0.02%. Calibration curves were determined only for the 2,4-, 3,5-, and 2,5-DNT isomers and for the 2,3,4- and 2,4,5TNT isomers because none of the other isomeric impurities were observed (> 0.02 %) in any production samples of TNT which were chromatographed. This observation is consistent with the theoretically expected TNT isomeric distribution as depicted in a toluene nitration chart which appears in reference (13). (13) P. de Beule, Bull. SOC.Cliim., Beiges, 42,27 (1933).
ACKNOWLEDGMENT The authors thank D. N. Thatcher of Eastern Laboratory, Explosives Du Company, for supplying
TNT samples and for much information. Frank Pristera of Picatinny Arsenal, Dover, N. J.9 supplied Samples Of 23334-and 23435-TNT3 and we are grateful to him. RECEIVED for review April 14, 1967. Accepted June 14, 1967.
Quantitative Gas Chromatographic Determination of Ethanolamilnes as Trifluoroacetyl Derivatives L. E. Brydial and H. E. Persinger Union Carbide Corp., Chemicals and Plastics, South Charleston, W. Va.
COMMERCIAL ETHANOLAMINES are mixtures of mono-, di-, triethanolamine, and low concentrations of various impurities, Determination of the individual components of ethanolamine mixtures is usually made by chemical methods which are nonspecific and, for the most part, inaccurate. Monoethanolamine can be determined by the Van Slyke manometric method (1) or by the colorimetric procedure of Critchfield and Johnson (2). Diethanolamine is usually determined by a method which is based on the reaction of mono- and diethanolamine with periodic acid (3). Triethanolamine is determined by a nonaqueous titration after the mono- and diethanolamine have been converted to less basic amides by reaction with acetic anhydride. The Van Slyke and periodate methods must be followed precisely to obtain reproducible results. Triethanolamine is also reported to react with periodate under certain conditions (3). The major limitation of the Critchfield and Johnson method is that it uses reagents which contain both di- and triethanolamine. The results of the triethanolamine determination are affected by the particular solvent and indicator employed, and show a n inverse relationship to the amount of acetic anhydride added ( 4 ) . The sum of the individual determinations for the components in ethanolamine mixtures usually totals more than 100 %. The present investigation for more reliable analytical methods for this important class of compounds was prompted by these shortcomings. In an evaluation of direct gas chromatographic methods, the most favorable results were obtained using a column of 5 % Carbowax 20M on Haloport F. However, analysis by direct procedure is complicated by the strong hydrogen Present address, Union Carbide Corp., Chemicals and Plastics, Bound Brook, N. J. 08805 (1) E. F. Hillenbrand, Jr., and C. A. Pentz, “Organic Analysis” Vol. 111, Mitchell, Ed., Interscience, New York, 1956, p. 148. 28, 430 (2) F. E. Critchfield and J. B. Johnson, ANAL.CHEM., ( 19 56). (3) G. F. Smith, Jr., “Analytical Applications of Periodic Acid
and Iodic Acid and Their Salts”, 5th ed., G. Frederick Smith Chemical Co., Colurnbus, Ohio, 1950, p. 99. (4) L. E. Brydia and H. E. Persinger, unpublished work, Union Carbide Corp., August 1964. 131 8
ANALYTICAL CHEMISTRY
bonding character of the ethanolamines which causes excessive peak tailing. This prevents an exact quantitative analysis. Because direct gas chromatographic procedures were not entirely satisfactory, chromatography of derivatives was investigated. Trifluoroacetic anhydride has been widely used to convert nonvolatile amino acids (5-8) and amines (9-11) to volatile N-trifluoroacetamide derivatives, but this reagent has not been used for the analysis of alcoholamines. In fact, very little quantitative analytical data on mixtures of amines using this reagent have been reported. This paper describes a gas chromatographic procedure for the quantitative determination of ethanolamines as their trifluoroacetyl derivatives.
