Trace Level Determination of Low-Molecular-Weight Alcohols in Aqueous Samples Based on Alkyl Nitrite Formation and Gas Chromatography

ANALYTICAL SCIENCES MAY 2001, VOL. 17 639 2001 © The Japan Society for Analytical Chemistry

Ha Thi-Hoang NGUYEN, Norimichi TAKENAKA,† Hiroshi BANDOW, and Yasuaki MAEDA

College of Engineering, Osaka Prefecture University, Gakuencho, Sakai, Osaka 599Ð8531, Japan

A simple and sensitive method for the determination of liquid methanol and ethanol at trace levels by an alkyl nitrite formation reaction has been established. Alcohol was allowed to react with nitrous acid, which was yielded from sulfuric acid and sodium nitrite in the solution, to form the corresponding alkyl nitrite in the hexane organic phase. Alkyl nitrites in hexane were analyzed by a gas chromatograph with an electron capture detector (GC-ECD). The detection limits, which were determined at a signal-to-noise ratio of 3, were 1.1 and 0.7 µg/L for methanol and ethanol, respectively, by 1 µL injection. The relative standard deviations for n = 8 were 4.0 and 3.3% for methanol and ethanol, respectively. The method was applied to determine the alcohol concentration in a rice paddy, pond water, tap water, and well water. Those aqueous samples were also spiked with standard alcohols; the average recoveries of spiked methanol and ethanol were 98 and 91% with relative standard deviations of 6.1 and 4.0%, respectively.

(Received August 17, 2000; Accepted October 2, 2000)

these reagents are very unstable, overall, it is not simple to Introduction determine alcohols at the trace level. Alkyl nitrites were found to be a good derivatization for alcohol determination, and Alcohols are a group of important organic compounds popularly alcohol in the gas phase at the trace level was successfully used as solvents in laboratories, industries, and households.1 In quantitatively analyzed in an earlier study: alkyl nitrite in the the natural environment, plant and biological objects are a gas phase was formed from a reaction of alcohol and nitrogen variety emission sources of alcohols.2Ð4 Moreover, alcohols are dioxide on a Pyrex glass surface.11 Alkyl nitrite can also be widely found in beverages, foods, and pharmaceuticals.1,5 formed from an alcohol and nitrous acid reaction, which is Methanol and ethanol are the most popularly used alcohols. conveniently applied for alcohol determination in the aqueous Recently, the use of alcohol fuel as an alternative fuel has phase. In this study, we propose a method for the determination received much attention as a solution for many problems caused of liquid alcohols at µg/L concentrations using the alkyl nitrite by gasoline fuel.6 The determination of alcohol is an important formation reaction. GC-ECD was chosen for an alkyl nitrite subject in the field of analysis. Moreover, in an examination of analysis because of its high sensitivity. The simultaneous the alcohol level due to emission from a plant as well as the existence of many substances in environmental samples usually ambient level of alcohol due to an alcohol-fuel vehicle, a causes many problems in quantitative analysis. In most cases, quantitative-analysis technique at the trace level is required. chromatograms with many unexpected peaks as well as ghost However, the determination of alcohols at the trace level is still peaks might be obtained. These problems can easily be avoided a challenging problem.7 There are many methods reported for by using GC-ECD, which has high selectivity. alcohol determination, such as a conductometric method,8 an In this method, alcohols were allowed to react with nitrous electrochemical fuel-cell method,9 and an infrared absorption acid, which was formed by the reaction of sulfuric acid and method,10 which have detection limits at several ppm (parts per sodium nitrite to form the corresponding alkyl nitrites: million). A method using a dehydrogenase-based biosensor, which received much development for many years, was still H2SO4 + NaNO2 → HONO + Na2SO4, (1) found to have difficulty in measuring alcohols at the ppb (or

µg/L) level. Gas chromatography and high-performance liquid ROH + HONO → RONO + H2O. (2) chromatography were also employed for alcohol determinations. In recent years, derivatization techniques for After the yielded alkyl nitrites were then analyzed by GC-ECD, HPLC analysis have received attention because of their highly the alcohol concentration was calculated using a calibrated sensitive detection characteristics. Several types of reagents, conversion factor of each alcohol to its nitrite. such as carbonyl azides and carbonyl chlorides, have been suggested as derivatizations of alcohols. However, the reactions for these derivatizations were under rather strict conditions, Experimental such as at high temperature in the dark etc. Moreover, because All analytical-grade chemicals including methanol, ethanol, † To whom correspondence should be addressed. sulfuric acid, sodium nitrite, and hexane, were obtained from 640 ANALYTICAL SCIENCES MAY 2001, VOL. 17

