Determination of Alkylpolynitrates by Electron Capture Gas Chromatography-Application to Air Pollution Ettore Cameral and Dario Pravisanil Research Dept. of Dr. Ing. M . Biazzi S . A., Dinamife S.p.A, Mereto di Tornba, Udine, Italy
ALTHOUGH IT IS KNOWN that electron capture detectors are very sensitive to the group -ON02 ( I ) , there has not been reported any application of this type of determination to the analysis in air of very small quantities of alkylpolynitrates, substances of a great significance in pharmaceuticals and of wide interest in the field of propellants and explosives. Very high sensitivity combined with simultaneous low sensitivity to other products-e.g., hydrocarbons and simple oxygenated compounds-make this method interesting not only for the usual analysis, but also for the quantitative determinations of the alkylpolynitrates as air pollutants. Conditions for the determination of three of the most important industrial alkylpolynitrates-Le., ethylene glycol dinitrate (EGDN), 1-2 propylene glycol dinitrate (PGDN) and glycerol trinitrate (NG)-have been considered in this research. Alkylpolynitrates are toxic substances and atmosphere pollution arising from manufacturing processes constitutes a problem to industry; the higher the vapor pressure of the substance, the more reason for worry. Thus, among the above-mentioned compounds, N G is by far the least dangerous from the pollution viewpoint, because its vapor pressure is low compared with that of the remaining two-e.g., at room temperature it is at least 70 times lower than that of EGDN (2,3). Two methods of analysis of low concentrations of nitroesters in the air are: (a) based upon the nitration of inetaxylenol in sulfuric acid and the colorimetric determination of the nitroxylenol isolated by steam distillation ( 4 ) , and (b) based on the finding that nitrites are formed at a defined rate during the saponification of nitroesters with KOH; the nitrites are diazotized and coupled, and the resulting highly colored salt is determined spectrophotometrically (5). In electron capture detection, the relationship between recorder response and vapor concentration is similar, under ideal conditions, to that of Beer's law for light absorption (6). Every increase of the vapor concentration, c, absorbs the same fraction of electrons incident upon it, that is the number of electrons N reaching the anode per second, is: N = N oe-KCX where No is the number of electrons reaching the anode per second in the pure carrier gas, K is the electron absorption coefficient of the vapor and X i s a proportionality constant. Present address, Dinamite S.p.A., Mereto di Tomba, Udine, Italy. (1) E. F. Darley, K. A. Kettner, and E. R. Stephens, ANAL.CHEM., 35, 589 (1963). (2) J. D. Brandner, Iizd. Eng. Chem. 30,681 (1938). (3) A. Marshall, J . SOC.Chem. Ind. 49, 32 (1930). (4) H. Yagoda and F. H. Goldman, J. Znd. Hyg. Tux. 25, 440 (1943). (5) J. E. Currah, Paper presented, Analytical Subject Division, Chemical Institute of Canada, Quebec City, February, 1953. (6) J. E. Lovelock, ANAL.CHEM., 35, 474 (1963).
Nevertheless, some factors such as the capacity of the electrode to function in different additional modes besides as an electron absorption detector, may be the cause of not infrequent errors in quantitative analyses. The results of our experiences prove that the use of the method of air gas chromatographic analysis by electron capture detection is not associated, within the conditions described, with anomalous quantitative errors, so that this method can be accepted as an alternative to the existing one. It is characterized by great specificity because a peak, and therefore a determination, is obtained for each individual nitroester. Moreover, its full value includes also nitro compounds, such as those derived from benzene and toluene, which are also of paramount importance in industrial hygiene. EXPERIMENTAL
Apparatus and Samples. An apparatus, C. Erba, Mod. C, Type AID/f, equipped with an electron capture detector with a 300 mc. tritium source, was used. The electrical potential to the cell was set at 20 volts which was near the inflection point on the current voltage curve and which gave a standing current of 3.5 A. In the selection of the column, particular attention was paid to working conditions, because of the instability of the compounds, and to the material used, because of the very high absorption, particularly in the instance of NG. Excellent results were finally obtained by using a glass column with 2-mm i d . , 25 cm long, packed with 10% Igepal CO-880 on 80-120 mesh Chromosorb P, siliconized. Operating parameters were: column temperature, 120" C (160" C in the case of NG); vaporizing temperature, 170" C ; nitrogen pressure, 2.5 kg per sq cm; nitrogen flow, 133 ml/min; chart speed, 1 inch per 3 minutes. Standards were prepared by nitrating pure alcohols and purifying the esters by recrystallization from solutions containing 15% (by volume) of nitroesters and 85% peroxide free diethyl ether. Procedure. All the attempts to inject, with a glass-Teflon gas-tight syringe directly into the apparatus, samples of 1-10 ml of air taken from polluted