hydrocarbons, accurate results within the confidence interval were obtained without calibrating. These results substantiate the assumption that effects other than change in density are not significant under these conditions. hrthermore, the validity of eliminating calibration by using the gas-density detector was independently substantiated by the analysis of the synthetic blends (6). DISCUSSION
Modifications of the approach are required for two components, hydrogen and ethane. Hydrogen causes a large Ap and AU (about 20 times larger than that of other components for a given weight). Flow rates must be increased not only to dissipate the energy developed by a large Ap, but to prevent highly diffusive hydrogen from reaching the sensing elements. Even with these precautions, results, although linear, are 5y0 low. Apparently other effects, such as viscosity effects, become significant with the 20-fold increase in flow rates. Ethane causes an extremely small Ap and AU because its molecular weight is nearly that of nitrogen. The
response is low enough to cause poor precision. The obvious solution is to use a carrier gas of a different molecular weight. For over five years, gas-density detectors have been used to analyze many digerent samples including olefins in naphtha, products of paraffin isomerization and light naphthas in crude (6, 6, 14). Calibrations of thermalconductivity cells would have been difficult and time consuming because of the large number of components, the presence of both liquids and gases in the samples, and the extreme nonlinear thermal-conductivity behavior of hydrogen in helium (13). With the gasdensity detector, no calibration was needed other than the 5% correction for hydrogen. Commercial models of the gas-density detector have been developed by the Gow-Mac Instrument Co. These more sophisticated detectors are made of metal, have interchangeable flowmeters, and may be operated a t 300” C.
raphy,” p. 31, Reinhold, New York, 1957. (3) Martin, A. J. P., U. S. Patent 2,728,219 (Dec. 27, 1955). (4) Martin, A. J. P., James, A. T., Biochem. J . 63, 138 (1956). (5) Martin, R. L., ANAL. CHEW 32, 336 (1960). (6) Martin, R. L., Winters, J. C., Ibid., 31,1954 (1959). (7) Munday, C. W., Primavesi, G. R., ‘‘Vapour Phase Chromatography,” D. H. Desty, ed., p. 146, Academic Press, New York, 1957. (8) Murray, K. E., AuslTakzn J . A p p l . Sci. 10, 156 (1959). (9) Nerheim, A. G. (to Standard Oil Company of Indiana), U. S. Patents 3,090,112 (May 28, 1963),and 3,082,618 (March 26, 1963); and Rushton, J. H., U. S. Patent 3,082,619 (hfaroh 26, 1963); and Tucker, E. B., U. S. Patent 3,050,984 (August 28, 1962). (10) Pecsok, R. L., “Principles a n i Practice of Gas Chromatography, p. 119, Wiley, New York, 1959. (11) Phillips, C. S. G., Timms, P. I,., J . C h T m t o g . 5,131 (1961). (12) Schmauch, L. J., ANAL. CHEM.31, 225 11959). ch, L. J., Dinerstein, R. A., 43 (1960). , J. C., Jones, F. S., Martin,
LITERATURE CITED
(15) Young; J. G., Second Symposium on Gas Chromatography, East Lansing, Mich., June 11, 1959. RECEIVEDfor review May 3, 1963. Accepted July 18, 1963.
(1) Desty, D. H., “Gas Chromatography,” D. H. Desty, ed., p. 200, Academic
Press, New York, 1957.
(2) Keulemans, A. I. M., “Gas Chromag-
Preparation of Methyl Esters for Gas Liquid Chromatography of Acids by Pyrolysis of Tetra methylammonium Sa Its ERNEST W. ROBB and JOHN
J. WESTBROOK 111
Philip Morris Research Center, Richmond, Vu.
b The tetramethylammonium salts of carboxylic acids are converted to the corresponding methyl esters in high yield when they are injected into a commercial gas chromatographic unit at a vaporizer temperature of 330” to 365” C. The tetramethylammonium salts are prepared either b y titrating the acids with tetramethylammonium hydroxide or by ion exchange on an anion exchange resin. This reaction has several advantages compared with existing methylation procedures in the gas chromatographic analysis of acids. However, the yield of methyl ester is lower if the sample size is less than 50 pg.; and oxalic, malonic, malic, and citric acids do not yield any methyl ester b y this procedure.