Wako Pure Chemicals Inc. (Osaka, Japan) and used without further purification. A commercially available ethanol solution of ethyl nitrite was used as its standard. Methyl nitrite standard was synthesized according to a procedure reported by Blatt.12 In order to make a calibration graph for alcohol determination, diluted alcohol standards were prepared by dissolved liquid alcohol (>99.5%) in Milli-Q water (resistance >18 MΩ cm). Liquid samples, such as water from a lake, and rice paddy were first filtered by a coarse filter, and then by a membrane filter. The liquid sample obtained through a membrane filter size of 0.45 µm was used for the analysis. After 3 g of sodium nitrite was dissolved in 30 mL of an alcohol sample, 2 mL of sulfuric acid was dissolved in a 20 mL alcohol sample solution; then 2 mL of hexane, which would be an organic solvent for any yielded alkyl nitrite, was added to the sodium nitrite/alcohol sample solution. All of the sulfuric acid/20 mL alcohol solution, which was set in a separating funnel, was slowly flowed into a sodium nitrite/alcohol sample solution, which was set in an Erlenmeyer flask while the reaction mixture (in Erlenmeyer flask) was being stirred by a stirrer. The flow rate into the sodium nitrite/alcohol sample was Fig. 1 Typical chromatogram of alkyl nitrites yielded from regulated by a screw of the separating funnel. Because alkyl alcohols in a rice paddy (enlarge 2:1). 1, peak of nitrous acid; 2, peak of methyl nitrite; 3, peak of ethyl nitrite; 4, peak of hexane. nitrites are harmful if inhaled continuously, the reaction should be run in a hood. After all of the sulfuric acid/alcohol solution had flowed completely into the sodium nitrite mixture, the reaction mixture was extracted, alkyl nitrite in the hexane phase Optimum concentrations of sulfuric acid and sodium nitrite for was washed with water, and the 1 µL hexane phase was injected alkyl nitrite formation reaction into GC-ECD. The yielded alkyl nitrites and that dissolved in The effect of the sulfuric acid and sodium nitrite hexane were analyzed by GC-ECD. A GC analysis was concentrations on the nitrite formation reaction was examined performed using a Shimadzu GC-4CM gas chromatograph fitted with a concentration range of sulfuric acid of 0.2 Ð 1.1 mM (Fig. with a Teflon column (3 mm inner diameter, and 4 m length). 3A) and a sodium nitrite concentration of 1 M (the sodium The carrier gas was nitrogen with a flow rate of 40 mL/min. nitrite concentration for the maximum yield of alkyl nitrite was The column was packed with tricresyl phosphate on a previously estimate). The results show that the yield of alkyl Chromosorb W-AW (60 Ð 80 mesh). The temperature of the nitrite reached the maximum value when the concentration of injection-port, detector, and column was 25ûC. A sulfuric acid was higher than 0.7 mM. The optimum chromatogram of the alkyl nitrites from alcohols from a rice concentration of sodium nitrite was also studied. The examined paddy is shown in Fig.1. concentration range of sodium nitrite was 0.2 Ð 1.8 M. The nitrite formation was recorded for the reaction of alcohols with a solution having a various sodium nitrite concentration and 0.7 Results and Discussion mM sulfuric acid. As shown in Fig. 3B, the optimum concentration of sodium nitrite was 0.9 M. Organic solvent for yielded alkyl nitrite We chose 0.7 mM sulfuric acid and 0.9 M sodium nitrite Several organic solvents, including isopropyl ether, 1-butyl concentrations for an alkyl nitrite formation reaction from ether, dimethoxy methane, hexane, cyclohexane, toluene, alcohol. benzene, and m-xylene, were examined for their applicability of alkyl nitrite solvent. After 60 and 40 pmol of methyl nitrite and Stable time of alkyl nitrite yielded ethyl nitrite standards, respectively, were injected to 2 mL of Figure 4 shows the stability of alkyl nitrite/hexane while it each organic solvent, 1 µL of the each solution was injected to was being stored from the time of extracting out of an the GC-ECD, and the peak area for each was recorded. The alcohol/nitrous acid reaction mixture. The alkyl nitrite dissolved efficiency was defined as the ratio of the peak area by concentrations were stable until 50 min of storage, and then 1 µL organic solvent added nitrites and the peak area by direct started to undergo decomposition. The result shows that alkyl injection to GC-ECD of 0.03 and 0.02 pmol of methyl and ethyl nitrites were stable in hexane sufficiently long for an analysis. nitrite standards, which contained the same amount of nitrites in 1 µL organic solvent injected if the alkyl nitrite completely Calibration graph, detection limit, and reproducibility of dissolved in the solvent. The results are shown in Fig. 2. The alcohol analysis dissolved efficiencies for most solvents were lower for methyl Figure 5 shows calibration graphs for methanol and ethanol. nitrite than for ethyl nitrite. Although cyclohexane gave a high Both calibrations are straight lines with zero intercept and with dissolved efficiency, it showed a bad effect on the ECD correlations of 0.9990 and 0.9962 for methanol and ethanol, sensitivity, and led to a bad base line. hexane was preferred respectively. The points of 0 µg/L shown in the figure were because of a high dissolved efficiency and good reproducibility made for blank values of methanol and ethanol. The response of its peak-area/GC-ECD. for an ethanol analysis was three to four-times higher than that Within the temperature range (15 Ð 30ûC), the variation in the for methanol, resulting from the lower dissolved efficiency in alkyl nitrite yield was insignificant. For five replicates, the hexane of methyl nitrite than that of ethyl nitrite. Moreover, relative standard deviations of the alkyl nitrite yield were 7.4 compared with ethanol, the reaction efficiency of methanol with and 4.1% for methanol and ethanol, respectively. nitrous acid was lower. The detection limit of alcohols was ANALYTICAL SCIENCES MAY 2001, VOL. 17 641

