ANALYST, JANUARY 1989, VOL. 114 41 Supercritical Fluid Chromatography of Coal-derived Polycyclic Aromatic Hydrocarbons on Packed Columns Ian K. Barker, Jacob P. Kithinji, Keith D. Bartle," Anthony A. Clifford, Mark W. Raynor, Gavin F. Shilstone and Peter A. Halford-Maw Department of Physical Chemistry, University of Leeds, Leeds LS2 9JT, UK The supercritical fluid chromatographic behaviour of coal-derived polycyclic aromatic hydrocarbons using modified HPLC equipment with carbon dioxide ( C 0 2 ) and methanol - CO2 as the mobile phases is described. Factors affecting the separation and retention of these compounds are considered. Keywords: Supercritical fluid chromatography; packed columns; polycyclic aromatic hydrocarbons The tumorigenicity of coal-derived oils has been shown to be due to the presence of polycyclic aromatic hydrocarbons (PAHs)' and there is much current interest in devising rapid analytical methods for PAHs that can be applied to coal processing and combustion products.Coal oils can be screened rapidly for the presence of groups of PAHs by 1 electrochemical methods,2 but complementary separation procedures, which allow more detailed identification, are necessary, as the tumorigenic activities of PAH isomers vary widely. 1 Although high-performance liquid chromatography (HPLC) is commonly used for the analysis of PAH mix- tures,'-s supercritical fluid chromatography (SFC) has a number of advantages over HPLC when the same columns are employed. Above its critical point, a substance such as carbon dioxide (CO,) has properties that make its use as a chromato- graphic mobile phase very favourable, particularly in terms of faster analysis times.Solute diffusion coefficients in super- critical fluids are considerably greater than in liquids; the resulting mass transfer coefficients lead to the generation of considerably larger numbers of theoretical plates per unit time in SFC compared with HPLC. Further, the reduced viscosities of supercritical fluids allow greater mobile phase velocities.6 The separation of PAH standards using packed columns has been described by Gere et al.7 and the possibility of separating coal-derived PAHs by SFC on amino and octadecylsilane (ODS) columns has been reported.8 This paper describes the construction of a packed column SFC system. The retention characteristics of PAHs were investigated using C 0 2 and modified C 0 2 and compared with those obtained by HPLC and GC.The rapid determination of PAHs in a coal tar is demonstrated. Experimental Apparatus A diagram of the supercritical fluid chromatograph con- structed mainly from standard HPLC components is shown in Fig. 1. Liquid C02 (British Oxygen) or C 0 2 - methanol (95 + 5 ) (Electrochem) is supplied from the cylinder dip tube to a Varian 8500 syringe pump through A in stainless-steel tubing fitted with a 5-pm in-line filter. The pump is cooled by circulating ethanol at -2 "C through a 10-m long copper coil wound round the cylinder head. The pump outlet is connected via stainless-steel tubing to a Rheodyne 7010 or 7125 injection valve with 10- or 20-pl sample loops and mounted in a - Rheodyne (7125) .-- ODS 2 column (25 x 4.5 mm i.d.1 Cecil UV detector High-pressure Capillary restrictor cell (>300 bar) in water-bath I Brookes flowmeter ( ~ 8 .8 I m-1) Galaxy computer Fig. 1. Schematic diagram of the supercritical fluid chromatograph * To whom correspondence should be addressed.42 ANALYST, JANUARY 1989. VOL. 114 chromatographic oven (Dupont) maintained above the critical temperature of C 0 2 (31 "C). A 1-m length of the connecting tube is located in the oven to ensure pre-heating. Standard HPLC columns (10 or 25 cm x 4.5 mm i.d.) are joined to the injector and to a high-pressure cell mounted in a Cecil ultraviolet (UV) spectrophotometer