Anal. Chem. 1995, 67,2015-2019
Determination of Aromatic Hydrocarbons in Gasolines by Gas Chromatography/Fourier Transform Infrared Spectroscopy John W. Mehl,* John W. Finkbeiner, and Frank P. DiSanzo
Mobil Research and Development Corporation,Paulsboro Research Laboratory, Paulsboro, New Jersey 08066
The total aromatic contents of gasolines as well as 25 individual gasoline-range aromatic hydrocarbons were quantitated by GC/FX'-IR with 1,2-dimethoxyethane as intemal standard. Reconstructed selective wavelength chromatograms (SWCs) were used for each analyte to give g d selectivities over respective coeluting hydrocarbons. Calibrations and quantitations were based on the peak areas from these SWCs, and all calibration curves were linear. An average relative standard deviation of 0.9%and an average accuracy of 0.9%were found. GC/m-IR provided analyses comparable to the current EPA-specified GC/MS method with less sample preparation than GC/MS and with the added advantage of the simultaneous measurement of the ethers and alcohols (oxygenates) in the fuels. As a result of clean fuel legislation, the levels of two classes of compounds present in reformulated gasolines currently are or soon will be regulated in many parts of the United States. These two classes of compounds are oxygenates, mainly ethers and/or alcohols added to reduce exhaust emissions, and aromatic hydrocarbons. The US. Environmental Protection Agency @PA) currently mandates' a gas chromatography/oxygenflame ionization (GC/O-FID) method,lS2to determine the oxygenates, a gas chromatography/mass spectrometric (GUMS) to determine approximately 25 individual aromatic compounds typically found in gasolines and to estimate the total aromatic levels, and American Standard of Testing Materials method D3606 to determine benzene. A complete analysis requires three instruments and the addition of five internal standards to each sample. The proposed inclusion of benzene in the GC/MS analysis3 may eliminate method D3606, but the amount of instrumentation and sample preparation will still be substantial. We have recently published methods for the determination of oxygenates* and benzene, toluene, ethylbenzene, and m-, p-, and &xylenes Os in gasolines by GC/ET-IR These methods used selective wavelength chromatograms (SWCs) at the analytes' respective unique infrared absorbances (the C-0 stretching region for the ethers and alcohols and the aryl C-H out-of-plane bending region for the aromatics) to obtain selectivity over (1) Environmental Protection Agency, 40 CFR Part SO, Dec. 13, 1993. (2)DiSanzo, F. P.J Chromatogr. Sci. 1990,28, 73-74. (3) Proposed ASTM GC/MS method for determining aromatics in reformulated gasolines. (4)Diehl, J. W.; Finkbeiner, J. W.; DiSanzo, F. P. Anal Chem. 1992,64,32023205. Finkbeiner, J. W.; DiSanzo, F. P. Anal. Chem. 1993,65,2493(5)Diehl, J. W.; 2496. 0003-2700/95/0367-2015$9.00/0 0 1995 American Chemical Society
coeluting hydrocarbons. The work below expands upon this GC/ ET-IR approach for quantitation to include the aromatic hydrocarbons specified by the EPA, to provide a means to estimate the total aromatic levels in fuels, and to provide simultaneous determination of the oxygenates along with the aromatics. EXPERIMENTAL SECTION GC/Fl'-IR System. A Hewlett Packard Model 5890 Series I1 GC/5965B IRD ET-IR spectrometer was configured as follows. GC Column. Hewlett Packard 60 m x 0.53 mm i.d. x 5.0 pm film HP-1. Carrier. Hz, 42 cm/s (5.1 psi) set at 300 "C. Injector. Electronic pressure controlled on-column with a Hewlett Packard 7673B autosampler. Injection volume was 0.5 pL. Injector was not heated but lagged the oven temperature by -5 "C. Oven Temperature Program. From 40 "C (0 min), 2 deg/min to 190 "C (0 min), then 30 deg/min to 300 "C (1.0 min). All of the analytes and the internal standard (ISTD) eluted before the oven reached 190 "C. The fast temperature ramp to 300 "C was to eliminate any carryover between samples. Electronic Pressure Program. From 5.1 psi (75 min), 99 psi/ min to 60 psi (until end of oven temperature program). All of the analytes and the ISTD eluted at 5.1 psi column head pressure. ?he fast ramp to 60 psi was to eliminate carryover between samples. FT-IR Spectrometer Detector. Wide band (4000-550 cm-l) MCT (nominal D* = 1 x 1O1O cm Hz 0.5/W). Light Pipe Temperature. 300 "C. Transfer Line Temperatures. 300 "C. The flow cell exit line was connected to vent outside the GC oven with a 0.25 m x 0.53 mm i.d. section of fused silica tubing. Narrower i.d. tubing caused back pressure in the light pipe, especially when the column head pressure was increased to 60 psi. Resolution. 8 cm-I. Scan Rate. S i interferogramswere coadded for 1spectrum/ S.