Fig. 2 Dissolved efficiencies of alkyl nitrites in different organic solvents: , for methanol; , for ethanol.

Fig. 4 Stability of alkyl nitrite in hexane vs. time. ( ) for the formation of methyl nitrite from methanol; ( ) for the formation of methyl nitrite from ethanol

is defined as RRF = Sa/Si, where Sa stands for the peak area of each nitrite yielded per unit amount (1 µg/L) of the

corresponding alcohol; Si is the peak area of hexane. The RRF and its relative standard deviation (RSDV) together with the retention time of alcohols are listed in Table 1. For n =8, the RSDV values of RRF were 4.00 and 3.31%, and the RSDV values of the retention time were 0.33 and 0.49% for methanol and ethanol, respectively. The RSDV values are small enough to assume the calibrated-conversion factor of each Fig. 3 Optimum condition for alkyl nitrites formation. A, acid alcohol to its nitrite to be constant within ca. 5% deviation. sulfuric; B, sodium nitrite. ( ) for the formation of methyl nitrite This method showed rather good reproducibility and could be from methanol; ( ) for the formation of methyl nitrite from ethanol. applied to trace-level measurements of liquid alcohols.

Effect of temperature on the determination of liquid alcohol The variation of the reaction temperature may effect the yield determined at a signal-to-noise of 3:1, which was found to be of alkyl nitrite due to the very low boiling point of alkyl nitrite, higher than three-times the standard deviation by blank values. especially methyl nitrite, at Ð12ûC. In this determination The detection limits of methanol and ethanol were 1.11 and 0.71 method, the alkyl formation reaction was at room temperature, µg/L, respectively, for 1 µL injection (Table 1). because it would be much more cumbersome to control at a The reproducibility of this alcohol determination method was very low reaction temperature. The yield of alkyl nitrite was examined for eight runs, including samples of 75 µg/L methanol examined in the 20 Ð 30ûC range, which may be the room- and 30 µg/L ethanol (run 1 to 4) as well as samples of 150 µg/L temperature variation, and at 0 to 5ûC. The reaction temperature methanol and 60 µg/L ethanol (run 5 to 8). was controlled by dipping the reactor in warm/ice water. A The relative response factor (RRF) for methanol and ethanol thermometer, which was set directly to the reaction solution, 642 ANALYTICAL SCIENCES MAY 2001, VOL. 17

Fig. 5 Calibration of alcohols. A, for methanol; B, for ethanol. Fig. 6 Formation of alkyl nitrite vs. reaction temperature. A, for room temperature range; B, for 0 Ð 20ûC range. ( ) for the formation of methyl nitrite from methanol; ( ) for the formation of methyl nitrite from ethanol Table 1 Standard deviation (SDV) and detection limit (DLa) of liquid alcohols determination

