Anal. Chem. 1980, 52, 2069-2072
column to confirm the identity of each component. T o emphasize the importance of confirmation, we ran the sample shown in Figure 3 on the SP-2100 capillary column and found that the concentration of 1,3,5-TCB and HCB were only one-third of the their concentration as determined on the Carbowax 20M column. T h e concentrations of all other components in the sample were in good agreement for both columns. For future analysis we intend to use the pentane extract preconcentration method because of its simplicity. Also, in practice we find that the resin column has a tendency to plug up when samples contain significant quantities of particulates. T h e resin column technique is useful for clear samples and our preliminary studies show it can be used to efficiently recover chlorobenzenes from air samples. Because no evaporation step is used in the technique and because good recoveries are achieved for many chlorinated volatiles, the resin column technique may prove useful for the combined determination of chlorobenzenes and chlorinated volatiles such as trihalomethanes, should a suitable chromatographic separation procedure be developed.
2089
LITERATURE CITED (1) "Water-Related Environmental Fate of 129 Priorhy Pollutants", Report EPA-440/4-79-029b, United States Environmental Protection Agency, 1979. (2) Niimi, A. J. Bull. Environ. Contam. Toxlcol. 1979, 23. 20-24. (3) Fischer, Annegret: Slemrova, Java Vom Wasser 1078, 51, 33-46. (4) Schwarzenbach, Rene P., MdnarXubica, Eva; Giger, Walter; Wakehem, Stuart G. Environ. Sci. Technol. 1979, 13, 1367-1373. (5) "Organic Chemicals in Drinking Water, Summary of the National Organics Monitoring Survey" United States Environmental Protection Agency, Criteria and Standards Division, Office of Drinking Water: Washington, D.C., 1976. (6) Junk, G. A.; Richard, J. J.; Griesen, M. D.; Witiak, D.: W M k , J. L.; Arguello, M. D.; Vick. R.; Svec, H. J.; Fritz, J. S.; Calder. G. V. J . Chromatogr. 1974, 99, 745-762. (7) Renberg, Lars Anal. Chem. 1978, 50, 1836-1638. ( 8 ) Tateda, Akira; Fritz. James S. J . Chromatogr. 1978, 152, 329-340. (9) Richard, John J.; Junk, Gregor A. J. Am. Water Works Assac. 1977, 69, 62-64. (10) Trussel, Albert R.; Umphres, Mark D.: Leong, Lawrence Y. C.; Trussell, R. Rhodes. In "Water Chlorlnation Environmental Impact and Health Effects"; Jolley, R. L., Gorchev, H., Hamiiton, D. H., Eds.; Ann Arbor Science Publishers: Ann Arbor, Mi, 1978; Vol. 2, pp 543-5533, (11) Glaze. William H.; Peyton, Gary G.; Rawley, Rlchard. Envlron. Scl. Technol. 1977, 1 1 . 685-690.
RECEIVED for review June 17,1980. Accepted August 11,1980.
Estimation of the Vapor Pressure of Petroleum Distillate Fractions from Gas Chromatographic Data F. T. Eggertsen," N. R. Nygard, and L. D. Nickoley Cal/Ink Division, Flint Ink Corporation. 1404 Fourth Street, Berkeley, California 94 7 10
The vapor pressure of petroleum distillate fractions is estimated conveniently by applying ideal gas and solution laws to gas chromatographic data. The vapor pressure is computed as a sunmation of the partial pressures for carbon number groups in the chromatogram, each partial pressure being derived as the product of mole fraction, determined from the chromatogram, and saturation pressure, Calculated using the Antoine equation. A rapid method of computation with a programmable calculator was used. The method assumes approximate conformance to Raoult's and Dalton's laws relating to partial pressures from a solution. The results generally agree well with values obtained manometrically. Advantages of the method are the general availability of suitable gas chromatographic equipment, convenience and speed of the procedure and calculations, applicability over a wide range of volatility, freedom from temperature equilibration problems, and small sample requirement.