T
chromatography of carboxylic acids is difficult because their polarity and tendency to dimerize cause trailing, poorly shaped peaks, and HE GAS LIQUID
1644
ANALYTICAL
CHEMISTRY
nonreproducible retention times unless recourse is had to chromatographic phases especially designed for acids. In addition, their polarity results in excessively long retention times for the higher molecular weight acids. For these reasons mixtures of acids are commonly converted to their methyl esters before they are chromatographed, and a considerable number of papers have dealt with methods for methylating acids. Vorbeck et al. (9), and Kirkland (6) have discussed the relative merits of commonly used methylation procedures. Recently two techniques have been described in which acids are converted to esters by a pyrolysis reaction, the esters formed being swept directly onto a chromatographic column. In one procedure a mixture of the dry potassium salts of the acids and ethyl potassium sulfate is heated in a pyrolysis device (6, 8). In the other, the acids, mixed with boron trifluoride etherate
and alcohol, are injcctcd into :t hcatctl chamber ( 2 ) . In an attempt to dcvise a procedure in which the methyl esters would be formed during the chromatographic step, but which would not require any special apparatus, the authors considered the pyrolysis of tetramethylammonium salts of acids. This reaction was reported by Prelog and Picentanida (7), who found that when the tetramethylammonium salts of acids were heated, trimethylamine was driven off and a residue of nearly pure methyl ester remained. The reaction was later applied to the methylation of some sterically hindered benzoic acids by Fuson, Corse, and Homing (3). They reported that pyrolysis a t 200” to 250’ C. gave 60 to 90% yields of methj.1 esters. We have found that this reaction proceeds rapidly a t the temperatures obtainable in the injection ports of commercial gas chromatography units, so that when solutions of tetramethyl-
-
4 -
2,
Figure 1. The methyl benzoate peak in the gas chromatography of tetramethylammonium ber zoate
-.
O n F & M Model 500 programmed temperature gas chromatograph), unit On Perkin-Elmer PR 154 Vapor Fractorneter. Both at column tmsmperature of 150' C.
---.
ammonium salts of acids are injected into the instrument, the methylation reaction occurs in a nearly quantitative yield in the vaporizer and the methyl esters formed are swept onto the column and chromatographed. EXPERIMENTAL
Chromatographic Equipment and Conditions. An I? & PI Scientific Corp. Model 500 programmed temperature chromatographic unit was used with a 4-foot by l/o-inch 0.d. Carbowax 20 M column (20% on 40/60 mesh Chroniosorb W). The helium flow rate was 120 ml. per minute, and the detcctor temperature was 267" C. The coiumn temperature was maintained a t 75" C. for 2.5 minutes after injection of the sample, then was programmed a t 11' C. per minute to a maximum temperature of 220" C. Except where specified as being otherwise, the injection port temperature was 365' C., and sample sizes were 1 to 5 pl. of a solution containing 20 to 25 mg per ml. of tetramethylammonium salt. Procedure : Preparation of Tetramethylammanium Salts. BY TmR.4TION. The acid was weighed out, dissolved in methanol, and titrated t o a phenolphthalein end point with tet,rainetliylammoni~mhydroxide in methanol, prepared from tetramethylammonium chloride and silver oxide by the procedure of Cundiff and Markunas (1). The titrated solution was diluted with niei,hanol to a volume such that the concentration of the acid was between 20 and 2.5 mg. per ml. BY IONEXCHANGE. Amberlite IRA-400 ion exchange resin, 2.0 grams, was converted to the hydroxide forin by the method of Hornstein et al. (4). The resin was washed with four 50-ml. portions of anhydrous methanol by stirring and decantation. Then a solution of 0.203 gram 3f benzoic acid in 10 nil. of methanol was added to the resin. After 10 minutes, the resin was transferred to a colunin (10 X 250 mm.) and washed with 200 ml. of methanol.