Selective Wavelength Chromatogram Reconstructions. The second difference type was used6z7(derivative function width, 75). A reference spectrum for the reconstructions was obtained by averaging the spectra from 0.1 to 0.5 min of each chromatogram. No compounds eluted during this period. As part of the SWC reconstruction, the software produced an absorbance spectrum (6)de Haseth, J. A; Isenhours, T. L. Anal. Chem. 1977,49,49. (7)Bowater, I. C.;Brown, R S.;Cooper, J. R; Wilkins, C. L Anal. Chem. 1986, 58,2195. Analytical Chemistty, Vol. 67, No. 13, July 1, 1995 2015
Table 1. Selective Wavelength Chromatogram Reconstruction Frequencies
frequency range (cm-l)
compound
1,2-dimethoxyethane0) benzene toluene
ethylbenzene m-xylene
p-xylene o-xylene
isopropylbenzene n-propylbenzene lethyl-3-methylbenzene 1-ethyl-4methylbenzene 1,3,5trimethylbenzene 1-ethyl-2-methylbenzene 1,2,4trimethylbenzene 1,2,3-trimethylbenzene indan
1,bdiethylbenzene 1,Cdiethylbenzene n-butylbenzene 1,2-diethylbenzene 1,2,4,5tetramethylbenzene 1,2,3,5tetramethylbenzene 1,2,3,4tetramethylbenzene naphthalene
2-methylnaphthalene 1-methylnaphthalene
1123-1131 670-678 724-732 694-702 687-695 790-798 736-744
695-703 695-703
775-783 807-815 831-839 741-749 801-809 762-770 739-747 698-706 826-834 694-702 748-756 863-871 844-852 798-806 777-785 803-81 1 782 -790
for each chromatographic data point by ratioing the reference spectrum against the spectrum stored for each point. Table 1 contains a list of the wavelengths used for the analytes and the ISTD, 1,2-&methoxyethane. These corresponded to the analytes' aryl C-H out-of-plane bending frequencies f4 cm-'. AU compounds were identified by retention time matching and spiking with pure standards (when available) as well as spectroscopically by GC/FT-IR/MS.8*9 Figure 1 shows a 600-900 cm-l SWC of a typical gasoline with the analytes labeled. Spectrometer Purge. Dry Nz, 40 psi (100 Wmin), was used. The restrictors were removed from the spectrometer's purge gas lines. This high purge rate greatly reduced any danger of explosion if HSreached the IR source because of a light pipe leak. No leaks were observed in 3 years of operation. Data System. Data were processed on a Dell 4560XE IBMcompatible personal computer with Hewlett Packard Petro-IRD GC/FT-IR quantitation software. Materials. Pure compounds were purchased from Aldrich Chemical Co. and Wiley Organics. Calibration solutions were prepared in 100 mL volumetric flasks with n-heptane as solvent. Concentrations of the calibration solutions were analyte dependent, and the procedure is described for each below. Benzene. Into respective volumetric flasks were added 0.1,0.5, 1.0, 3.0, and 5.0 mL of benzene along with 10.0 mL of ISTD, 1,2dimethoxyethane while the weights of the benzene and ISTD were accurately recorded. The flasks were then diluted to the mark with solvent. Toluene. Into respective flasks were added 1.0, 3.0, 5.0, 10.0, 20.0, and 25.0 mL of toluene along with 10.0 mL of ISTD while the weights were accurately recorded. The flasks were then diluted to the mark with solvent. ~~
(8) Wilkins, C. L.; Giss, G. N.; White, R L.; Brissey, G. M.; Onyiriuka, E. C.
Anal. Chem. 1982,54, 2260-2264. (9) Laude, D. A; Brissey, G. M.; Ijames, C. Anal. Chem. 1984,56,1163-1168.
F.;Brown, R
S.: Wilkins, C. L.
2016 Analytical Chemistry, Vol. 67, No. 13, July 1, 1995
I b lir (Inl".)
t4
kb
7b
Figure 1. GCIFT-IR selective wavelength chromatograms showing the aromatics found in gasoline: (top) 0-40 min; (bottom) 40-75 min.
