18/9 ball joints whicb fit on the titration flask. One automatic buret is for Karl Fischei reagent, the other for water-in-methanol. A magnetic stirrer (Arthur H. Thomas Co.), a heatin ‘mantle, and a Variac are also required, as well as an instrument for determining the end point of the titration. The dead-sto method described by Foulk and Bawden (6)is satisfactory, and tge Fischer Titrimeter or any p H meter may be used. This part of the apparatus is not absolutely necessary, as the end point may be determined visually. PROCEXWRE
Standardization of Reagents. The apparatus, except the distilling flask, is assembled aa shown in Figure 1 and the reagents are standardized according to the procedure of Almy, Griffin, and Wilcox (9). It is more convenient t o use this method than a distillation. Table I shows that factors obtained by both procedure are essentially the same.