A solution of 0.50 gram of tetramethylammonium chloride (Eastman Organic Chemicals, practical grade) in 15 ml. of methanol was added to the column, and the eluate (volume 15 ml., elution rate 0.6 ml. per minute) was collected and a 5-~1, aliquot was chromatographed. Separation of a Mixture of Acids. A solution of 0.375 grFm each of butyric, valeric, caproic, caprylic, capric, lauric, myristic, and palmitic acids in 25 ml. of methanol was prepared. Five milliliters of this solution was titrated with tetramethylammonium hydroxide as described above and diluted t o 25.0 ml. with methanol. A IO-pl. aliquot was chromatographed. A second 5.0-ml. portion of the solution of the acids was added to 6.25 grams of Amberlite IRA 400 which had been converted to the hydroxide form and wadied with methanol as described above. The resin was then filtered, washed on the filter with 200 ml. of methanol, and added to a solution of 5.0 grams pf tetramethylammonium chloride in methanol. The mixture was stirred for 10 minutes and filtered, and the filtrate was diluted to 25.0 ml. with methanol. A 10-pl. aliquot was chromatographed. Determination of Ester Yields. Purified samples of the methyl esters were prepared from commercial samples by preparative scale gas chromatography. Standard solutions of the esters in methanol were chromatographed and linear plots of peak area us. concentration were obtained. Peak areas were measured by cutting out the peaks and weighing the paper. The amount of methyl ester formed when the tetramethylammonium salts were chromatographed was then estimated by comparing the area of the ester peak with the standard plot. All yield figures are the average of two or more determinations. The difference between duplicate determinations was about 3%. RESULTS
AND DISCUSSION
In a preliminary experiment, a solution of tetramethylammonium benzoate was chromatographed on two commercial gas chromatography units, the Perkin-Elmer Model 154 and the F & M Model 500, using the same column and chromatographic conditions. In both cases, peaks due to trimethylamine, methanol solvent,, and methyl benzoate were observed. There was no benzoic acid peak. In both cases, the yield of methyl benzoate, calculated from the peak area, was 91%. However, there was a great difference in the shape of the two peaks, as can be seen in Figure 1. The authors believe this difference is due not only to the difference in temperature in the injection ports of the two instruments, but also to differences in the geometry of the vaporizer heating elements. If some parts of the vaporizer chamber are cooler than others, as is the case in the Perkin-Elmer unit, the
c--i
240
200
PREHEATER
280 TEMPERATURE,
320
360
400
OC
Figure 2. Yield of methyl benzoate as a function of injection port temperature
tetramethylammonium salt left on the cooler parts when the solvent evaporates will be converted to methyl ester more slowly, causing trailing of the methyl ester peak. In any case, all subsequent work was done on the F & M instrument, and sharp, symmetrical ester peaks were always obtained. The optimum injection port temperature was then determined by chromatographing tetramethylammonium benzoate in methanol on the F & M Model 500 unit a t a series of different injection port temperatures. The results are shown in Figure 2. Only small amounts of ester are formed below 220' C.; the yield increases sharply as the temperature is increased to 260' C., and a plateau is found between 330" and 370" C. In this range the average yield of methyl benzoate was 99%. All subsequent work was done a t an indicated injection port temperature of 365" C. It is possible that the optimum decomposition temperature of other tetramethylammonium salts may be quite different from that of tetramethylammonium benzoate; however, the high yields of methyl esters (Table I) indicate that this is not the case for the acids studied.
Table 1.
Yields of Methyl Esters from Various Acids
Acid Acetic Butyric Valeric Caprylic Lauric Myristic Palmitic Benzoic Cinnamic
Yield of methyl ester, % 94 87 92 87 93
Oxalic
Malonic Succinic Levulinic Glycolic Lactic Malic Citric
VOL. 35, NO. 1 1 , OCTOBER 1963
86
88 99 87 0 0
94 89 93
97 0 0
1645
o
a
MICROGRIIMS
80
120
TElR&METHYLIIMM0NlUM
160 SUCCINIITE
2w
+
240
INJECTED
Figure 3. Yield of dimethyl succinate as a function of sample size --_t
2
To ascertain the generality of this reaction we next investigated a number of different acids. The yields of methyl esters are listed in Table I. The fatty acids gave yields in the range 85 to 95%, with no apparent relationship between yield and chain length. Substantially complete conversions to methyl esters were found for aromatic acids, hydroxy acids, and dicarboxylic acids, except for oxalic, malonic, citric, and malic acids. In the case of these four acids no ester peak mas observed. Evidently the salts of these acids decompose entirely to lower molecular weight products a t 365" C. During this study we observed that the yields of ester depended somewhat on the sample size. This effect was investigated systematically by chromatographing various volumes of a solution of tetramethylammonium succinate, and also by chromatographing a constant volume of each of a series of tetramethylammonium succinate solutions of varying concentration. In both cases the yield of ester depended on the amount of tetramethylammonium succinate being chromatographed, the yield being less with very small amounts of the salt. The combined data from these two experiments are plotted in Figure 3. A low yield of methyl ester for small amounts of acids would constitute a serious disadvantage of this procedure, especially if it were used to study mixtures of acids in which some of the acids occur in very small quantities. However, in a mixture of acids, the yield of a given methyl ester does not depend on the amount of the corresponding tetramethylammonium salt but on the total amount of salts in the mixture. This was shown by the following experiment.
