Anal. Chem. 2000, 72, 1301-1306
C60 and C70 HPLC Retention Reversal Study Using Organic Modifiers Yves Claude Guillaume,*,† Eric Peyrin,‡ and Catherine Grosset‡
Laboratoire de Chimie Analytique, Faculte´ de Medecine Pharmacie, Place Saint Jacques, 25030 Besanc¸ on Cedex, France and Faculte de Pharmacie, Domaine de la Merci, 38700 La Tronche Cedex, France
A novel equation (Guillaume Y. C. et al. Anal. Chem. 1998, 70, 608) modeling the weak polar solute retention in reversed-phase liquid chromatography (RPLC) was applied to fullerene molecules C60 and C70. In RPLC, with an organic modifier (OM) /water mobile phase, the fullerene cluster solvation energies were calculated for OM ) methanol, ethanol, propanol, butanol, and pentanol. An enthalpy-entropy compensation revealed that the type of interactions between fullerenes and the stationary phase was independent of both the fullerene and organic modifier structures. The energetics of OM and OM-water cluster exchange processes in the mobile phase were investigated in relation to the carbon atom number of the hydrophobic chain of the OM. Two linear correlations were found between the Gibbs free energy changes in the solvent exchange processes which confirmed that (i) a reversal elution order existed for C60 and C70 when methanol was changed into ethanol, propanol, butanol, pentanol and that (ii) the mobile phase was dominant in governing selectivity changes in nonpolar solutes. A number of recent reports have examined the role of the mobile phase in the retention process in reversed-phase liquid chromatography (RPLC). Nevertheless, it should be stressed that retention is a result of the chemical potential difference between the mobile and stationary phases and that most chromatographic studies, which are designed to show only the effects from one or the other, are not totally deconvoluted. Lochmuller and Hangac1 reviewed and studied the use of mobile-phase additives for selective interactions with solutes and selectivity adjustment. Carr’s group2 performed a study measuring diffusion coefficients of alkylbenzenes and alkylphenones in water/methanol and water/ acetonitrile mixtures. Miyabe and Takeuchi3 studied surface diffusion and heat of adsorption of small molecules. Guillaume et * Corresponding author: (phone) 333.81.66.55.46 (fax) 333.81.66.55.27 (email)
[email protected] † Faculte ´ de Medecine Pharmacie. ‡ Faculte ´ de Pharmacie. (1) Lochmuller, C. H.; Hangac, H. H. Crit. Rev. Anal. Chem. 1997, 27(1), 2748. (2) Li, J.; Carr, P. W. Anal. Chem. 1997, 69(13), 2550-2553. (3) Miyabe, K.; Takeuchi, S. Anal. Chem. 1997, 69(13), 2567-2574. 10.1021/ac990842k CCC: $19.00 Published on Web 02/11/2000
© 2000 American Chemical Society
al.4 used a chemometric methodology to study the separation of 10 benzodiazepines in methanol/water and acetonitrile/water mixtures. Nyredy5 published a thorough study of a solvent classification for liquid chromatography and discussed both solvent strength and solvent selectivity factors, on the basis of Snyder’s solvent groups. Vailaya and Horvath6 published the first original work on the solvophobic theory in many years and reexamined a large set of retention data with nonpolar and weakly polar eluites. These authors showed that this theory can be useful for the evaluation of physicochemical parameters associated with retention of hydrocarbon solutes and that the mobile phase is dominant in governing selectivity changes in nonpolar solutes. Fullerenes,7-14 having a closed carbon cage molecule Cn with only pentagons and hexagons, provide an excellent descriptive picture of nonpolar solute retention in RPLC. Various chromatographic techniques have been used to separate fullerenes.15-21 All the previously reported fullerene separations used poly (octadecylsiloxane) (ODS) as a stationary phase and a nonpolar solvent such as hexane or a gradient of hexane with methylene chloride as a (4) Guillaume, Y. C.; Cavalli, E. J.; Peyrin, E.; Guinchard, C. J. Liq. Chromatogr. Relat. Technol. 