Determination of ethers and alcohols in gasolines by gas

taining minimum levels prescribed by law as well as optimum levels for engine .... Absorbance Algorithm for Reconstruction of Gas Chromatograms from...
1 downloads 0 Views 356KB Size
9202

Anal. Chem. 1892, 64, 3202-3205

Determination of Ethers and Alcohols In Gasollnes by Gas Chromatography/Fourler Transform Infrared Spectroscopy John W. Diehl,’ John W. Finkbeiner, and Frank P. DiSanzo Mobil Research and Development Corporation, Paubboro Research Laboratory, Paulsboro, New Jersey 08066

INTRODUCTION In response to clean fuel legislation, research is underway in the petroleum industry to develop fuels which reduce vehicle exhaust emissions. Part of this effort is the addition of certain ethers and alcohols such as methyl tert-butyl ether (MTBE) and ethanol to gasolines. The capability of accurately measuring these compounds is important in maintaining minimum levels prescribed by law as well as optimum levels for engine performance. A number of techniques have been developed to determine ethers and alcohols such as ASTMD4815,’the oxygen specific FID (0-FID),2and atomic emission detection (AED1.3 All of these methods have had some problem such as hydrocarbon interference with D4815 and reliability with the 0-FID and AED. This compound class has distinct absorbances in the mid-infrared.4 Solution infrared spectroscopy cannot readily deal with mixtures of ethers and alcohols which can potentially occur by deliberate blending or from impurities in the feed ethers such as methanol, tert-butyl alcohol, and even tert-amyl methyl ether in the case of MTJ3E. Gas chromatography (GC), however, to first resolve the compounds followed by infrared spectroscopy addresses this problem. GC/Fourier transform infrared (FTIR)spectroscopy has already been demonstrated as a good quantitative tool for organic acids6 and various pollutanta.6 The results of our investigation into the applicability of GC/FTIR for oxygenates in fuel are presented below.

EXPERIMENTAL SECTION A Hewlett-Packard Model 5890 Series I1 GC/5965B IRD was configured as follows. Column: J&W 60-m X 0.32-mm4.d. 5.0-pm film DB-1. The column was connected to a 0.5-m section of J&W 0.53-mm4.d. deactivated fused silica tubing with a Swagelok low dead volume 1/16-in.union. This retention gap was installed in the injector to allow use of the HP 7673B autosampler. Carrier: Hz, 42 cm/s set at 300 OC (approximately 20 psi). Oven temperature program: 20 O C (0 min) 2 deg/min to 80 OC (0 min) 30 deg/min to 300 OC. Injector: Cool on-columncapillaryinjector with heater turned off. Injector temperature was approximately 5 OC higher than the oven temperature. FTIR Spectrometer Parameters. Detector: wide band MCT (4000-550 cm-1) (nominal D* = 1.0 X 1Olo C~HZO.~/W). Light pipe temperature: 250 OC. Transfer line temperature: 250 OC. Resolution: 8 cm-l. Scan rate: six interferograms coadded for 1 spectrum per second. Selective absorbance reconstructions: second difference reconstruction. Derivativefunction width = 75.78 Table I contains (1)ASTM D4815-88. (2) DiSanzo, F. P. J. Chromatogr. Sci. 1990, 28,73-74. (3) Slatkavitz,K. 3.; Uden, P. C.; Barnes, R. M. J. Chromatogr. 1986, 355, 117. (4) Silverstein, R. M.; Bassler, G. C. Spectrometric Identification of Organic Compounds, 2nd ed.; John Wiley & Sons, Inc.: New York, 1967. (5) Olson, E. S.; Diehl, J. W.; Froehlich, M. L. Anal. Chem. 1988,60, 1920-1924. (6) Gurka, D. F.; Pyle, S. M. Enuiron. Sci. Technol. 1988,22,963-967. (7) de Haseth, J. A,; Isenhour, T. L. Reconstruction of Gas Chro-

matograms from Interferometric Gas Chromatography/Infrared Spectrometry Data. Anal. Chem. 1977,49,49.

