V O L U M E 21, NO. 9, S E P T E M B E R 1 9 4 9
1073
Table 11. Analysis of Deuterium in Water Gas
Sample 1
L
% N1
Present 32.7
?& Air
Present 3.3
1.1
0.6
6.9
0.2
I
20.3
P
3.7
1
22.6
2
12.8
5 Air presumably collected and before
Experimental, hloie % D (as CHPD) 1.37 1 .37 1 .27 1.26 2.73 2.72
Experimental Average, Mole % D 1.32
Calculated, Mole % D (as Dz0) 1 42
2.70
2.83
26.3"
4.71 4.75 4.84 4.70 0.3 4.80 4.78 26.ga 14.54 14.63 14.79 14.62 25.8" 14.60 14.68 leaked into gas sampling bulb aftrr sample had been analysis was made.
ages of 55, and pattern values were corrected for C13 content (1.1%). The pattern values compare favorably; the fragmentation values from the present work are slightly higher, possibly because of a higher ionization chamber temperature. The sensitivity difference between methane and methyl deuteride for thi8 work (methyl deuteride lower than methane by 1.5%) comparea more favorably with that of Evans et al. (sensitivities equal within 1%) than with those of Turkevich et al. (methyl deuteride higher than methalie by 5%). A series of water standards of known isotopic composition was prepared by weight dilution methods, using normal distilled water and 99.87% deuterium oxide as the ingredients. Using the same apparatus and technique as in the preparation of methyl deuteride, two successive gas samples were prepared from each water standard. Two analyses of each sample were made, using the mass spectrometer. Table I1 shows that air or nitrogen or both are present in large amounts in four of the eight samples. From the good results obtained, it is apparent that no special precautions need to be taken to prevent this contamination. LITERATURE CITED
was taken. Without removing the syringe from the neoprene serum stopper, 50 to 100 mg. of water were injected and a second gas sample was taken. Additional samples of gas were taken without removing the syringe. The mass spectral patterns of nine samples of methyl deuteride, prepared in three batches from two different vials of deuterium oxide, did not vary by more than 0.2 for the 16 mass peak and not more than 0.1 for the other mass peaks.
(1) Evans, Yl.,Bauer,, N.,and
Comparisons of patterns and sensitivities for methane arid methyl deuteride with those of Evans et al. ( 1 ) and of Turkevich et al. (6) are given in Table I. All three sets of data were obtained o n Consolidated mass spectrometers at electron-accelerating volt-
(6) Turkevich, J., Friedman, L., Solomon, E., and Wrightson, F. M.: J . Am. Chem. Soc., 70, 2638 (1948). ( 7 ) Zerewitinoff, Ber., 40, 2023 (1907).
(1946). -,
Beach, J. Y . , J . Chem. Phgs., 14, 701
\--
(2)
Feldman, J., Myles. M., Wender, I., and Orchin, M.,I d . Eng.
Chem., 41, 1032 (1949). (3) Fisher, R. B., Potter, R. A,, and Voskuyl, R. J., ANAL.CHEM.. 20, 571 (1948). (4) Keston, A. S., Rittenberg, D.. Chem., 122, 227 (1937).
and Schoenheimer, R. J., J . Riol.
( 5 ) Orchin, M., and Wender, I., h s a ~CHEM., . 21, 875 (1949).
RECEIVEDJanuary 18. 1949.
Determination of Acrylonitrile and Alpha, Beta-Unsaturated Carbonyl Compounds Using Dodecanethiol DONALD W. BEESING, WILLARD P. TYLER, DONALD M. KURTZ, AND STUART A. HARRISON' The B. F. Goodrich Research Center, Brecksville, Ohio Methods are described for the determination of from 2 to 200 mg. of acrylonitrile and a,@-unsaturatedaldehydes and esters by the addition of excess primary mercaptan and determination of the unreacted mercaptan iodometrically or amperometrically. The salts of the corresponding acids do not react with the mercaptan under the conditions employed and other unsaturated compounds in general do not interfere.