rooms were not successful because of marked absorption of nitroesters on the walls. However practical and alluring, the method had to be given UP * For air sampling, air was bubbled at a rate of 1 liter per minute through two impinger sampling glass tubes, connected in series by means of a small hose and containing 10 ml of ethyl alcohol each, Sampling efficiency was already proven (5) by bubbling continuously through the alcoholic solution contained in the two glass tubes for a known time a nitrogen flow measured with a rotameter and containing known contents of EGDN. The quantities of EGDN absorbed by the alcoholic solution, spectrophotometrically analyzed agreed within experimental error with the theoretical data. An apparatus particularly suitable for sampling is the midget impinger sampling device (Mine Safety Appliances). EGDN and PGDN are generally present in workrooms in quantities ranging from 1 to 5 mg/cu m (14-79 pphm at atmospheric pressure and 20" C). VOL. 39, NO. 13, NOVEMBER 1967
1645
The procedure consists of bubbling 10 liters of air through 20 ml of alcohol contained in two tubes, then thoroughly mixing the 20 ml of alcohol together, and injecting 5 pl. With higher concentrations, it is advisable to dilute the alcoholic solution accordingly. The areas (or heights) of the peaks thus obtained, must be compared with those of standard solutions, to be renewed every week, having concentrations approximately corresponding to the unknown ones. To minimize the possibility of errors originating from unsteady performance of the apparatus or from the influence of impurities, if any, on the baseline, the following method was adopted: three single injections of three standard solutions were made, then double injections of the samples at the unknown concentration in a number not higher than five and, then the three single injections of the standard solutions were repeated. With the average results of the two series of standard injections, the calibration curve can be plotted and it tends to a straight line. Figure 1 represents the chromatogram of PGDN, mononitrotoluene (MNT) and EGDN, carried out, in this instance, at 110 C and 1 Kg/sq cm to emphasize the presence of MNT, which in some cases, may also pollute the work atmosphere. The concentration of N G in the air is usually lower than 0.1 mg/cu m (1.6 pphm) and its presence as a pollutant does not therefore cause any worry at all. For this reason and because of the lower sensitivity of the electron capture detector to this compound, if an analysis must be made, it is necessary to bubble a quantity of air between 100-300 liters into the 20 ml. of alcohol. The most common case is the presence in the atmosphere of EGDN and N G at the same time. This does not influence the analysis of one compound with respect to the other. Figure 2 shows the chromatogram of NG in presence of EGDN in a quantity 10 times larger.
r
d,
P 2
N aa
m
t
w
VI
z
0 0,
VI
w a
a
W P
O
RESULTS AND DISCUSSION
To measure the accuracy of the method, the results of the analysis of 10 alcoholic solutions, at known concentrations, respectively, of PGDN, EGDN, and NG, were considered. Every analysis was repeated three times. The corresponding standard average relative deviations are: PGDN, 3.27 %; EGDN, 3.88%;; NG, 7.24%. Likewise, the standard average relative deviation obtained in similar analyses of alcoholic solutions containing EGDN and NG (in the ratio 10:1) at the same time are: EGDN, 4.95 % and NG, 4.26%;;. The standard average relative deviation calculated on all of the 150 analyses carried out is : 4.72 %, which is a good value for such delicate analyses. The level noise present was 2 X 10-liA. From the measure of the amperes, corresponding to the recorder response, obtained by injections of 5 p1 of a solution at a known concentration and assuming that the detection occurs at a 3 to 1 signal-to-noise level, the sensitivity of the method-Le., the minimum detectable quantity, was calcug; MNT, 2 X g ; EGDN, lated: PGDN, 5 X 2 x io-log; NG, 2 x 10-9g.
1646
ANALYTICAL CHEMISTRY
a 0
0
w
E
II
8
6 4 2 MINUTES
0
Figure 1. Chromatogram of 5 pl sample of ethyl alcohol solution containing 0.5 pg PGDN, 1.5 pg MNT, and 0.5 pg EGDN per ml
8
6 4 2 MINUTES
0
Figure 2. Chromatogram of 5 p1 sample of ethyl alcohol solution containing 5.5 pg NG and 55 pg EGDN per ml
The degrees of sensitivity ascertained correspond to the possibility of detection in the air according to the procedure described-Le., witha sampling of 10 liters of air, of 0.02mg/cu m. (0.3 pphm) PGDN; 0.08 mg/cu m (1.4 pphm) MNT; 0.08 mg/cu m (1.3 pphm) EGDN and, by bubbling of 100 liters of air, of 0.08 mg/cu m (0.8 pphm) NG. Practical analytical tests in polluted workrooms gave results agreeing with the ones obtained with the spectrophotometric method. The possibility of simplifying the air sampling procedure to the point of enabling sampling and direct injection, with a syringe, of about 1 ml of air would be highly desirable and really interesting. ACKNOWLEDGMENT
We thank F. Dal Dan for stimulating discussions and M. Biazzi for permission to publish this paper.
RECEIVED for review May 22,1967. Accepted August 4,1967.