Table I. Reconstruction Freauencies frequency range compound (cm-l) compound methanol 1055-1063 diisopropyl ether ethanol 1052-1060 isobutyl alcohol 2-propanol 1141-1149 ethyl tert-butyl tert-butyl alcohol 1207-1215 ether 1-propanol 1056-1064 1,2-dimethoxyethane methyl tert-butyl 1205-1213 (ISTD) ether 1-butanol 2-butanol 1128-1136 tert-amyl methyl ether

frequency range (cm-1) 1122-1130 1037-1045 1199-1207 1123-1131 3665-3673 1185-1193

Table 11. Analyte Densities at 25 OC density density (g/mL) compound (g/mL) compound methanol 0.781 P-butanol 0.787 ethanol 0.782 diisopropyl ether 0.717 2-propanol 0.771 isobutyl alcohol 0.788 tert-butyl alcohol 0.764 ethyl tert-butyl ether 0.737 1-propanol 0.791 1-butanol 0.792 methyl tert-butyl ether 0.733 tert-amyl methyl ether 0.768 Table 111. Selectivities over Toluene compound selectivity compound selectivity methanol 43 diisopropyl ether >lo00 42 isobutyl alcohol 24 ethanol 4 ethyl tert-butyl 152 2-propanol 77 ether tert-butyl alcohol 22 1,2-dimethoxyethane >lo00 1-propanol 124 (ISTD) methyl tert-butyl ether 1-butanol 143 2-butanol >lo00 tert-amyl methyl 119 ether the frequencies for each compound. These were the absorbance maxima in the region of interest &4cm-l. A reference spectrum for each sample’s reconstructions was obtained by averaging the spectra from 0.1 to 0.5 min run time. No analyte or gasoline component eluted during this retention time window. Pure ethers and alcoholswere obtained from AldrichChemical Co. and Wiley Organics. Twenty calibration solutions were prepared in the 0-20 vol % range by pipetting and weighing aliquota in 1-mL incrementa of the analytes with 10 mL of 1,2dimethoxyethane which was the internal standard (ISTD) and diluting to 100 mL with toluene. Multilevel calibration curves were developed on a weight/weight basis. Calibrations in a base fuel did not improve accuracy. Gasolines were analyzed by weighingthe internal standard into a known volume and weight of sample and determining the weight of the ether or alcohol from the calibration curves. The volume percent was calculated by dividingthe determined weight by the density of the compound of interest with this volume dividedby the starting samplevolume. Table I1 contains the densities used for these measurements. There were some small differences between the literature values and those measured in the laboratory. (8) Bowater,I. C.; Brown, R. S.; Cooper,J. R.; Wilkine, C. L. Maximum Absorbance Algorithm for Reconstruction of Gas Chromatograms from Gas Chromatograph/Infrared Spectrometry Data. Anal. Chem. 1986, 58, 2195.

0003-2700/92/0364-3202$03.00/0 0 1992 American Chemlcal socclty

ANALYTICAL CHEMISTRY, VOL. 64, NO. 24, DECEMBER 15, 1092

8208

Flgurr 1. 1250-1000cm-1 selective absorbance chromatogram of the anaiytes and ISTD at 1 volume % each in gasoline. Unlabeled peaks

are from hydrocarbons.

300-

GASOLINE 3288-2808

CM-1

ES0'

,

150-

100-

Flguro 2. 3200-2800cm-1 selective absorbance chromatogram of a typical gasoline highlighting the large number of coeluting and potentially

interferlng compounds.

RESULTS AND DISCUSSION Figure 1shows a GC/FTIR chromatogram of the analytes reconstructed from 1250 to lo00 cm-l. Most analyses were performed in thisC-O stretchingregion. All of the compounds were chromatographically well resolved except for isobutyl alcohol and ETBE. Better separation could have been obtained by starting at a lower oven temperature, but at a sacrificeof analysis time. The absorbancemaxima of isobutyl alcohol and ETBE, 1041 and 1203 cm-1, respectively, were sufficiently different that they were measurable even with coelution. Figure 2 shows a 3200-2800-~m-~ chromatogram of a typical gasoline which highlights the potential hydrocarbon interferences. Gasolines usually contain (24412 saturates and olefiis and C H 1 2 one- and two-ring aromatics all of which have strong absorbances in the 3200-2800-~rn-~ spectral region. These classes also have some absorbance in the 1250-1000 cm-1 range resulting from methylene twisting and wagging, and aromatic C-H in-plane bending.4 Of these, the aromatic absorption is the most intense. This can be