Vapor pressure is often an important specification for petroleum distillates and formulations containing them, not only for quality control purposes but also for ensuring conformance with safety and environmental regulations. Such data are needed for calculating volatile emissions from industrial facilities. A number of methods are available for measuring vapor pressure manometrically; for example, the isoteniscope (I) or the well-known Reid (2) method published by the American Society for Testing and Materials (ASTM). In other ASTM methods the vapor pressure of a petroleum distillate fuel is derived from ASTM D 86 distillation data using a graphical correlation procedure (3),and the Reid vapor pressure of a gasoline can be estimated from the C4 and C5 0003-2700/80/0352-2069$01 .OO/O
contents as determined by gas chromatography (GC)(4). The classical gas saturation method depends on measuring the mass of volatiles carred away by a gas (5). T h e amount of volatiles in a saturated gas may be determined by GC analysis (6) or with a thermal analyzer employing a hydrogen flame ionization detector (7). Also, an effusion method has been developed on the basis of measuring mass flow through an orifice (8). In the present study GC simulated distillation analyses (9) are performed in a manner to produce data from which vapor pressures can be calculated. The method was developed primarily for application to oils of low volatility, such as heavy ink oils, for which vapor pressures were not available and were difficult to measure by conventional means. T h e technique also proved to be useful for low-boiling petroleum distillate fractions, including gasolines.
EXPERIMENTAL SECTION Apparatus. A Varian 1200 gas chromatographequipped with hydrogen flame ionization detection was used. The separating columns were l/g in. coiled stainless steel, 1and 10 ft long, packed with 5% SE30 silicone on H/P Chrommrb W, 80-100 mesh. The gas flows were as follows: helium carrier, 28 mL/min; hydrogen, 35 mL/min; and air, 300 mL/min. The chromatograms were recorded on a Varian Model 20 strip chart recorder, and the peak areas were integrated with a planimeter or a Columbia Scientific Industries Supergrator-3Aprogrammable computing integrator. Reagents. Reagent quality petroleum ether 20-40 "C was used as a solvent. Chromatographic quality n-paraffin liquids were used for calibration purposes, including c&9, C12,C14,and Cle Also a paraffin wax blend containing n-paraffin up to about Cw was employed as a calibration standard. Sample Preparation. Make a solution containing 0.2-0.3 g of sample in 2 mL of petroleum ether (or other noninterfering solvent). Add 1 drop of a pure n-paraffin (C9,Clz, C14,or CIS) as a marker which is selected not to interfere with the sample 1980 American Chemical Society
2070
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
5
0
10
20
I5
ILmq rn,""t.l
Flgure 1. Chromatogram of a heavy oil: sample 3, Table V; 1-ft GC column, 95-285 OC.
Table I. Vapor Pressure of a Heavy Oila vapor pressure at 104 "C, torr partial re1 re1 moles mole satn pressure area (A/(C no.)) fraction pressure Cno. (A) X 10' (F) (PI (FP) 0.010 1.900 0.019 15 0.4 2.67 0.886 0.019 6.25 0.022 16 1.0 17 2.1 12.35 0.044 0.445 0.020 18 3.9 21.67 0.078 0.220 0.017 19 5.6 29.5 0.106 0.104 0.011 20 6.4 32.0 0.115 0.0349 0.004 21 7.5 35.7 0.128 0.0203 0.003 6.9 31.4 0.113 0.00950 0.001 22 23 6.2 27.0 0.097 0.00445 0.000 24 5.5 22.9 0.082 0.00209 0.000 0.00098 0.000 25 4.0 16.0 0.058 0.000 0.046 0.00046 26 3.3 12.7 27 2.8 0.00021 0.000 10.4 0.037 17.7 31b 5.5 0.00001 0.000 0.064 total 278.2 1.000 0.094 Sample 3, Table V ; GC curve, Figure 1. 28-34.