N
0 CHEMICAL method for determining acrylonitrile in ap-
preciable quantities has been described in the literature. A method for determining small amounts in air by hydrolysis to ammonia was developed by Petersen and Radke (11) and a physical method by infrared spectroscopy was described recently by Dinsmore and Smith (9). In studying the disappearance of mercaptan (thiol) in butadiene-acrylonitrile copolymerizations, it was found that primary mercaptans disappeared rapidly in the presence of a base; the 1 Present address, Research Laboratories, General Mills. Inc., Minneapolis. Mion.
primary mercaptan had reacted with the scryloiiitrile in what appeared to be a quantitative manner. At about the same time, the reaction of acrylonitrile as well as other open-chain a,@-unsaturated nitriles with certain organic mercaptans was patented by Harman (4). Since that time, Hurd and Gerahbein (6) have reported that acrylonitrile and alkanethiols or thiophenols react practically quantitatively in the presence of a small amount of alkaline condensing agent. This paper describes the conditions under which acrylonitrile and some a,@-unsaturated esters and aldehydes can be determined accurately by adding an excess of n-dodecyl or other pri-
1074
ANALYTICAL CHEMISTRY
mary mercaptan and titrating the unreacted mercaptan iodometrically or mperometrically with silver nitrate. REAGENTS
Alcoholic Mercaptan Solution. Dissolve 25 grams of crude ndodecyl mercaptan (dodecanethiol) in 1 liter of ethanol (Formula 2-B) or 2-propanol. n-Hexyl mercaptan (hexanethiol) was also tested with satisfactory results and presumably an equivalent quantity of any primary mercaptan may be used, but the commercially available dodecyl mercaptan is a good compromise in solubility and odor. Dilutions of this solution are used for determination of less than 10 mg. of acrylonitrile or other active compound. No secondary or tertiary mercaptans were included in this study. Basic Catalyst. Dissolve 1 gram of potassium hydroxide in 20 ml. of alcohol, or use undiluted Triton B (40y0aqueous solution of trimethylbenzylammonium hydroxide). Standard Iodine Solution. Use 0.05 to 0.1 A' iodine solution in aqueous potassium iodide. Standard Silver Nitrate Solution. Use 0.005 to 0.025 N silver nitrate for amperometric titration. ANALYTICAL PROCEDURES
Procedure for 20 to 200 Mg. of Reactive Compound. Pipet 10 to 50 ml. of alcoholic mercaptan, dependin on the optimum excess of mercaptan to be used (see Table into a 125-ml. iodine flask. Reigh into the flask sufficient sample to contain 20 to 200 mg. of acrylonitrile or other active unsaturated compound. If necessary, neutralize any acidity with base. Add 1 ml. of alcoholic potassium hydroxide or 0.25 ml. (5 drops) of Triton B in excess. Stopper the flask and allow it to stand the optimum time for completion of reaction (see Table I). rlcidify with 1 to 2 ml. of glacial acetic acid, dilute to about 75 ml. with alcohol, and titrate xith iodine solution of appropriate strength to a faint yellox end point permanent for 30 seconds of swirling. If a reaction time of 2 minutes is used, dilute the same volume of mercaptan solution as uTas used for the experimental determination to about 75 ml. Kith alcohol and run a blank by direct titration with iodine. If a longer reaction time is used, run a blank containing the mercaptan and the basic catalyst through the entire procedure and also run a direct blank titration of mercaptan with no base. The difference between these blank values represents the error due to air oxidation of mercaptan in the basic solution. A part of this error proportional to the amount of excess mercaptan present in the experimental run must be subtracted from the direct titration blank. Because the greater part of the active compound reacts with mercaptan within 2 minutes, this method of blank correction is satisfactory. Slternatively, oxygen may be swept from the alcoholic mercaptan with an inert gas before addition of sample or basic catalyst. yoactive compound = ( B - a).\-TVx E x 100
f),
where A = ml. of iodine required for the unknolyn, B = ml. of iodine required for the direct mercaptan blank (corrected if necessary), il' = normality of iodine, E = milliequivalent weight of reactive compound, and W = weight of sample. In the above method, the concentration of reactive compound should not be less than 0.5 mg. per ml. of solution during reaction. Procedure for 2 to 30 Mg. of Reactive Compound. Pipet the proper atnount of mercaptan solution, diluted if necessary, and weigh the sample into a beaker, keeping the ratios such that the concentration of reactive compound does not fall below 0.2 mg. per ml., preferably 0.5 mg. per ml., and the requirements of Table I are met. Add the basic catalyst as in the iodometric method and allow the alkaline mixture to stand the required length of time. Acidify with glacial acetic acid, and add sufficient ammonium hydroxide.to neutralize the acid and to make the solution 0.25 M in ammonia. Dilute to 100 ml. with alcohol, add about 1 ml.,of 0.1 M ammonium nitrate, and titrate amperometrically with 0.005 to 0.025 N silver nitrate, using the general technique described by Kolthoff and Harris (7). Make blank runs on the mercaptan solution under the same conditions as used in the determination, and correct the blank proportionally as in the iodometric procedure if necessary. Calculate as in the iodometric method, substituting silver nitrate volumes and normality for those of iodine, Notes. AS low as 5 mg. of acrylonitrile have been determined by the iodometric method, but with loss in accuracy. The amperometric method should be limited to a maximum of