seen in Figure 3 where the gasoline from Figure 2 was reconstructed between 1250 and lo00 cm-l. Even though there are no ethers or alcohols in this fuel, a number of peaks can be seen. By GC/FID, 2-methylpentane which eluted a t 12 min was about 5.5 wt 5% of the sample while benzene a t 20.5 min was about 1.5 wt %, yet their peak areas in the 1250-1o00-cm-1 region were approximately equal. GC/FTIR was not found to be as selective over hydrocarbons as oxygen-specific detector^,^^^ and Table I11 contains selectivities relative to toluene. These were calculated by dividing the chromatographic peak area of 1volume % of an analyte a t ita respective selective wavelength reconstruction region by the chromatographic peak area of 1volume % of toluene a t the analyte's wavelength region. The selectivities ranged from 4 to >1000. Reconstructions a t the narrow frequency ranges listed in Table I1 were necessary to completely eliminate coeluting hydrocarbon interferences. No background hydrocarbon signals were present in any selective absorbance reconstructions of a number of gasolines

5204

ANALYTICAL CHEMISTRY, VOL. 64, NO. 24, DECEMBER 15, 1992 GRSOLINE 1250-1088 CM-l 140-

120-

100-

00-

eo-

A

0-

2

4

6

0

IE

IO

14

IE

I0

20

22

24

Fwr 8. 1250-1000cm-1selective absorbance chromatogram of the same gasoilne as in Flgure 2 which shows that the gasoilne hydrocarbons have some signal In the C-0 stretching region. GRSOLINE

128s-1213sm-l

I C

f

Flgwo 4. 1205-1213-cm-1selective absorbance Chromatogram of the gasdlne from Figures 2 and 3. This frequency range was used for MTBE quantitation. No signal Is present at the MTBE retention time of approxlmately 12 min.

and blending feeds. This can be seen in Figure 4 where a typical gasolinehas been reconstructed a t MTBEs frequency range of 1205-1213 cm-1. No peak is evident a t the MTBE retention time of 12 min. Figure 5 shows the 1205-1213-cm-1 chromatogram of the 1volume % mixture from Figure 1.The narrow reconstruction frequency reduced the MTBE signal by approximately 50% compared to the wider range of 125010oO cm-1. There was a problem in the case of l-butanol's C-O stretching absorbance at 1042 cm-l. This alcohol coeluted with benzene, a common component in gasoline8 typically found in the 1-2 w t % range. The selectivity of l-butanol over benzene a t 1042 cm-l was only 3. l-Butanol had to be analyzed a t the O-H stretching frequency of 3669 cm-l where the selectivity was 143. Since this is a weaker IR band the detection l i i i t for this compound was 0.5 volume % compared to 0.1% for the other analytes. Although no quantitation problems were encountered with the 12 compounds addressed here, this was definitely a situation where both gas chromatography and infrared spectroscopy were needed to ensure good analytical data. If

other ethers or alcohols are to be determined, interference problems have to be addressed, especially if coelution with an aromatic hydrocarbon can occur. All calibration curves were linear even up to 95 volume 7% with correlation coefficients >0.999. This was more in line with the results reported by Gurka and Pyles than the nonlinear ones reported by Olson et In the latter case, the analytes and their respective internal standards not only coeluted chromatographically but had relatively close C - 0 absorptions. This may explain the nonlinear calibrations. All calibration curves were found to be stable for several months. Table IV contains precision and accuracy data for the analytes a t the 5,10, and 15 volume % levels in gasoline (7.7 mg/mL =I approximately 1 volume %). The average relative standard deviation (RSD) (n- 1) was 0.3 % ,and the average percentage accuracy was 0.8 7%. The RSD with split injection was 1.0%, and on-column injection was definitely the more precise injection technique. There was some peak shape distortion with C4 alcohols above the 10% level apparently from overload of the 5.0-pm film column, and this