10
5
20
15 flrn.,
mi"",.,
Flgure 2. Chromatogram of a light oil: sample 5, Table V; 1 0 4 GC column, 65-250 OC. 50
5
0
10
15
time, minuter
Figure 3. Chromatogram of a solvent: sample 9, table V; 1 0 4 GC column, 65-250 OC. "C7
C no.
Table 11. Vapor Pressure of a Light Oila
C no. 13
1.3
14 15 16 17 18
2.9 6.9 10.6 12.2 7.8 4.2
19
20 total a
re1 area
4.0
mole fraction
vapor pressure at 104 "C, torr satn partial pressure pressure
(F)
(PI
(FP)
0.033 0.069 0.153 0.221 0.239 0.144 0.074 0.067 1.000
8.55 4.05 1.90 0.886 0.445 0.220
0.282 0.279 0.291 0.196 0.106 0.032 0.008 0.002 1.196
0.104
0.035
Sample 5, Table V; GC curve, Figure 2.
chromatogram. Weigh the marker if it is desired to use the marker also as an internal standard to check recovery of the sample. Alternatively, samples of low viscosity may be injected neat. Procedure. For heavy distillates from about Cl2 to Ca use the 1-ft GC column with the flow conditions as already specified and with the initial temperature at 95 OC. Inject 1-5 pL of sample and heat the column at 10 OC/min to 285 "C to obtain a chromatogram like Figure 1. For light fractions (C4-C2,) use the 10-ft column in order to achieve sufficient separation of the low-boiling components. Also use a lower starting temperature, e.g., 50 OC for samples containing C6s and 40 "C if C4s and C5s are present. Program heat at 10 OC/min to 250 "C. Typical chromatograms are shown in Figures 2-4. Obtain chromatograms for suitable n-paraffin standards under the same conditions as for the samples in order to identify the n-paraffin peaks in the sample chromatograms. Calculation of Vapor Pressures. The calculations, examples of which are given in Table I and in less detail in Tables 11-IV are made as follows.
0
I
2
3
4
5
iime. minutea
Figure 4. Chromatogram of a l w t solvent: sample 10, Table V; 1 0 4
GC column, 50-250 OC. Table 111. Vapor Pressure of a Solvent vapor pressure at 20 "C, torr mole satn partial re1 fraction pressure pressure C no. area (F) (P) (FP) C8 0.2 0.005 11.0 0.0 57 cyclic-C, 1.5 0.039 7.4 0.290 c 9 10 8 0.248 3.1 0.765 20.1 0.414 0.90 0.374 c,, c,, 12.5 0.234 0.26 0.061 c,, 3.5 0.060 0.08 0.005 total 1.000 1.552 a
SamDle 9. Table V;GC curve. Figure 3.
Divide the chromatogram into areas approximately by carbon number, as shown in the figures, with the boundaries of these areas located midway between the retention times of adjacent n-
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
Table IV. Vapor Pressure of a Light Solventa
C no. POUP br-C, n-C, cyclic-C, br-C, n-C, cyclic-C, (1) (2)
br-C, n-C, total a
re1 area
vapor pressure at 20 “C, torr mole satn partial fraction pressure pressure (F) (PI (Fa
1.0 4.3 3.0 11.0 8.9 6.0 1.3 2.0 0.5
0.030 0.128 0.089 0.280 0.227 0.153 0.037 0.045 0.011 1.000
150 121.5 96 53 35.6 28.0 21.5 15.5 10.5
4.50 15.55 8.54 14.84 8.08 4.28 0.80 0.70 0.12 57.41
Sample 10, Table V; GC curve, Figure 4.