located immediately below the oven. The cell (Fig. 2) has a path length of 10 mm and a volume of 8 p1 and is similar to that described by Hewlett-Packardg; polished quartz windows, 5 mm in diameter and 3 mm thick, were held in place by metal end pieces and 0.007 in PTFE ring seals.The outlet of the UV cell was connected to a restrictor to maintain supercritical (>75 bar) conditions. The restrictor was made from either 0.010 or 0.006 in i.d., in o.d. stainless-steel tubing held at 40 "C in a water-bath or a Tescom back-pressure regulator fitted with Buna N O-rings. All the connecting tubing had an internal diameter of 0.010 in and an outer diameter of in. Operation of the Chromatograph Complete filling of the pump cylinder is readily achieved if the pump head is cooled to 0 "C. The system can be operated under the flow control of the pump or under pressure control by means of a pressure transducer and microcomputer (Gemini) interfaced to the pump as described by van Leuten and Rothman.1 0 Pressure programming is achieved by varying the inlet pressure by means of a programme written in BASIC and ramp rates of up to 50 bar min-1 can be achieved. Samples were injected as solutions containing 3-5 g 1 - 1 of PAHs in either dichloromethane or methanol. Carbon dioxide is virtually transparent in the UV region above 190 nm and so (a) Inlet 3.5 crn + , \ Outlet Quartz windows (5 x 3 r n r n ) A / \ / \ \ / / / \ \ \ \ \ 10 rnrn \ \ \ \ / / \ \ \ / \ \ ' 4 PTFE seals Fig. 2. and ( b ) cross-section of the detector cell Diagrams of the high-pressure UV flow cell. ( u ) Detector cell there are few restrictions on the choice of wavelength when using either UV or fluorescence detection.Capillary tubing restrictors were less easily blocked than the back-pressure regulator. Samples Coke-oven tar distillation fractions (anthracene oils) were supplied by British Coal and gasifier tars and coal tar pitches by British Gas. Samples were stirred with HPLC grade dichloromethane or methanol and the solutions filtered before injection. Results and Discussion Injection The best results were achieved by the injection of 10-pl sample volumes. A 20-pl loop gave rise to peak broadening and splitting, particularly at low pressures. The effect was most marked for dichloromethane and toluene solutions and was attributed to an effect well known in HPLC": if a sample is injected in a large volume of solution in a solvent that is stronger than the mobile phase, then the resulting peaks may be distorted.Reducing the sample size generally led to symmetrical peaks. Column Packing The retention of PAHs was investigated for a range of different (standard) HPLC column packings with C 0 2 as the mobile phase (Table 1). Very rapid elution was observed from silica, generally with poor peak shapes which were attributed to adsorption on the active sites (residual silanol groups) on the silica. The retention times were longer on silica modified with aminopropyl groups (normal phase), but poor peak shapes were observed for late eluting compounds; this effect was attributed to the precipitation of PAHs caused by their poor solubility in the mobile phase, particularly at the lower pressures at the end of the column.The best chromatograms were obtained on octadecylsilane (ODS) modified silica (reversed-phase) columns, although the retention of PAHs was similar to that found on diol columns. A typical chromatogram of a number of standard compounds run on an ODS column is shown in Fig. 3. Effect of Modifier, Temperature and Pressure on Retention The addition of methanol to the C 0 2 mobile phase substan- tially reduced the retention times of the PAEls on both the diol and ODS columns (Table 1). For polar solutes this effect is commonly attributed to competition between the modifier and the solute for active sites on the stationary phase.]