Table I. Optimum Excess of Mercaptan and Reaction Time Compound Acrylonitrile. acrylates, maleates Methacrylates crotonatee aldehydes Ketones (semihuantitative' estimation)
Optimum Excess
of Mercaptan, % 25-100
100-150
200-400
Optimum Reaction Time, Min 2 10-15 20-30
30 mg. because of errors involving fouling of the platinum electrode with silver mercaptide. Acrylonitrile and acrylates are too volatile to weigh from Hill weighing bot"G1es. Stoppered or screw-cap vials are satisfactory. The amperometric procedure was used in this investigation primarily to avoid interferences peculiar to the iodometric titration, and for the determination of small quantities of reactive compound. The potent,iometric rnet'hod of Tamele and Ryland (16) could probably be used to replace the amperometric method of determining the excess mercaptan, although it was not used in this work. DlSCUSSION AND RESULTS
Because acrylonitrile was studied most extensively, this discussion of method, errors, and results concerns this compound unless otherwise stated. The mercaptan method for determining acrylonitrile is IICCUrate and precise if the following errors are kept to a minimum: the error due to air oxidation of mercaptan, the cyanoet,hylation error due to reaction of acrylonitrile with the alcoholic solvent, and t,he error due to alkaline hydrolysis of the acrylonitrile. The oxidation error can be controlled least readily. High results will be obtained if the reaction time is prolonged without blank correction. This error will increase if too large an excess of mercaptan is used or the solution is too concent,rated. However, the error from this source is negligible with a 2-minute reaction time and may be minimized for longer reaction intervals by making the proportional blank correction. The cyanoethylation of alcohol causes a negative error, but this is of importance only if the basic catalyst is in contact with the solution before the mercaptan is added. The reactmionwith mercaptan is much more rapid than with alcohol, but the latter reaction is favored to some extent by excess basic catalyst and by long reaction times. Hydrolysis of acrylonitrile in the presence of water and base can be a serious source of error if more than 25% water is present' in the reaction mixture, but the error is negligible under normal operating conditions. The optimum amount of catalyst seems to be about 50 mg. of potassium hydroxide or 250 mg. of Triton B. Somewhat smaller quantities may be advantageous in the amperometric procedure where the solution volume is low and where water may be present, Up to five times these amounts will not cause appreciable error in iodometric acrylonitrile determinations, but may cause errors in running other determinations with longer reaction times because of oxidation and hydrolysis. The acrylonitrile-mercaptan reaction does not go to completion, or the accuracy is affected by other factors, if the concentration of acrylonitrile in the reaction mixture is less than about 0.5 mg. per ml. However, the error is under 1% for 0.2 mg. per ml., and this figure is used, as the lower limit for the amperometric method, as the accuracy of this method is of the same magnitude as this error. The acrylonitrile used for this study was freshly distilled research grade material. The initial value found by the iodometric method was 99.7% for one such sample. From the reproducibility shown in Table I1 it is apparent that this material was essentially of this purity and that the maximum error in accuracy of the method is of the order of -0.3%. Upon standing in a s t o p pered transparent bottle a t room temperature, the purity decreased over a period of about one month to 98.5%. The dr-
1075
V O L U M E 21, NO. 9, S E P T E M B E R 1 9 4 9 crease was consistent and outside the errors of reproducibility. S o polymer formation was observed. Other samples of slightly lower initial purity behaved in a similar manner. Alcoholic solutions of acrylonitrile are stable for a t least 10 days, whereas water solutions decrease in concentration rapidly owing to hydrolysis or volatilization. This is of importance in analysis of gas streams. I t has been demonstrated by the authors that methanol is a satisfactory solvent for collecting traces of acrylonitrile from gas streams, whereas water is a very poor solvent because of the high partial pressure of the acrylonitrile above the solution. For collecting larger quantities in a vapor stream, the gases were passed into basic alcoholic mercaptan sohtion to prevent loss of acrylonitrile due to the sweeping out effect of the gases. Table 11. Reproducibility
I
[odometric method Sample weights mg. Average of me& deviations, % Number of results lmperometric method Sample weights mg. 30 Mean deviatio;, % 10.3 Mean error", o/o -0.5 Number of results 15 Based on iodometric method for larger quantities.