A mixture of butyric, succinic, valeric, and cinnamic acids was titrated with tetramethylammonium hydroxide and diluted so that the concentration of each acid in the solution was 5 pg. per microliter and the total concentration of acids was 20 fig. per microliter. A 5+1. sample of this solution was chromatographed. From Figure 3, we can predict that if the amount of each acid determines the yield of its ester, each 1646
ANALYTICAL CHEMISTRY
4
e
6
13
I2
14
6
8
M NLTES
Figure 4. Chromatogram of a mixture of tetramethylammonium salts. Sample preparation and chromatographic conditions are described in Experimental Section 1.
2. 3. 5.
5. 6. 7. 8.
9. 10.
Trimethylamine Methanol Methyl butyrate Methyl valerate Methyl caproate Methyl caprylate Methyl caprate Methyl laurate Methyl myristate Methyl palmitate
eater should be formed in about 75% yield. On the other hand, if the total amount of acids present governs the yield of each ester, the yields of esters should be greater than 90%. The observed yields were: methyl butyrate 99%) methyl valerate 102%, methyl succinate loa%, and methyl cinnamate 91%. This result implies that, in the analysis of a mixture of acids, a reasonably complete conversion of each acid to its methyl ester can be expected as long as the total amount of acids being chromatographed is greater than approximately 50 pg. The tetramethylammonium salts of carboxylic acids can be conveniently prepared by ion exchange chromatography. Acids are quickly and completely absorbed from methanol solution by a strongly basic anion exchange resin such as Amberlite IRA 400 in its hydroxide form (4). Addition of methanolic tetramethylammonium chloride to the reiin causes the acids to appear in the eluate as their tetramethylammonium salts. The eluate can be injected into the gas chromatographic unit as it comes off the resin, or it can be concentrated by evaporation of the solvent before gas chromatography. In an experiment using benzoic acid, a yield of 102% of methyl benzoate mas observed in the gas chromatogram, based on the amount of benzoic acid added t o the resin. In another experiment, an aliquot of a solution containing eight aliphatic acids was titrated with tetramethylammonium
hydroxide, and a aecond aliquot was added to the ion exchange resin, and the acids were removed from the resin with tetramethylammonium chloride. l h e two solutions of tetramethylammonium salts thus obtained were chromatographed. In the two chromatograms the peak heights for corresponding esters were the same within the experimental error. This demonstrates that losses of acids in the ion exchange procedure are negligible. The chromatogram from the titrated solution is shown in Figure 4. The trailing peak underlying the methyl caproate and methyl caprylate peaks is due to water formed in the neutralization reaction. Methanol was used as the solvent for the tetramethylammonium salts throughout this study. Under the chromatographic conditions used, methanol covers the peak from methyl propionate. However, any other solvent capable of dissolving these salts could be used in place of methanol. We have found, for example, that ethylene glycol monomethyl ether is a good solvent for tetramethylammonium salts. In a particular application of this method, a solvent could undoubtedly be found which would not interfere with any of the ester peaks being studied. The trimethylamine formed in the pyrolysis reaction did not interfere with any of the ester peaks, since its retention time vas less than that of methyl formate on the Carbon-ax column. This method of methylation appears to offer several advantages for the gas
chromatographic analysis of acids. Most notable are its convenience and rapidity. A separate methylation step is not required. Furthermore it can be used with a commercially available gas chromatographic instrument without any modification. The method is particularly convenient when combined with the use of ion exchange resins for the analysis of mixtures, since acids can be separated from neutral and basic compounds, convertell to their methyl esters, and examined by gas chromatography in a short, sing;le step procedure. A further advantage here is that the acids can be concentrated after their conversion to nonvolatile tetramethylammonium salts, eliminating the possibility of losing volati>eacids or methyl esters during solvent removal. The variable yields of esters (Table I) make this procedurc inapplicable for
assay-type analyses. The yields are high enough that the procedure can be used for qualitative analyses. It should also be adequate for trace analyses where high precision is not required, particularly if the standard plots of ester peak area us. concentration are prepared by putting known amounts of acids through the entire procedure. The most serious defect of the method is that the salts of some acids-e.g., oxalic, malonic, malic, and citric among the acids we have studied- are decomposed instead of being converted to their methyl esters, and so cannot be detected or analyzed by this procedure.
LITERATURE CITED
(1) Cundiff, R. H., Rlarkunas, 1’. C., ANAL.CHEM.28,792 (1956). ( 2 ) Drawert, F., preprints of papers t o be
presented at the Fourth International Gas Chromatography Symposium, edited by M. vain Swaay, pp. 132-8, Butterworths, London, 1962. (3) Fuson, R. C., Corse, J., Horning, E. C., J . Am. Chem. Soc. 61, 1290 (1939).