? However, other effects that can intluence retention are the inter m o 1 e c u 1 a r attraction between the met h a n o 1 and so 1 u t e molecules and the increased solvating power due to the presence of methanol.13 For PAHs, however, the reduced retention is probably caused principally by the last effect.As has been observed previously in SFC, increasing the column temperature at constant pressure results in an increase in retention because of the increase in the free volume of the mobile phase which leads to a reduction in the solubility and a shift in partition in favour of the stationary phase. A 20 "C rise in temperature from 34 "C results i n a 100% increase in the capacity factor, k ' , for chrysene. A further increase in temperature causes an increase in the vapour pressure of the PAHs and, consequently, there is an increase in the concen- tration in the mobile phase, which reduces the value of k ' .A graph of In k' versus the reciprocal of the temperature (1173 shows the typical turnover (Fig. 4). These graphs do not intersect for the PAHs so no selectivity can be induced by changing the operating temperature. 14 This dependence of retention on temperature in SFC has recently been explainedANALYST. JANUARY 1989, VOL. 114 43 Table 1. Retention times of PAHs at 45 "C on various 25-cm long packed columns with supercritical COz and modified C02 mobile phases Retention timeimin Mobile Pressure/ Column phase bar packing Toluene CO, . . . . 148 Silica 0.53 Aminopropyl 0.65 Diol 0.65 Octadecylsilane 0.50 CO2 - MeOH (95 + 5 ) . . 148 Silica - C 0 2 . . . . 215 Aminopropyl - Octadecylsilane - Diol - D i d - Naph- thalene 0.70 1.30 1.10 0.90 0.45 0.73 0.80 0.62 0.55 Fluorene 0.97 2.30 1.70 1.55 0.56 0.84 1.25 0.83 0.80 Phen- anthrene 1.22 4.70 2.65 2.30 0.59 1.06 2.40 1 .00 1.10 Pyrene 1.32 10.70 4.90 4.90 0.68 1.45 5.20 1.98 2.05 Chrysene 1.82 20.00 8.10 8.10 0.83 1.80 9.00 2.98 3.05 ! 5 7 3, 0 2.0 4.0 6.0 Ti me/m in Fig.3. Chromatogram of a standard mixture of 1) naphthalene; (2) fluorene; (3) phenanthrene; (4) fluoranthene; (5(, pyrene; (6) benz- [alanthracene; and (7) chrysene. Column: ODS, 25 cm X 4.6 mm i.d. packed with 5-pm particles. Inlet pressure, 212 bar; temperature, 40 "C: CO, flow-rate. 2.5 I min-1 at STP. Restrictor: 2 m x 0.25 mm i.d. Detector wavelength. 254 nm 1.6 0.8 t -J 0 2.6 2.8 3.0 3.2 IO~T-~IK-I Fig. 4. Graphs of In k' versus 1/T.(0) Pyrene; (A) phenanthrene; (0) fluorene; and (0) naphthalene. ODS column; inlet pressure, 148 bar. Other conditions as in Fig. 3 quantitativelyls: the retention is dependent on the fugacity coefficient (a) of the solute in the supercritical mobile phase 7.04 c 5.42 E .- 1 .- E" 4- 3.80 0 C (u (u .- 4- 4- a 2.18 n u ~ r L 0.55 155 174 194 215 Column inlet pressure/bar Fig. 5. Plots of the retention time of (0) pyrene; ( A ) phenanthrene; (0) fluorene; and (0) naphthalene as a function of the CO, inlet pressure on an ODS column at 40 "C. Other conditions as in Fig. 3 and on the density (p) of the mobile phase. Hence a linear relationship is observed when the total contribution of these terms, i.e., [ln(k') + In(@) - ln(p)], is plotted against the reciprocal of the temperature.The effect of pressure on retention is illustrated in Fig. 5 , which shows that increasing the pressure at constant tempera- ture results in a decrease in retention because of the increased density of the mobile phase, which in turn leads to increased solubility and a shift in partition in favour of the mobile phase. 