20-200
*0.22 126
10 11.0 -0.9 16
INTERFERENCES
OTHER APPLICATIONS O F THE METHOD
In general, a,@-unsaturatedesters and aldehydes react quantitatively in this method, while the salts of the corresponding acids do not react at all under the conditions employed, an important consideration from an analytical standpoint. Ketones react only semiquantitatively. Table I11 gives a summary of results on some other compounds studied. The compounds were purified to some extent and analyses by another method are given in several cases. Inasmuch as the most satisfactory method that the authors have found for determining unsaturation of this type of compound is the pyridine sulfate dibromide method of Rosenmund, Kuhnhenn et al. (IS)as modified by Rowe, Furnas, and Bliss ( I J ) , using mercuric acetate as catalyst, results are given by this method in some cases. Ethyl acrylate gives results comparable to those for methyl m y l a t e and the method has been used for control analysis of both compounds. Cinnamaldehyde and crotonaldehyde, and presumably other a,@-unsaturated aldehydes, differ from the other compounds in their action in that the reaction product with mercaptan is un-
Table 111. Summary of Results Purity b y Mercaptan Method Pyridine ReMean Sulfate Excess action deviaNo. Method, mercap- time, Purity, tion. of % tan. % min. % ' % results
Compound Methyl acrylate, freshly distilled ... 10-25 Ethvl acrvlate. vaciinm di"stilled- - - ~- ~ ~ ... 30 Diethyl maleate EastmanKodak, p i r e 10-164 35-175 Ethyl crotonate, vacuum distilled 95.3a 65-95 Allyl crotonate, vacuum distilled 98.8' 75-260 Llethyl methacrylate, vacuum distilled 97.5' 30-120 Cinnsmaldehyde, vacu u m distilled from KaHCOs ... 130-170 Crotonaldehyde distilled from NAH,COI . 85 Vinyl acetate, distilled ... 170-250 Mesityl oxide. distilled 97.9O 160 400 q-Hydroxybenz ylideneacetone, recrystallized 100.3b 100-400 a With mercuric acetate as catalyst. b Without catalyst. 1
..
___.
2
99.8
10.1
4
2
98.4
to.0
2
2
99.3
t0.1
4
10-30
95.1
t0.2
5
10-25
98.7
t0.3
11
10-30
98.8
10.2
10
2-10
99.4
t0.2
8
10-20 99.4 2-20 75-83 2-30 45-49 25 94.4
to.1
4 9
3-10
26-81
... ... ...
..,
stable in dilute acetic acid, so that low results are obtained, together with a distinct odor in the case of cinnamaldehyde, if the mixture stands for more than about 2 minutes in the acid solution before iodometric titration. Several compounds listed in Table I11 do not give quantitative results. Vinyl acetate probably does not react in the same manner as the other compounds, because it is not an a,&unsaturated carbonyl compound and is known to act as an acetylating agent. The reaction is probably not quantitative, although no attempt was made to prove the purity of the samples examined. The a,+% unsaturated ketones apparently do not react quantitatively, although they may approach complete reaction with a very large excess of mercaptan. No type of unsaturation other than a,@-carbonylunsaturation has been found to react under the conditions herein described. Among the other unsaturated and aromatic conpounds studied, and found not to react, are acetylene, allyl alcohol, allyl acetate. styrene, butadiene, myrcene, vinyl chloride, 2-vinylpyridine, and benzanilide. I n the presence of hydrogen cyanide the solution must, be acidified with hydrochloric acid instead of acetic acid before titration.