(4) Hornstein, I., Alford, J. A,, Elliott, L. E., Crowe, P. F., ANAL.CHEM.32, 540 (1960). ( 5 ) Hunter,’I. R , J . Chromatog. 7, 288
(1962).
(6) Kirkland, J. J., ASAL.
CHEx
1520 (1961). ( 7 ) Prelog, V., Picentanida, RZ.,
33,
z.
Physiol. Chem. 244, 56 (1936). (8) Ralls, J. W., ASAT,. CHEM. 32, 332
(1960).
ACKNOWLEDGMENT
(9) Vorbeck, Xi. L., hlnttick, I,. R., Lee, E’. A . , Pederson. C. S.,Ihid.. 33. 1.512 (i96ij. REPEIVED for review J:tnuary 30, 1063. Accepted Jiily 18, 1963. I
The authors thank R. D. Carpenter for his helpful advice and suggestions.
,
Gas Chr oma t o g ra p hic Dete rmination O f Free Tobene Diisocyanate in Adducts with T rimethy Io Ipropa ne N. R. NEUBAUER,
G .R. SKRECKOSKI, R. G. WHITE,
and A. J. KANE
Nafional Aniline Division, Allied Chemical Corp., Buffalo, N. Y.
b The volatile components of toluene diisocyanate (TDl)-trimethylolpropane (TMP) adducts are vaporized and the amount of free TDI i s determined b y gas liquid chromatography. Conditions are chosen to minimize the thermal dissociation of the adduct. Peak heights of the free TDI and ctn internal marker added to the unknowti are compared to the corresponding peaks of a sample of known free TDI concentration. The method has been used in laboratory studies of TDI-TMP reactions. When compared with a wet chemical method, the gas chromatographic method exhibits greater precision and accuracy in the range of 0.5 to 10% free TDI.
U
are now receiving increasing attention from the paint industry. Thev are solutions of resins prepared by reaction of the -NCO group of diisocyanatcs with the -OH group of polyols. A i,ypical component of two-package urethane coatings is the adduct formed by reacting toluene diisocyanate (TDI) and trimethylolpropane (TMP). Reaction of TDI and ThSP a t sufficiently low temperature takes advantage of thi: relatively greater reactivity of the -KC0 group para to the niethyl group and produces a triisocyanate adduct of high molecular RETHANE COATINGS
weight containing little unreacted TDI (IO). Because they are free of residual TDI, polyisocyanates of this type offer the advantage of decreased tosicity in addition to greater ease in formulation. Although preparation of polyisocyanate adducts of this type is reasonably successful with Nacconate 100 (2,4-TDI), preparation from the more economical Nacconate 80 (80-20 mixture of 2,4-TDI-2,6-TDI) is complicated by the presence of a higher proportion of the less reactive -NCO groups in the ortho position. Various statements in the literature indicate that residual T D I content of adducts prepared from the 80-20 mixture ranges from 0.1 to 15y0depending upon conditions of the reaction ( 2 , 5, 6, IO). Measurement of the free TDI content of these adducts is complicated by the fact that classical methods for the determination of isocyanates are end-group analyses-reaction with amines and infrared spectroscopywhich do not differentiate between mono-, di-, and triisocyanate ( I , 12, I S ) . Methods to separate unreacted TDI from the adduct have been suggested. In certain cases, the low molecular weight diamine produced by the acid hydrolysis of TDI can be separated from the high molecular weight triamine produced by the hydrolysis of the
adduct. Estimation of the diamine can then be made colorimetrically (8). Free TDI has been separated from adducts by vacuum distillation in specially designed apparatus (11) and l y solvent estraction (8). These metliodo, however, do not lend themselves t o routine analysis of a large number of samples. Gas liquid chromatography (GLC) has been discussed briefly as a method for determination of mono-, di-, and triisocyanates (4). The present study concerns the gas chromatographic determination of free TDI in TDI-TAIP adducts. EXPERIMENTAL
The TDI-ThIP adducts were 1 ) ~ pared in a solvent mixture consisting of 2 parts Cellosolve acetate (Union Carbide Chemical Corp.) and 1 part xylene by weight. The final product consisted of 40% solvent by weight. The preparation of the TDI-TMP adducts was carried out by conventional laboratory procedure. The reaction was considered complete when the measured XCO content was constant and about equal to the calculated value. Apparatus. A Perkin-Elmer >lode1 158 vapor fractonieter was used n-ith :t hot-wire detector. Modification was made t o permit control and nieHsurcnient of the vaporizer tempcr;iturc,. VOL. 35, NO. 1 1 , OCTOBER 1963
1647