C2-Benzenes. Into respective flasks were added 0.5, 1.0, 3.0, 5.0, and 10.0 mL of each compound along with 10.0 mL ISTD while the weights were accurately recorded. The flasks were then diluted to the mark with solvent. C3-Benzenes, WBenzenes, C5-Benzenes, Indan, and Naphthalenes. Into respective flasks were added 0.1, 0.5, 1.0, 3.0, and 5.0 g of each compound along with 10.0 mL of ISTD while the weights were accurately recorded. The flasks were then diluted to the mark with solvent. These solutions bracketed the concentration ranges over which the analytes occurred in gasolines. The manner in which the compounds were combined in solutions was not critical for calibration as long as no compound contained an impurity which
Table 2. GC/FT.IR Selectlvlties of Coelutlng Aromatic Compounds
compounds
selectivity
m-xylene over p-xylene b-xylene over m-xylene lethyl-3-methylbenzene over l-ethyl-4-methylbenzene lethyl4methylbenzene over lethyl-3-methylbenzene 1,Miethylbenzeneover n-butylbenzene n-butylbenzene over 1,4diethylbenzene
'lo00 14 7 35 'lo00 'lo00
Table 3. Preclslon and Accuracy Data.
compound
actual
benzene 1.75 toluene 11.19 4.31 ethylbenzene m-xylene 4.31 4.28 p-xylene o-xylene 4.37 11.50 isopropylbenzene n-propylbenzene 11.72 11.76 leth yl-Smethylbenzene 12.01 lethyl-4-methylbenzene 34.53 1,3,5trimethylbenzene 11.98 lethyl-2-methylbenzene 34.91 1,2,4trimethylbenzene 35.17 1,2,3-trimethylbenzene 11.61 indan 11.63 1,Sdiethylbenzene 11.56 1,4diethylbenzene 11.44 n-butylbenzene 11.91 1,2-diethylbenzene 1,2,4,5tetramethylbenzene 11.59 1,2,3,5tetramethylbenzene 11.72 1,2,3,4tetramethylbenzene 11.62 11.20 naphthalene 11.67 2-methylnaphthalene 11.59 1-methylnaphthalene
determined
RSD
accuracy
1.76 11.40 4.36 4.37 4.27 4.42 11.55 11.80 11.85 12.01 35.01 12.10 35.41 35.76 11.79 11.48 11.51 11.37 11.86 11.60 11.70 11.66 11.25 11.77 11.79
0.6 1.1 1.5 2.1 0.2 0.7 0.7 0.8 1.0 0.9 0.7 0.9 0.8 0.5 0.7 1.1 1.1 1.0 1.1 0.9 1.1 0.9 1.0 1.1 1.1
0.6 1.9 1.2 1.4 0.2 1.1 0.4 0.7 0.8 0.0 1.4 1.0 1.4 1.7 1.6 1.3 0.5 0.6 0.4 0.1 0.2 0.4 0.5 0.9 1.7
I
i
a
Results are in mg/mL. n = 10 over 48 h. Average accuracy, 0.9%. Average RSD,0.9%.
*. would interfere with another aromatic in the same solution. In
this case, BTEX were added together in one set of five solutions, the C3-benzenes and indan were added together in another set of five solutions, and the other compounds were added together in another set of five solutions. Each of these solutions contained 10.0 mL (-8.6 g) of ISTD. Calibration curves were the plots of the respective selective wavelength (SW) chromatographic peak area of the analyte divided by the SW chromatographic peak area of the ISTD vs the weight of the analyte in the calibration solution divided by the weight of the ISTD in the solution (Le., &/Ai vs W J K , where a is analyte and i is ISTD). All calibration and quantitation routines were performed with the commercial software provided with the FT-IR spectrometer. Gasolines were prepared by adding 1.0 mL of ISTD to 9.0 mL of sample while the weights were accurately recorded. After GC/ FT-IR data acquisition, the weights of the analytes were determined from the calibration curves. Weight % was calculated by dividing the determined weight of an analyte by the sample weight and multiplying by 100. GC/MS Aualyses. The GC/MS analyses were performed according to the method outlined in ref 3. Gasoline samples were prepared by accurately weighing 0.10 g each of four internal
1
n
n
I
I
I(
Y
Figure 2. GC/FT-IR chromatogramsshowing the SWC frequencies used for the estimated aromatics: (top) 40-50 min; (middle) 50-60 min; and (bottom) 60-70 min. Analytical Chemistry, Vol. 67, No. 13, July 1, 7995
2017
Table 4. Analysis of a Full Range Gasoline.