4 1 8
In addition to the interferences that would normally be encountered with iodometric or argentometric tit rations, some substances react with mercaptan or catalyze air oxidation of mercaptan. Elemental sulfur appears to react with mercaptan, causing high results. Benzoyl peroxide and presumably other peroxides interfere, because they are capable of reacting both with mercap tan and with iodides. Hydroquinone, and to even a greater extent, quinone, catalyze air oxidation of mercaptan in basic solution, causing high results. Other phenolic polymerization inhibitors have not been studied. NATURE OF REACTION BETWEEN a,B-UNSATURATED CARBONYL COMPOUNDS AND PRIMARY MERCAPTAN
The reaction between a mercaptan and an a#-unsaturated ketone has been considered as a 1,4 addition to a conjugated series of double bonds (8):
0 RCH=CH
e
R"
+R
m e R SR' Axx=
8" R"
SR'
0
II
R~HCH~CR" X nuiriber of compounds of the above type have bpen prepared (8, 9, 12, 15). The reaction between a mercaptan and an a,d-unsaturated ester apparently occurs in the same manner (6,6,9, I O ) . The reaction of acrylonitrile with a primary mercaptan had not been reported, as far as was known, when this study was started. The production of a P-thioalkyl (or aryl) ether of a nitrile had been forecast, however, by the work of Bruson and Riener ( 1 ) on the cyanoethylation of active hydrogen groups and of Gershbein and Hurd (3) on the reaction of hydrogen sulfide with acrylonitrile. Based on the authors' work (see Tables I and 111),the substitution of a methyl group for each hydrogen on the 8-unsaturated carbon of an ester progressively slows up the reaction with mercaptan. This is also true of methyl substitution on the a-unsaturated carbon. Again, although the a,@-unsaturatedesters react readily with a primary mercaptan, the salts of the acids do not react. These effects are predictable by modern electronic considerations. CONCLUSIONS
The conditions for determining acrylonitrile accurately by adding an excess of n-dodecyl (or any primary) mercaptan in the presence of a basic catalyst were found. Optimum conditions are
1076
ANALYTICAL CHEMISTRY
a Zniinute reaction interval, small excess of mercaptan, use of a basic catalyst, and mixture of the mercaptan with acrylonitrile before addition of the base. This method with modification in some cases can be employed for accurately determining a,@-unsaturatedesters and aldehydes. The salts of the corresponding acids and other simple or conjugated unsaturated compounds do not react. LITERATURE CITED
(1) Bruson, H. .4., and Riener, T. W., J . A m . Chem. SOC.,65, 23 (1943). (2) Dinsmore, H.L., and Smith, D. C., ANAL.C H E M 20, . , 19 (1948). (3) Gershbein, L. L.,and Hurd, C. D., J . Am. Chem. SOC..6 9 , 241 (1947). (4) Harman, M. W. (to Monsanto Chemical Co.), U. S. Patent 2,413,917(Jan. 7, 1947). (6) Hurd, C. D., and Gershbein, L. L.,J . A m . Chem. Soc., 69,2328 (1947).
(6) Jacobson, R. A. (to E. I. du Pont de Nemours & Co.), U. 9. Patent 2,199,799(May 7,1940). (7) Kolthoff,I. M., and Harris, W. E., IND.ENQ.CHEM., ANAL.ED., 18, 161 (1946). (8) Nicolet, B. H., J . Am. Chem. Soc., 53,3066 (1931). (9) Zbid., 57,1098 (1935). (10 )Nicolet, B. H., J.Biol. Chem., 95, 389 (1932). (11) Petersen, G.W.,and Radke, H. H., IND, ENQ.CHEM.,ANAL. ED.,16, 63 (1944). (12) Posner, T., Ber., 35,809 (1902). (13) Rosenmund, K.W.,Kuhnhenn, W., Rosenberg-Gruszynski, D., and Rosetti, H., 2. Untersuch. Nahr.-u. Gennussm., 46, 164 (1923). (14) Rowe, R. G., Funas, C. C., and Bliss, H., IND. ENG.CEEM,, A x . 4 ~ ED., . 16, 371 (1944). (15) Ruhemann, S.,J . Chem. SOC.,87,461 (1905). (16) Tamele, M.W., and Ryland, L. B., IND.ENO.CHEM.,ANAL.ED., 8 , 16 (1936). RECEIVEDD r m u b e r 8. 1948.