EXPERIMENTAL Apparatus. An F and M Model 810 gas chromatograph equipped with dual columns and thermal conductigty detector was used. The detector was equipped with w-2 filaments and was operated at a current of 185 mA. The injection port and detector were maintained at 250” C, and the column was operated isothermally at 172” C. Helium was used as carrier gas a t a flow rate of 30 ml per minute. Chromatographic Columns. The columns were aluminum, 5 feet long with 0.25-inch 0.d. The columns were packed with 5 z neopentylglycol succinate (F and M Scientific Corp., catalog L P 93) on 60- to 80-mesh Chromosorb-G (Johns-Manville Products Corp.). Chemicals. Trifluoroacetic anhydride was obtained either from Eastman Kodak (catalog 7386) or Matheson, Coleman and Bell (catalog TX 1285) and was used as received. Monoethanolamine (MEA), diethanolamine (DEA), and triethanolamine (TEA) were products from the Union Carbide Corp. (5) W. M . Lamkin and C. W. Gehrke, ANAL.CHEW., 37, 383 (1 965). (6) P. A. Cruickshank and J. C. Sheehan, Zbid., 36, 1191 (1964). (7) K. Blau and A. Darbe, J . Cliromarog., 17,445 (1965). (8) S. Makisumi and H. A. Saroff, J . Gas Clirornarog., 3, 21 (1965). (9) W. J. A. Vanden Heuvel, W. L. Gardiner, and E. C. Horning, ANAL.CHEW,36, 1550 (1964). (10) R. A. Morrissette and W. E. Link, J . Gas Chromatog., 3, 67 ( 1 965). (11) W. H. McCurdy, Jr., and R. W. Reiser, ANAL.CHEM., 38, 795 (1966).
Table I. Response Factors and Derivative Data for Components of Ethanolamine Mixtures TFAA No. of Relative derivative groups response Component Abbreviation (mol wt) reacting factor MEA 253.11 2 0.62 Monoethanolamine DEA Diethanolamine 393.18 3 0.19 TEA Triethanolamine 431.24 3 1 .oo 2,2’-[2-(2-Hydroxyethoxy)ethylimino]diethanol HEEID 481.30 3 1 .00a a Pure material not available for determination of response factor. Response factor of 1.00 assigned.
T F A ANHYDRIDE X 1024
Table 11. Weight Per Cent Analyses of Trifluoroacetyl Derivatives of Ethanolamines by Gas Chromatography Added Determineda * MEA DEA TEA MEA DEA TEA 3.3 2.6 94.1 3.4 2.6 94.0 2.3 8.3 89.4 2.3 8.8 88.9 33.7 32.3 34.0 34.1 32.1 33.8 89.7 5.0 5.3 89.8 4.9 5.3 5.6 89.0 5.4 5.1 88.5 5.8 a
0
2
4
6
Approximate retention time (minutes) 4.9 8.6 11.0 23.5
8
IO
12
16
14
18
20
22
24
CH2-CH2-O-CO-CF3
/ N-CO-CF3
26
MINUTES
Figure 1. Chromatogram of trifluoroacetyl derivatives of ethanolamines Procedure. One milliliter of trifluoroacetic anhydride was added to a 2-ml serum bottle, and the bottle was stoppered with a rubber septum cap. Air was evacuated from the bottle with a 10-ml hypodermic syringe. The derivative was prepared by slowly adding 0.05 ml of sample t o the serum bottle by means of a 0.25-ml hypodermic syringe with continuous shaking. (If the sample is added too rapidly, the pressure buildup will force the syringe needle to be separated from the body of the syringe. This is especially important when analyzing samples which are high in monoethanolamine.) After the sample was added, the syringe was removed, and the serum bottle was shaken occasionally for 5 to 10 minutes. Five-microliter samples (equivalent t o 0.25 111 of ethanolamines) were introduced into the gas chromatograph by means of a 10-111 hypodermic syringe. The peak areas were measured, and the weight per cents were calculated using the appropriate response factors. Response Factors. The response factors were obtained by analyzing synthetic mixtures which were prepared using the highest purity ethanolamines available. RESULTS AND DISCUSSION
Trifluoroacetic anhydride reacts with the amino groups of mono- and diethanolamine and the hydroxyl groups of mono-, di-, and triethanolamine as shown in the following equations:
All results are averages of two or more determinations. Results are not corrected for the small amount of water present.