1997, 20(11), 1741-1756. (5) Nyiredy, S. In Chromatography; Kaiser, Ed.; Incom: Duesseldorf, Germany, 1997; pp 231-239. (6) Vailaya, A.; Horvath, C. J. Phys. Chem. 1997, 101, 1(30), 5875-5888. (7) Kroto, H. W.; Heath, J. R.; O’Brien, S. C.; Curl, R. F.; Smalley, R. E. Nature (London) 1985, 318, 162. (8) Wudl, F. Acc. Chem. Res. 1992, 25, 143. (9) Fowler, P. W.; Manolopoulos, D. E. An atlas of fullerenes; Clarendon: Oxford, UK, 1995. (10) Kroto, H. W. Nature (London) 1987, 329, 529. (11) Schamaltz, T. G.; Seitz, W. A.; Klein, D. J.; Hite, G. E. J. Am. Chem. Soc. 1988, 110, 113. (12) Achiba, Y.; Kikuchi, K.; Auihara, Y.; Wakabayashi, T.; Miyake, Y.; Kainosho, M. In The chemical physics of fullerenes 10 (and 5) years later; Andreoni, W., Ed.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1996; pp 138-139. (13) Manolopoulos, D. E. J. Chem. Soc., Faraday Trans. 1991, 87, 2861. (14) Fowler, P. W.; Manolopulos, D. E.; Baten, R. C. J. Chem. Soc., Faraday Trans. 1991, 87, 3103. (15) Taylor, R.; Hare, J. P.; Abdul-Sada, A. K.; Kroto, H. W. J. Chem. Soc., Chem. Commun. 1990, 20, 1423. (16) Ajie, H. J.; Alvarez, M. M.; Anz, S. J.; Beck, R. D.; Diedrich, F.; Fostiropoulos, K.; Huffman, D. R.; Ktratschmeir, W.; Rubin, Y.; Schriver, K. E.; Sensharma, D.; Whetten, R. L. J. Phys. Chem. 1990, 94, 8630. (17) Allemand, P. M.; Koch, A.; Wudl, F. J. Am. Chem. Soc. 1991, 113 (3), 1050. (18) Vassalo, A. M.; Palmisano, A. J.; Pang, L. S. J. Chem. Soc., Chem. Commun. 1992, 60, 1. (19) Jino, K.; Yamamoto, K.; Veda, T.; Nagashima, K.; Itoh, C.; Fetzer, C.; Bigg, W. R. J. Chromatogr. 1992, 594, 105. (20) Meier, M. S.; Selegue, J. P. J. Org. Chem. 1992, 57, 1924. (21) Diederich, F.; Whetten, R. L. Acc. Chem. Res. 1992, 25, 119
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000 1301
mobile phase. In this paper, the retention behavior of fullerene molecules in various organic modifier (OM)/water mixtures was studied. The water fraction varied from 0.07 to 0.22 and the OMs used were methanol, ethanol, propanol, butanol, and pentanol. Theory. The structure of an alcohol(A)/water(W) mixture depends on the hydrogen bonds22,23 between the bulk water and, respectively, the hydrophilic (eq 1) and the hydrophobic ends of the alcohol molecule (eq 2)
A + W h AW
(1)
A + λW h AWλ
(2)
In high-alcohol fractions, the presence of a 1:1 hydrogen bond alcohol/water fraction (eq 1) is highly predominant.24 Therefore, over the water fraction range, the fullerene molecule CX (X ) 60 or 70), in the mobile phase, is preferentially solvated by the free alcohol, A, and the alcohol/water clusters, AW. This solvation equilibrium is represented by:
A + CX h CXA
(3)
AW + CX h CXAW
(4)
In a previous study, a mathematical model was presented to describe, in high-alcohol fractions, the variation in the retention factor k′ of weak polar solutes.25 By applying this model to totally apolar molecules, such as fullerene, the following equation was obtained:
Ln k′CX ) ln KCX,S + ln $ - ln(1 + KCX,AξAdA + KCX,AWξAWdAW) (5)