b RRF Retention time 10% variation of ethyl nitrite yield might be included if the Run MethanolEthanol Methanol Ethanol room temperature was changed by ca. 10ûC. At low temperature (0ûC, 5ûC), while no increase in the ethyl nitrite yield with a 1 0.0100 0.0426 1.26 2.19 temperature reduction was shown, the yield of methyl nitrite 2 0.0111 0.0431 1.27 2.20 showed a tendency to increase at low temperature with 1.34 Ð 3 0.0102 0.0398 1.26 2.19 2.00%/ûC. From this result, we proposed that at a lower 4 0.0099 0.0437 1.27 2.21 5 0.0099 0.0414 1.27 2.20 temperature, down to the boiling point of methyl nitrite, a high 6 0.0102 0.0442 1.26 2.20 yield of methyl nitrite from methanol could be obtained. 7 0.0100 0.0435 1.26 2.18 8 0.0103 0.0427 1.27 2.20 Application of the method to measure alcohols concentration in an aqueous sample and spike results Average 0.0102 0.0426 1.27 2.20 The alcohol concentrations in liquid samples, taken from rice SDV 0.00041 0.00141 0.00421 0.0108 paddy, a pond, tap water, and a well in Sakai city, Osaka, were RSDV 4.00% 3.31% 0.33% 0.49% determined. The results are listed in Table 2. The alcohol DL (µg/L) 1.11 0.71 concentrations were a few µg/L for these samples. In the rice paddy, the methanol concentration was rather high compared a. The detection limit was determined at a signal-to-noise ratio of 3:1. with those in other samples. The liquid samples were also b. RRF: relative response factor (mentioned in the text). spiked with 20 µg/L methanol and 12 µg/L ethanol and with 40 µg/L methanol and 24 µg/L ethanol. The spiked results are listed in Table 2. The recoveries for spiked samples were from showed a 1 Ð 2.5ûC variation during the reaction. Three runs 90 to 105%. The good reproducibility and applicability of the were observed for each reaction temperature (0, 5, 20, 23, 25 present method to determine alcohols in aqueous samples could and 29ûC), and the relative standard deviations were 0.85 Ð be proved. 4.93%. The variation in the alkyl nitrite formation response vs. reaction temperature is shown in Fig. 6. The results indicate that in the room-temperature range, the change of yielded alkyl References nitrite was negligible. Although the yield of ethyl nitrite decreased with an increase in the room temperature, which was 1. S. H. Chen, H. L. Wu, C. H. Yen, S. M. Wu, and S. J. Lin, near the boiling point of ethyl nitrite (17ûC), the decrease of J. Chromatogr. A, 1998, 799, 93. ethyl nitrite formation was 0.72 Ð 1.08%/ûC. Therefore, a ca. 2. W. Kirstine, I. Galbally, Y. Ye, and M. Hooper, J. ANALYTICAL SCIENCES MAY 2001, VOL. 17 643

Table 2 Relative recovery of the analysis of alcohol spiked in fresh water, replicate n = 2

Paddy field Pond water Tap water Well water

Conc. (µg/L) of alcohol in samples methanol 17.4 1.0 2.0 1.2 ethanol 3.6 6.1 4.5 2.6 Recovery (%) Spiked 20 µg/L methanol 97.7 91.8 106.3 88.9 12 µg/L ethanol 89.7 92.2 93.0 88.9 Recovery (%) Spiked 40 µg/L methanol 105.8 98.8 98.0 93.8 24 µg/L ethanol 96.9 84.3 87.8 94.2

Geophys. Res., 1998, 103, 10605. Fujita, J. Geophys. Res., 1998, 103, 22281. 3. R. C. MacDonald and R. Fall, Atmos. Environ., 1993, 27A, 8. T. Maekawa, J. Tamaki, N. Miura, N. Yamazoe, and S. 1709. Matsushima, Sens. Actuators B, 1992, 9, 63. 4. M. S. Lamanna and A. H. Goldstein, J. Geophys. Res., 9. W. J. Criddle, T. P. Jones, and M. J. H. Neame, Meas. 1999, 104, 21247. Control., 1984, 17, 10 5. G. Huang, G. Deng, H. Qiao, and X. Zhou, Anal. Chem., 10. A. W. Jones, K. M. Beylich, A. Bjorneboe, J. Ingum, and J. 1999, 71, 4245 Morland, Clin. Chem., 1992, 38, 743. 6. T. Y. Chang, R. H. Hammerle, S. M. Japar, and I. T. 11. H. T. Nguyen, Y. Fujio, N. Takenaka, H. Bandow, and Y. Salmeen, Environ. Sci. Technol., 1991, 25, 1190. Maeda, Anal. Chim. Acta, 1999, 402, 233. 7. E. C. Apel, J. G. Calvert, J. P. Greenberg, D. Riemer, R. 12. A. H. Blatt, “Organic Synthesis”, 1966, vol. 2, John Wiley Zika, T. E. Kleindienst, W. A. Lonnerman, K. Fung, and E. and Sons, New York, 204.