compound/class
GC/FT-IR
GC/MSb
MTBE‘ benzene toluene ethylbenzene m-xylene p-xylene m- p-xylene o-xylene isopropylbenzene n-propylbenzene lethyl-3-methylbenzene lethyl4methylbenzene 1,3,5trimethylbenzene lethyl-2-methylbenzene 1,2,4-trimethylbenzene 1,2,3-trimethylbenzene indan 1,3-diethylbenzene 1,Cdiethylbenzene n-butylbenzene 1,Cdiethyl + n-butylbenzene 1,2-diethylbenzene 1,2,4,5tetramethylbenzene 1,2,3,5tetramethylbenzene 1,2,3,4tetramethylbenzene naphthalene 2-methylnaphthalene 1-methylnaphthalene estimated C10+ aromatics total aromatics
10.7 f 0.1W 1.15 i 0.02e 8.41 f 0.14 2.35 f 0.07 3.61 f 0.12 1.50 f 0.01 5.11 i: 0.13 1.77 f 0.02 0.09 f 0.01 0.37 i 0.01 1.56 f 0.04 0.55 f 0.02 0.63 f 0.02 0.42 i: 0.01 1.84 f 0.05 0.33 5 0.01 0.15 f 0.01 0.09 f 0.01 0.25 f 0.01 0.21 f 0.01 0.46 f 0.02 0.04 f 0.01 0.09 i: 0.01 0.19 f 0.01 0.30 f 0.01 0.18 i 0.01 0.18 f 0.01 0.08 f 0.01 1.31 f 0.03 27.4
nfl 1.2@ 8.35 2.29 nr nr 5.12 1.72 0.05 0.38 1.46 0.58 0.66 0.41 1.63 0.45 0.19 0.10 nr nr 0.48 0.03 0.13 0.26 0.29 0.25 0.23 0.11 1.20 27.6
+
Results are in wt %. G U M S analyses were performed according to the method described in ref 3. SWC frequencies were 1205-1213 cm-1 as specified in ref 10. f 2 times the SD provided in ref 4. e f 2 times the SD listed in Table 3. ’Not reported. GC/MS precision not listed in ref 1 or 3.
standards,benzene-&, toluene&, ethylbenzene$lo, and naphthalene ds, into 10.00 g of a sample.
RESULTS AND DISCUSSION Note that, in Figure 1, 1-ethyl-%methylbenzenewas only
partially resolved from l-ethyl-4-methylbenzene, m-xylene was unresolved from pxylene, and 1,4-diethylbenzenewas unresolved from n-butylbenzene. Table 2 shows the selectivities of these compounds over the respective coeluting compounds. Selectivity was calculated by dividing the respective SW chromatographic peak area of the analyte at 10 mg/mL concentration by the area of the peak produced by 10 mg/mL of the coeluting compound after reconstruction at the analyte’s wavelength region. All calibration curves were linear. Precision and accuracy data are shown in Table 3. The average relative standard deviation was 0.9%,and the average accuracy was 0.9%. The detection limits The detection limits for BTEX have been presented previ~usly.~ for the other aromatics were -0.05 wt % (50 ng injected), with a signal to noise ratio of 10. It should be noted that although multiple perdeuterated internal standards have been used for GC/ FT-IR quantitation? only 1,2-dimethoxyethane,the same ISTD used to determine oxygenates by GC/FT-IR,4J0was needed for this work. This differed from the GC/MS approach, which required multiple internal standards.lI2 A major difficulty in analyzing petroleum streams is that there are usually many isomers present for which pure standards are (10)Proposed ASTM GC/FT-IR method for determining oxygenates and aromatics in reformulated gasolines. 2018 Analytical Chemistty, Vol. 67, No. 13, July 1, 1995
Table 5. GCFT-IR Analyses of the Same Gasoline Performed 6 Weeks Apart but with the Same Calibration Curves
compound/class benzene toluene ethylbenzene m-xylene @xylene o-xylene isopropylbenzene n-propylbenzene lethyl-3-methylbenzene lethyl-4methylbenzene 1,3,5trimethylbenzene lethyl-2-methylbenzene 1,2,4trimethylbenzene 1,2,3-trimethylbenzene indan 1,3diethylbenzene 1,4diethylbenzene n-butylbenzene 1,2diethylbenzene 1,2,4,5tetramethylbenzene 1,2,3,5tetramethylbenzene 1,2,3,4tetramethylbenzene naphthalene 2-methylnaphthalene 1-methylnaphthalene estd C10+ aromatics total aromatics
8/16/94
9/30/94
re1 diff (%)b
0.94 4.85 1.28 3.86 1.36 2.02 0.05 0.37 2.30 0.57 0.83 0.58 2.88 0.68 0.44 0.21 0.40 0.54 0.03 0.31 0.57 0.13 0.65 0.05 0.05 10.70 37.0
0.94 4.82 1.23 3.84 1.34 2.00 0.04 0.35 2.22 0.51 0.83 0.59 2.87 0.69 0.44 0.21 0.42 0.49 0.06 0.34 0.62 0.13 0.64 0.05 0.07 11.00 36.7
0.0 0.6 4.0 0.5 1.5 1.0 20.0 8.0 3.5 10.5 0.0 1.7 0.3 1.5 0.0 0.0 5.0 9.3 100.0 9.7 8.8 0.0 1.5 0.0 40.0 2.8 0.8
a Results are in wt %. Absolute value of (8/16/94 result - 9/30/ 94 result)/8/16/94 result x 100. Average relative difference, 8.6%.