Where (i) k′CX is the retention factor of the fullerene CX, (ii) KCX,S is the equilibrium constant for the CX transfer from a total aqueous phase25 to the RP18 stationary phase, (iii) KCX,A is the equilibrium constant for the CX solvation by free alcohol, A, (iv) KCX,AW is the equilibrium constant for the CX solvation by alcoholwater clusters, AW, (v) $ is the phase ratio, (vi) ξA, dA, and ξAW, dAW are, respectively, the volume fraction and molar density of free alcohol, A, and alcohol-water cluster, AW. Thermodynamic Considerations. ∆GoCX,S, ∆HoCX,S, ∆SoCX,S, ∆GoCX,A, ∆HoCX,A, ∆SoCX,A, ∆GoCX,AW, ∆HoCX,AW, ∆SoCX,AW are the Gibbs free energy, enthalpy, and entropy for the CX transfer from a total aqueous phase to the RP18 stationary phase, respectively, for the CX solvation by free alcohol and alcohol water cluster. These energies can be determined using the well-known thermodynamic equation:
Ln K ) -∆H°/RT + ∆S°/R
(6)
∆G° ) ∆H° - T∆So
(7)
ln K ) -∆G°/RT
(8)
( )
∂ ∆Go ∆Hο )- 2 ∂T T RT
(9)
Enthalpy-Entropy Compensation. Investigation of the enthalpy-entropy compensation temperature is a thermodynamic approach to the analysis of a chemical equilibrium of a K constant. Mathematically, enthalpy-entropy compensation can be expressed by the formula:26
∆Hoβ ) β∆So + ∆Goβ
(10)
where ∆Goβ is the Gibbs free energy of a physicochemical interaction at a compensation temperature β. According to eq 10, when enthalpy-entropy compensation is observed for a group of organic modifiers in a particular chemical interaction, all the organic modifiers have the same free energy, ∆Goβ, at a compensation temperature β. Therefore, if, enthalpy-entropy compensation is observed for the five organic modifiers, all will have the same type of interaction at the compensation temperature, β, although their temperature dependencies might differ. Application of eq 10 to the two chemical processes, i.e., the Cx solvation by (i) free alcohol A and (ii) alcohol-water clusters AW, gives:
∆HoCX,A ) β∆SoCX,A + ∆(GoCX,A)β
(11)
o o o ) β∆SCX,AW + ∆(GCX,AW )β ∆HCX,AW
(12)
Combining eqs 6 and 11 gives:
ln KCX,A ) ln(KCX,A)β ln(KCX,A)β ) -
∆HoCX,A 1 1 R T β
(
)
(13)
∆(GoCX,A)β Rβ
(14)
Combining eqs 6 and 12 gives:
ln KCX,AW ) ln(KCX,AW)β ln(KCX,AW)β ) -
o ∆HCX,AW 1 1 R T β
(
o ∆(GCX,AW )β Rβ
)
(15)
(16)
Equations 13 and 15 show, that if a plot of lnKCX,A and lnKCX,AW against -∆HoCX,A and -∆HoCX,AW, respectively, is linear then the type of interactions is identical for the five organic modifiers.
For an equilibrium physicochemical process of the K constant: (22) Katz, E. D.; Ogan, K.; Scott, R. P. W. Anal. Chem. 1989, 61, 349. (23) Dethlefessen, C.; Sorensen, P. G.; Hvidt, A. J. Solution. Chem. 1984, 13, 191 (24) Alam, M. K.; Callis, J. B. Anal. Chem. 1994, 66, 2293. (25) Guillaume, Y. C.; Peyrin, E.; Guinchard, C. Anal. Chem. 1998, 70, 608.
1302
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
EXPERIMENTAL SECTION Apparatus. The HPLC system consisted of an L7100 Hitachi pump (Merck, Nogent sur Marne, France) and a 7125 injection (26) Sander, L. C.; Field, L. R. Anal. Chem. 1980, 42, 2009.
Table 1. Measured Values for the Logarithms of the Fullerene Retention Factors at Different Water Fractions for T ) 40 °C and in a (A) Methanol/Water Mixture and a (B) Ethanol/Water Mixture water fraction (v/v)
Ln k′C60
Ln k′C70
0.07 0.08 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22
1.26 1.31 1.39 1.50 1.60 1.66 1.70 1.82 1.92 2.00 2.07 2.16 2.23 2.31 2.40 2.47
1.54 1.75 1.97 2.18 2.40 2.61 2.82 3.03 3.25 3.46 3.68 3.89 4.11 4.32 4.53 4.75
water fraction (v/v)
Ln k′C60
Ln k′C70
0.71 0.79 0.87 0.94 1.02 1.09 1.18 1.25 1.33 1.41 1.49 1.56 1.64 1.72 1.79 1.88
-0.51 -0.43 -0.35 -0.27 -0.19 -0.11 -0.03 0.04 0.12 0.20 0.28 0.36 0.44 0.52 0.60 0.68
(A)
Figure 1. The C60 (a) and C70 (b) molecular structures. For C70, carbon atoms were represented by their van der Waals sphere.