paraffins. Measure the area of the segments with a planimeter or more conveniently with an automatic integrator. Divide each area by the carbon number of the corresponding n-paraffin and normalize to a total of unity. The numbers obtained represent the mole fractions of the various carbon number groups. To calculate partial pressure at a given temperature, multiply each mole fraction by the saturation pressure, generally taken to be that for the n-paraffin. A summation of these partial pressures so calculated is the total vapor pressure. Compute the saturation pressures using the Antoine equation B In P = A - C+T where A, B , and C are constants and T = K. The constants, taken from Reid et al. (IO) were listed only up to C m The values from Czl through CB1were obtained by extrapolation through linear regression of the values from Clo to Czp. For carbon numbers C8 and lower, divide the C number segments into subgroups as shown in Figures 3 and 4 and Tables I11 and IV. The subgroups for components emerging before and after the n-paraffin consist mainly of branched and cyclic hydrocarbon types, respectively, and are so designated (“br” and “cyclic”). The saturation pressure corresponding to each peak or the center of the segment for each subgroup is determined from a plot of retention time vs. log vapor pressure of the n-paraffins. In making calculations for materials containing hydrocarbons above Cn,certain simplifying approximations can be made. The mole fraction of the heavy components may be computed in groups of two or more carbon numbers with no significant effect on the
accuracy. Also the partial pressures above about CZzmay be disregarded because their contribution at 104 “C and lower is small (less than 0.01 torr). Simplified Computation with a Programmable Calculator. Because of its low cost and wide availability, one of the popular types of programmable calculator was used here for processing the GC data. Obviously, if more powerful and efficient computation equipment is at hand a much more time-saving and comprehensive program could be used. The computations involve four steps, all of which are readily performed with a programmable calculator. To do this a program has been developed for the Texas Instruments Programmable 58/59 calculators as follows: (1) Compute from the Antoine equation the estimated vapor pressure at the desired temperature for each carbon number group in the sample. (2) Compute the amounts in moles and the totalmoles. (3) Compute mole fractions. (4) Compute the partial pressure for each carbon number group and collect the sum to obtain total vapor pressure. The 90-step program can easily be entered manually into a TI Programmable 58 or a magnetic card prepared for entry into a TI Programmable 59. For computations at a given temperature (saturation pressure fixed), the program reduces to 60 steps and requires about 10 min for a typical calculation. A TI 59 program for calculating vapor pressures using the Antoine equation has been reported previously ( 2 1 ) .
RESULTS AND DISCUSSION Results are given in Table V for products covering a wide range of volatilities and including oils, kerosene range solvents, and mineral spirits. Also shown are other properties from available specification data, including vapor pressures obtained manometrically, boiling range, and flash point. T h e vapor pressures for the heavy oils were computed a t 104 “C because interim local regulations required data at that temperature. Specification vapor pressures for these products were not available. The vapor pressures of the light oils and solvents were calculated at other temperatures, at least one of which allows a comparison with specification values. The agreement is quite satisfactory, probably within the usual variability range for these commercial products. A further evidence of the reliability of the measurements is that the vapor pressure generally decreases with increasing flash point, as is t o be expected. The precision of the method was determined for two of the samples listed in Table V. T h e values for no. 5 and no. 10 were 1.8% and 1.5%, respectively, for sets of four. The accuracy of the method was tested by direct comparison with the Reid method employing the same samples. In order
Table V. Vapor Pressure Comparisons-GC vs. Specification Values vapor pressure, torr sample
boiling range, “ C
flash point, “C
T,“ C
GC
104 104 104 104
0.05 0.04
104 20 66 104 66 104 20 104 20 37.8 20 37.8
1.2 0.012 0.46 4.3 0.41
spec values (manometric)
Heavy Oils 1. Ink 2. Ink 3. Ink 4. Ink
Oil Oil Oil Oil
1 0 (Chevron) 6013 (Bray Oil Co.) 4313 (Bray Oil Co.) 2302 (Witco)
329-394 > 260 >310
204 191 177 168
0.09 0.21
Light Oils and Solvents 5. 535 (Magie Oil Co.) 6. 47 (Magie Oil Co.) 7. 470 (Magie Oil Co.)
274-313 242-278 243-268
141 104
110
8. 440 (Magie Oil Co.)
232-258
99
9. 350H (Chevron)
156-207
38
77-102