ANALYTICAL CHEMISTRY, VOL. 64, NO. 24, DECEMBER 15, 1992

8205

I VOLUME X RNRLYTES I N GRSOLINE l2Q5-1P13 CM-1

40-

, MTBE

20-

; r

n

A

e-' 2

4

8

I

I0

12 Ttmm (.I".>

Table IV. Precision and Accuracy Data. determined (mean actual RSD % compound mg/mL) (mg/mL) (n- 1) accuracy 39.11 39.14 0.5 0.1 methanol 0.1 0.7 39.19 38.93 ethanol 0.1 0.2 38.52 38.49 2-propanol 0.3 37.45 1.6 36.85 tert-butyl alcohol 0.4 0.2 39.58 39.43 1-propanol 1.1 0.4 36.50 36.91 methyl tert-butyl ether 0.1 0.4 39.18 39.18 2-butanol 2.3 0.2 36.08 36.93 diisopropyl ether 0.2 ether isobutyl alcohol 0.3 39.08 39.18 1.3 0.4 36.50 36.04 ethyl tert-butyl ether 2.0 0.3 38.42 39.41 1-butanol 1.0 0.4 38.15 38.52 tert-amyl methyl ether 84.25 82.67 0.2 1.9 methanol 81.49 0.1 82.07 0.7 ethanol 81.43 80.96 0.1 0.6 2-propanol 78.21 78.50 0.5 0.4 tert-butyl alcohol 83.76 1.3 82.67 0.2 1-propanol 74.22 0.3 0.1 74.33 methyl tert-butyl ether 0.3 83.75 82.07 1.5 2-butanol 72.97 72.51 0.2 0.6 diisopropyl ether 82.73 80.91 0.2 1.0 isobutyl alcohol 73.24 73.94 0.1 1.0 ethyl tert-butyl ether 82.29 83.70 0.2 1.0 1-butanol 76.13 77.05 0.1 1.2 tert-amyl methyl ether 118.16 117.83 0.3 0.3 methanol 117.68 117.56 0.1 0.1 ethanol 116.48 116.48 0.1 0.0 2-propanol 112.98 1.3 114.52 0.2 tert-butyl alcohol 119.11 0.0 119.06 0.3 1-propanol 0.2 110.62 110.86 0.8 methyl tert-butyl ether 0.5 0.1 119.43 119.31 2-butanol 0.5 0.1 108.78 108.63 diisopropyl ether 0.1 0.3 118.74 118.35 isobutyl alcohol 0.6 1.2 111.17 ethyl tert-butyl ether 109.84 0.6 1.7 118.03 120.06 1-butanol 0.6 113.79 1.6 115.63 tert-amyl methyl ether 0

1

n = 10. 7.7 mg/mL = approximately 1 volume %.

had an adverse effect on accuracy. Accuracy was improved when these solutions were diluted 5:l with toluene. Sample

14

I8

I@

. - .A

a0

21

Table V. Analysis of Fuels fuel interlab' (volume 7%) A MTBE 14.6 0.3b B MTBE 14.5 0.9 C MTBE 13.9 f 0.8 D MTBE 14.6 f 0.8 E MTBE 15.2 f 0.9 F ETOH 9.7 0.6 G ETOH 9.7 f 0.7 H ETOH 9.6 f 0.5 I ETOH 9.6 f 0.7

*

14

0-FID 14.W 15.1 14.7 14.8 15.5 9.8 9.5 9.9 9.7

GC/FTIR 14.4 f 0.2d 14.6 i 0.2 13.9 i 0.2 14.8 f 0.2 14.9 i 0.2 9.0 0.1 9.4 f 0.1 9.3 f 0.1 9.0 f 0.1

*

0 Mean of D4815 results from 7 to 9 participating laboratories. Standard deviation, 28 = 95% confidence limit. e Precision data not available for 0-FID.d From Table IV.

b

dilution was not normally performed because of analysistime requirements. Table V shows the analyses of a number of test fuels compared to 0-FID, and the mean of the results from 7 to 9 laboratories employing D4815. Although the GC/FTIR results seemed to be lower than those of the 0-FID,all resulta were well within the precisions of the methods. The RSD's reported for D4815 ranged from 2 to 7 95, which were much worse than the 0.1-0.895 reported here for GC/FTIR (Table IV). Precision data was not available for the 0-FID.

CONCLUSION GC/FTIR is definitely a precise and accurate technique for measuring C 1 4 4 alcohols and C5 and C6 ethers in gasolines. The use of the correct absorbance reconstruction frequencies gives good selectivity over hydrocarbons as well as very linear and stable calibration curves. These features should make it applicable to the analysis of other ethers and alcohols as well as other compound classes as long as possible chromatographic and spectroscopicinterferences are kept in mind. RECEIVEDfor review July 2, 1992. Accepted September 30, 1992.