valve (Interchim Rheodyne, Montluc¸ on, France) fitted with a 20µL sample loop and an L4500 diode array detector (Merck, Nogent sur Marne, France). A Lichrocart 125 mm × 4 mm i.d. RP18 column (5-µm particle size with pore size of 100 Å) (Merck, Darmstadt, Germany) was used with a controlled temperature device in an interchim 701 oven. The diameter of the fullerene molecules is obviously inferior to the pore size (for example, the diameter of the C60 molecule is 7.1 Å)7 and, since the CX molecule and the free alcohol (or its clusters) are in rapid equilibrium,25 the size selectivity27,28 does not intervene. The mobile phase flow rate was fixed at 1 mL/min. Solvents and Samples. RPLC grade alcohol, i.e., methanol, ethanol, propanol, butanol, pentanol (Carlo Erba, Val de Reuil, France), was used without further purification. Water was obtained from an Elgastat option I water purification system (Odil, Talant, France) fitted with a reverse osmosis cartridge. The mobile phase used for the study was an alcohol/water mixture. The range of the water fraction (v/v) in this mixture was 0.07-0.22. Fullerenes C60 and C70 were obtained from Sigma Aldrich (Saint Quentin, France). The structures of these compounds are given in Figure 1. Each solute or a mixture of these was injected when the two peaks were well-resolved. Sodium nitrate was used as a dead-time marker. Temperature Studies. Compound retention factors were determined over the temperature range 20-45 °C for the calculation of the thermodynamic data. Six values were included in this range (20, 25, 30, 35, 40, 45 °C). Before each experiment, the mobile phase was prewarmed to the column temperature to decrease the possibility of a “cold-to-hot” temperature gradient, down the column, of several degrees. The chromatographic system was allowed to equilibrate at each temperature for at least 1 h prior to each experiment. To study this equilibration, the retention time of the compound C70 was measured every hour for 7 h and again after 22, 23, and 24 h. The maximum relative difference in the retention times of this compound between (27) Meier, M. S.; Selegue, J. P. J. Org. Chem. 1992, 57, 1924. (28) Gugel, A.; Mullen, K. J. Chromatogr. 1993, 628, 23.
(B) 0.07 0.08 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22
these different measurements was always 0.6%, making the chromatographic system sufficiently equilibrated for use after 1 h. RESULTS AND DISCUSSION Validation of the Solute Equilibrium Model. To obtain the constants KCX,S, KCX,A, and KCX,AW at 25 °C, the retention factor k′ values of C60 and C70 were determined for the five organic modifiers, i.e, methanol, ethanol, propanol, butanol, pentanol, and for a limited-variation range of water fractions, Φ (0.07-0.22), in the alcohol/water mixture. Sixteen Φ values were included in this range for each used alcohol, i.e., 0.07, 0.08, 0.09, 0.10, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19, 0.20, 0.21, 0.22 (Table 1). For a water fraction over 0.22, for methanol, Cx has a high retention time and its peak is very wide, making the measurement of the retention time difficult. All the experiments were repeated three times. The validation of the k′ values was less than 1%, indicating a high repeatability and good stability for the chromatographic system. Between the different mobile phases, the variation coefficients of the dead time were independent of the water fraction. The average dead time value was 1.08 min. The alcohol/water single Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1303
Table 2. Ln KCX,S Values for C60 and C70 for the Five Organic Modifiers with Standard Deviations (in Parentheses)
Figure 2. Correlation between the predicted (eq 5) and the experimental retention factors for the two fullerenes C60 and C70 and methanol as organic modifier. The slope is 1.03 with a correlation coefficient of 0.988 determined by linear regression.
association model25 was used to determine for each Φ value the corresponding fraction of the free water and alcohol-water clusters. Using a weighted nonlinear regression (WNLIN),25,29 the data were fitted to eq 5. After the WNLIN procedure, the calculated parameters could be used to estimate the k′ values at different water fractions. The correlation between predicted and experimental k′ values exhibited a slope equal to 1.03 with r2 > 0.94 (Figure 2). For example, the KCX,S values are given in Table 2. From these data, the following conclusions can be drawn: Between the five organic modifiers (OM), the variation coefficients of the K values obtained were KC60,S, showing that the affinity for the RP18 stationary phase was stronger for C70, i.e., for the more hydrophobic species and fullerene, which has the highest chromatographically accessible surface area. Thermodynamic Parameters. The previous experiments carried out at 25 °C were repeated at six other temperatures. The model parameters (eq 5) corresponding to each temperature and organic modifier were determined. The corresponding different (29) Bevington, P. R. Data reduction and error analysis for the physical sciences; McGraw-Hill: New York, 1969.