+ 2 CFsCOOH
\H
/
N-CH2-CHzOH
\
(1)
\
$. 3
CFa-C
?’ \\0 /
CH2-CH2-0-CO-CF3
/ N-CHr-CH~-O-CO-CF3 \CO-CFs
CFs-COOH
(2)
/ N-CHz-CHy-0-CO-CF3 + 3 CF3-COOH \CH~--CI~Z--O-CO-CFS
(3)
$. 3
0 //
//
/CH,CH*oH+
N-CH2-CHyOH
\CHz-CHy
OH
3
CF3-C\
CF3-C
7’ \\0
CH2-CH2-0-CO--CFI
iH2-cH*oH \
N-H
H
4-2
CF3-C\
7-
*3-1 0
The products of these reactions were verified using nuclear magnetic resonance and mass spectrometry. VOL. 39, NO. 1 1, SEPTEMBER 1967
1319
Table 111. Comparison of Analyses by Two Independent Laboratories of Samples of Known Composition Weight - .Der cent determineda,* Weight per cent added Laboratory A Laboratory B -__-_..___ MEA DEA TEA HEEID MEA DEA TEA HEEID MEA DEA TEA HEEID ... 0.81 98.92 0.27 Nil 0.80 98.91 0.30 Nil 0.83 98.93 0.24 1.04 1.77 96.92 0.26 0.88 1.70 97.16 0.26 1.12 1.81 96.86 0.20 3.18 94.99 0.20 1.07 4.14 94.53 0.26 1.04 3.91 94.81 0.24 1.02 1.80 4.82 93.12 0.25 1.94 4.58 93.23 0.26 1.97 4.64 93.15 0.24 5.55 5.85 88.40 0.20 5.23 5.95 88.66 0.16 5.02 5.19 88.95 0.24 a All results are averages of two or more determinations, using at least two separate derivative preparations. * Results are not corrected for the small amount of water present.
Reactions of the ethanolamines with trifluoroacetic anhydride are quantitative under the conditions of the method. Derivative preparation is simple and rapid. Although the preparation of a fresh derivative just prior to chromatographic analysis is recommended, derivatives have been reanalyzed 3 weeks after preparation without any significant change in the chromatographic results. Complete elution of all volatile derivatives requires less than 25 minutes, and total analysis time using a planimeter is approximately 1 hour. The total elapsed time for an analysis is less than 0.75 hour when an automatic data reduction system is used, but only a third of this time requires direct operator attention. The total analysis time required for the determination of the separate components of a n ethanolamine mixture by the usual wet chemical procedures greatly exceeds the time required to obtain the corresponding data by gas chromatographic analysis of their trifluoroacetyl derivatives. In addition to a shorter analysis time, the gas chromatographic procedure provides impurity information which is unobtainable from the analytical methods which are usually employed for the analysis of these compounds. For example, high purity triethanolamine usually contains from 0.1 to 1 % of 2,2'[2-(2-1iydroxyethoxy)ethylimino]diethanol. This tertiary amine impurity is titrated in the nonaqueous titration which is employed for the quantitative determination of TEA. Therefore, the purity of TEA samples is invariably lower than the reported values. However, 2,2'[2-(2-hydroxyethoxy)ethylimino]diethanol and other impurities are determined individually by the present method. Ethanolamines also contain low concentrations (98 %). These data were obtained from duplicate determinations of four separate derivative preparations of the sample. The results showed an absolute standard deviation of 0.022, 0.03%, and 0.02 for DEA, TEA, and HEEID, respectively. The MEA content of the sample was below the limit of detection-e.g., 0.05%. The relative standard deviation was 2.0%, 0.03 and 5.4%, respectively. The initial attempt to devise a gas chromatographic procedure based on the trifluoroacetyl derivatives of the ethanolamines was conducted using a hydrogen flame detector. The results obtained with this detector were not reproducible, while those obtained using a thermal conductivity detector were. Therefore, the method was developed using a thermal conductivity detector. Because the concentration level of impurities in highly refined ethanolamines is frequently below the detection limit of the thermal conductivity detector, the flame ionization detector was reinvestigated. An increase in sensitivity and a slight reduction in peak tailing of the trifluoroacetic acid were obtained; however, a decrease in precision, as yet unexplained, was observed.
z,
RECEIVEDApril 5, 1967. Accepted May 31, 1967. Work presented in part a t the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 6,1967.