difficult or impossible to obtain. This was the case for gasoline range aromatics with carbon numbers above nine, as can be readily seen in the 600-900 cm-I SWC in Figure 1. Some method of estimating these compounds must be employed such that, when this estimate is added to the sum of the levels of the known analytes, a total aromatics value is obtained. The GC/MS approach1q2used one C4-benzene isomer‘s calibration curves to estimate other C4-benzenes, one &benzene isomer for other C 5 benzenes, etc. In this work, 1,2,3,5tetramethylbenzene’s 844-852 cm-* calibration curve was used to estimate the aromatic compounds for which standards were unavailable. SWCs were generated at the estimated compounds’ respective wavelengths, and the respective amounts were calculated from the areas of the peaks and the 1,2,3,5tetramethylbenzenecurve’s slope and intercept. Figure 2 is a 600-900 cm-l SWC which shows the estimated compounds with the respective reconstruction frequencies labeled. These frequencies were obtained from examination of each compound‘s vapor phase infrared spectrum. The mass spectrum was also examined to ensure that the infrared absorbances were definitely due to an aromatic hydrocarbon. The sum of the estimated compounds’ results was the total estimated amount of C10+ aromatics. This was added to the sum of the determined amounts of analytes to provide the total amount of aromatics in the gasoline. Table 4 shows the GC/FT-IR results of the analysis of a full boiling range (C4-C12) gasoline compared to GC/MS results. There was good agreement between GC/FT-IR and GC/MS results for the measurement of the individual isomers as well as for the total. This agreement indicated that the SWC approach
for GC/FT-IR provided selectivity over the hydrocarbons, primarily paraffins, olefins, and naphthenes," which coeluted with the aromatics comparable to that of the reconstructed ion chromatograms (RIC) used by the GUMS approach? Note that GC/MS did not report methyl tert-butyl ether. Determination of ethers and alcohols in gasolines by GUMS may be possible, but the use of GC/MS for this application is not widespread. Since molecular ions are not very abundant in these compounds' electron impact spectra,12 chemical or some other mode of ionization may be needed to obtain good selectivity over the coeluting hydrocarbons. Also note that GUMS reported the coeluting pairs of m-xylenelp-xylene and 1,4diethylbenzene/nbutylbenzene as sums because the pairs were not spectroscopically distinguished. Using different ions for the RICs or chansing chromatographicconditions (such as a higher resolution column, a d8erent phase, etc.) may address this GC/MS limitation, but these improvements may also increase analysis time, decrease loading capacity/detection limits, or cause other problems avoided by the GC/FT-IR approach. (11) DiSanzo, F. P.; Giarrocco, V. J. 1. Chromatogr. Sci. 1988,26, 258. (12) Silverstein, R M.; Bassler, G. C. Spectrometric Identification of Organic Compounds, 2nd ed.; John Wiley & Sons, Inc.: New York, 1967.
GC/ET-IR calibration curves were found to be very stable, as can be seen in Table 5. A gasoline was analyzed twice over a 6 week period without recalibrating the instrument. Despite the time lapse, the results were essentially unchanged. The average relative difference was 8.6%, which included a number of analytes at