16 The thermodynamics of this phenomenon have recently been studied by Yonker et al. 17 Comparison of SFC With HPLC and GC for the Separation of PAHs The retention behaviour of PAHs in SFC with C02 on reverged-phase columns is compared in Table 2 with that observed on similar columns in the HPLC mode18 and with retention on a non-polar stationary phase using GC.19 The retention of PAHs in SFC resembles that found in GC and gives smooth graphs when the logarithm of the capacity factor is plotted against either the relative molecular mass or the boiling point of the PAHs (Fig.6). The retention indices given in Table 2 show how the isomeric pairs phenanthrene - anthracene, benzo(b]fluoranthene - benzo[k]fluoranthene, benzo[a]pyrene - benzo[e]pyrene and dibenz[a,h]anthracene - picene either co-elute in SFC or have fairly similar retention, as in GC. There is little sign of the effect of molecular shape on retention in SFC, whereas in reversed-phase HPLC the length to breadth ratio exerts a 5trong influence and the above pairs of isomers are well separated.*() Alkyl derivatives are also well separated from their parent PAHs using SFC.44 0 6 6 . ANALYST, JANUARY 1989, VOL.114 Table 2. Comparison of the retention indices of PAHs in SFC and HPLC with octadecylsilane packings and in GC. The retention index, I , of a compound x is defined by the equation log R, + log R,l Z, = log R, + log R,, + 1 - log R,l where n and n + 1 are the bracketing standards naphthalene, phenanthrene, chrysene and picene SFC HPLC" GCt Mobile phase CH,CN CH,CN (SO+ (90+ -H20 -H20 co2 20) 10) H2 Stationary phase ODs-2 column Spheri- LiChro- 1 Naphthalene . . 2.00 1 -Methylnaph- thalene . . 2.14 Fluorene . . 2.43 Phenanthrene . . 3.00 Anthracene . . 3.00 2-Methylphen- anthrene . . 3.12 Fluoranthene . . 3.28 Benzo[b]- fluorene . . 3.40 Pyrene . . . . 3.48 Benz[ a] - anthracene . . 3.88 Chrysene . . 4.00 4-Methyl- chrysene . . 4.11 Benzo[b]- fluoranthene 4.36 Benzo[k]- fluoranthene Benzo[e]pyrene Benzo[a]pyrene 4.67 Perylene . . . . 4.75 Dibenz[a,h]- anthracene . . 5.00 Picene . . . . 5.00 * Reference 18. t Reference 19. 2 2.00 2.51 3.00 3.00 3.38 3.58 3.92 4.00 4.38 4.43 4.56 4.59 5 .OO sorb ODS 2.00 2.56 3.00 3.00 3.41 3.63 3.92 4.00 4.43 4.67 5.00 sorb VYDAC RP- 18 2.00 2.73 3.00 3.14 3.42 3.62 4.00 4.00 4.40 4.48 4.40 4.63 4.78 5.00 201 TP 2.00 2.73 3.00 3.24 3.49 3.56 4.00 4.00 4.30 4.43 4.25 4.52 4.74 5.00 SE-52 2.00 2.20 2.70 3.00 3.01 3.20 3.45 3.69 3.52 3.99 4.00 4.20 4.43 4.44 4.52 4.54 4.57 4.96 5.00 0.37 i e 5 e 4 * 3 e2 0 1 0 1 0 148 245 343 440 537 Boiling pointPC Fig. 6. Plots of log R, against (0) the relative molecular mass and 0) the boilin point of 1) fluorene; (2) phenanthrene; (3) pyrene; [4) chrysene; 8) benzo[e\pyrene; and (6) picene on an ODS column.Other conditions as in Fig. 3 J I I I I I I I I I 1 I 0 1 2 3 4 5 6 7 8 9 Time/m in Fig. 7. Chromatograms of heavy anthracene oil. (a) SFC on a 25-cm ODS column; CO, inlet pressure, 180 bar. ( b ) GC on a 14-m capillary column. (1) Benzofluoranthenes; (2) benzopyrenes; and (3) pcrylene 0 2 4 6 n5 6 1 I 8 10 12 14 16 ' Ti me/mi n 8 Fig. 8. Pressure-programmed chromatogram (148-292 bar) of coal tar pitch on a 25-cm ODS column at 40 "C. UV detection wavelength, 254 nm. (1) Phenanthrene; (2) fluoranthene; (3) pyrene; (4) benz[a]- anthracene; ( 5 ) chrysene; (6) benzofluoranthenes; (7) benzopyrenes; (8) dibenzoanthracenes and phenanthrenes; and (9) coronene Rapid Analysis of Coal-derived Oils by SFC The rapid analysis of PAH mixtures on packed columns is made possible by the high mobile phase flow-rates, which are a consequence of the low viscosity of supercritical fluids and the high diffusivities of solutes, which in turn result in higher efficiencies per unit time than are possible in HPLC.Moreover, COz is particularly suitable because of its low critical temperature and excellent solvating power for non- polar solutes. These advantages, coupled with the separation of PAH isomers such as chrysene - benz[a]anthracene