Reduction of Error in Flame Photometry CHARLES E. BILLS', FRANCIS G . MCDONALD, WILLIAM NIEDER?vIEIER*,AND MELVIN C. SCHWARTZ Research Laboratory, Mead Johnson and Company, Evansville, Znd. In the determination of sodium and potassium with a Perkin-Elmer Model 18 flame photometer, much of the mutual interference of these elements was eliminated by the use of a newly designed amplification circuit. Further reduction in interference, both mutual and from foreign solutes, was effected by working at lower concentration ranges than have previously been recommended. With interferences thus minimized, most biological materials can be analyzed with no more than the instrument error of about 20/,, but when the test element is in low ratio to the total ash, it may be necessary to correct the determinations by the use of an empirically derived correction curve.
S
TUDENTS of flame photometry are agreed that this new analytical art is subject to numerous errors, some of which are understood and can be dealt with, whereas others remain obscure (1, 2, 4-7). Most of the published work has been done with the instrument described by Barnes, Richardson, Berry, and Hood ( I ) or its commercial counterpart, the Model 18 Perkin-Elmer flame photometer (8). This consists essentially of B Meker burner, an atomizer for spraying solutions into the flame, and a photoelectrir colorimeter for measuring the characteristic light emitted. The original work (1) was done with solutions of known composition. I t brought out the importance of constant conditions in the flame and atomizer, and the fact that foreign substances in the solution decrease or increase the light output. The interference of foreign substances or even of too much sodium or potassium ha4 troubled all subsequent workers, and various means have been proposed to overcome it. Thus Berry, Chappell, and Barnes (2) have suggested that, in suitable situations, counterbalancing quantities of the interfering substances be added to the standard solutions with which the photometer is calibrated. Obviously, this device would work only in routine analyses of a particular material. These authors also proposed, tor a corrective of more general applicability, the use of lithium tls an internal standard in flame photometry. They claimed That the internal standard was highly effective against errors caused by fluctuating gas pressure, and partially effective against errors caused by fluctuating air pressure and the presence of toreigri molecules. The present paper is based on nearly three years of expcrience with flame photometry, with emphasis on the reduction of errors. The study was made in connection with a survey of the 1 2
Present address, Route 12, Stringtown Road, Evansville, Ind. Present addreas, Hillman Hospital, Birmingham. Ale.
sodium and potassium content of foods and public water supplies (5).
INSTRUMENTAL IMPROVEMENTS
The Jyork was begun with a Model 18 Perkin-Elmer flame photometer. (Production of the Model 18 instrument was discontinued in 1946 i-n favor of a new model which the manufacturer considers greatly improved.) In this instrument the glass atomizing chamber was held in position around the base of the burner by a coiled spring. The arrangement was such that the chamber was readily displaced, as by touching the intake tube with the beaker holding the test solution. Such displacements occasionally caused large errors in reading, and the defect WBB corrected by clamping the chamber in a reproducible position. Less obvious, but more important, were certain defects in the electrical circuit. The instrument when calibrated for the most desirable range, 0 to 10 p.p.m., was operating a t practically ita limit of sensitivity, and therefore showed considerable drift and instability. Moreover, the amount of interference given by sodium with potassium readings and by potassium with sodium readings was both considerable and inconstant. These s h o r t comings were substantially reduced by employing a new electrical circuit, for which the authors are indebted to R. L. Schoene of the Schoene Electronics Laboratory, Evansville, Ind. The original circuit is shown in Figure 1. With this circuit, amplifier drift occurring between the time of setting the zero and taking the reading appears in the dial reading as an error. To obtain higher sensitivity 4,ithout undue drift a two-stage modified Wynn-Williams bridge circuit was adopted, giving en increase in electrical sensitivity of many times with only about the original amount of drift. To eliminate from the reading any error caused by this drift, a null balance push button was provided. The resulting circuit is shown in Figure 2. In use, the modified instrument is balanced by adjusting the appropriate