and the ben- zopyrenes, and of alkyl derivatives from the parent PAHs discussed above, make chromatography with supercritical C02 on reversed-phase columns an ideal method for the analysis of coal-derived oils. Fig. 7 shows the rapid separation of such a mixture, the resolution of the benzofluoranthene and benzopyrene isomers occurring within 9 min and being almostANALYST, JANUARY 1989.VOL. 114 45 1 5 0 1 2 3 4 5 6 7 8 9 1 0 Ti rneirnin Fig. 9. Chromatogram of coal tar pitch on an ODS column (10 cm X 4.6 rnm i.d.) at 40 “C. Inlet pressure, 148 bar; UV detection wavelength, 254 nm. Peak identification as in Fig. 8 as good as that obtained by capillary GC. Programming the column inlet pressure from 148 to 292 bar allows both optimisation of the separation of the lower PAHs (Fig. 8) and elution of the PAHs (Fig. 9) up to coronene (relative molecular mass 300) as the density and hence the solubility in CO? are increased. On short (10 cm) columns, isoconfertic (constant density) analyses extending up to coronene are possible in less than 10 min.For mixtures containing predomi- nantly alkylated derivatives, all the methyl and dimethyl compounds are well separated from the parent PAHs. The support of this work through grants from the Science and Engineering Research Council (SERC) and the British Gas Corporation, and through studentships awarded by the Royal Society of Chemistry, the British Council (J. P. K.) and the SERC (G. F. S.) is gratefully acknowledged. 1 . 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 1.5. 16. 17. 18. 19. 20. References Lee, M. L., Novotny, M. V., and Bartle, K. D.. “Analytical Chemistry of Polycyclic Aromatic Compounds,” Academic Press, New York. 1981. Bartle, K. D., Taylor, N., Pappin, A . , Wallace, S . , and Mills, D. G., Fuel, 1987, 66, 10.50.Bjorseth, A . , Editor, “Handbook of Polycyclic Aromatic Hydrocarbons,” Marcel Dekker, New York, 1983. Wise, S. A., in Bjorseth, A . , and Ramdahl, T . , Editors, “Handbook of Polycyclic Aromatic Compounds ,” Volume 2, Marcel Dekker, New York, 1985, p. 183. Wise. S. A., Benner, A. B., Liu, H., and Byrd, D. G., Anal. Chem., 1988, 60, 630. Bartle, K. D . , Barker, I. K., Clifford, A. A , , Kithinji, J . P., Raynor, M. W., and Shilstone, G. F., Anal. Proc., 1987, 24, 299. Gere, D . R . , Board, R., and McManigill, D., Anal. Chem., 1982, 54, 736. Christensen, R. G., J . High Resolut. Chrornatogr. Chrornatogr. Comrnun., 1985, 8, 824. McManigill, D., Board, R . , and Gere. D . K., Publication No. 43-59.53-1647. Hewlett-Packard, Avondale, PA, 1982. van Leuten, F. T., and Rothman, L. D., Anal. Chem.. 1976. 48, 1430. Snyder, L. R., and Kirkiand, J . J . , Editors, “Introduction to Modern Liquid Chromatography,” Second Edition, Wiley, New York, 1979, p. 805. Levy, J . M., and Ritchey, W. M., J. Chromutogr. Sci., 1986, 24, 242. Raynor, M. W., Kithinji, J . P . , Barker, I. K., Bartle. K. D., and Wilson, I . , J. Chromatogr.. 1988, 436, 497. Chester, T. L., and Innnis, D. P., J . Nigh Resolut. Chromatogr. Chrornutogr. Cornmun., 1985, 8, 561. Bartle, K. D., Clifford, A. A . , Kithinji, J . P . , and Shilstone, G. F., J. Chern. SOC., Faraday Trans. I , in the press. Klesper, E., and Leydendecker, D . , Znt. Lab., 1986, 17. Yonker, C. R . , Gale, R . W.. and Smith, R . D., J . Phys. Chern., 1987. 91, 3333. Wise. S. A., Bennet, W. J . , and May, W. E.. in Bjorseth, A , , and Dennis, A. J . , Editors. “Polynuclear Aromatic Hydrocar- bons: Chemistry and Biological Effects,” Battelle Press, Columbus, OH, 1980, p. 791. Lee, M. L., Vassilaros. D . L., White, C . M., and Novotny, M., Anal. Chern., 1979, 51, 768. Wise, S. A.. Bennet, B. A , , Guenther, F. R . , and May, W. E., J . Chromatogr. Sci., 1981, 19, 457. Paper 8102823 K Received July 13th, 1988 Accepted August 5th, 1988