1304
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
organic modifier
Ln KC60,s
Ln KC70,s
methanol ethanol propanol butanol pentanol
18.4(0.04) 18.2(0.02) 18.3(0.03) 17.9(0.03) 18.0(0.02)
21.3(0.04) 21.2(0.03) 21.1(0.02) 20.9(0.03) 21.3(0.02)
Van’t Hoff plots (eq 8) for the different chemical processes weredetermined. The correlation coefficients for the linear fits were in excess of 0.983. The typical standard deviations of slope and intercept were, respectively, 0.006 and 0.03. When methanol was taken into consideration, the enthalpy and entropy of solvation for C60 by the free methanol were -9.1 kJ/mol and 26.1 J/mol/ K, respectively, and those same values corresponding to the methanol/water clusters were -7.8 kJ/mol and 20.3 J/mol/K, respectively. The values again agreed with those reported in the literature.25,30 Enthalpy, entropy, and Gibbs free energy changes for the fullerene molecule transfer from the mobile to the RP18 stationary phase: From the results obtained, it can be noted that both ∆HoCX,S and ∆SoCX,S were negative. Negative ∆SoCX,S indicates an increase in the order of the chromatographic system as the fullerene was transferred from the total aqueous phase to the RP18 stationary phase. Negative ∆HoCX,S indicates that it was energetically more favorable for the solute to be in the RP18 stationary phase. The enthalpy and entropy for C70 were lower than those for C60. This indicated that both the affinity for the nonpolar stationary phase and the chromatographic system order were stronger for C70, i.e., for more hydrophobic species. Enthalpy, entropy, and Gibbs free energy changes for the fullerene molecule solvation by alcohol and its clusters: ∆HoCX,A < ∆HoCX,AW, indicate that the fullerene molecule preferred to be solvated by the weak polar entity, i.e., the free alcohol. ∆SoCX,AW < ∆SoCX,A with positive values would seem contradictory for the apparent lower degree of freedom of the fullerene solvated by the free alcohol and the alcohol/water clusters. This phenoenon can be explained by a contribution of the hydrophobic interaction.31-33 It is also seen in the case of the two following equilibria:
C60 + C70AW h C70 + AWC60
(17)
C60 + C70A h C70 + AC60
(18)
The energetics of these solvent exchange processes can be investigated in relation to the carbon atom number of the hydrophobic chain of the alcohol molecule A. The equilibrium (30) Guillaume, Y. C.; Guinchard, C. Anal. Chem. 1997, 69, 183. (31) Horvath, C. S.; Melander, W. R. J. Chromatogr. 1976, 125, 129. (32) Melander, W. R.; Horvath, C. S. In High Performance Liquid Chromatography, advances and perspectives; Horvath, C. S., Ed.; Academic Press: New York, 1980; vol.2, p 113. (33) Horvath, CS.; Melander, W. R. Am. Lab. 1978, 17.
Figure 4. The C60 and C70 chromatogram in a (a) methanol/water mixture (92-8(v/v)) and a (b) ethanol/water mixture (78-22(v/v)). The mobile phase flow-rate was 1 mL/min, and the column temperature was 25 °C. Numbers above peaks refer to CX retention time in minutes.
Figure 3. Solvent exchange process free energy ∆Gex,A (A) and ∆Gex, AW (B) in relation to the carbon atom number of the alcohol hydrophobic chain.
constants, Kex, of these two solvent exchange processes were:
Kex,A ) Kex,AW )
KC60,A KC70,A
(19)
KC60,AW KC70,AW
(20)
Values of KCX,A and KCX,AW were obtained as described above. The values of the free energy changes ∆Gex,A and ∆Gex,AW corresponding to the two solvent exchange processes (eqs 17 and 18) were calculated from eq 8. Two linear correlations between ∆Gex and n were found (Figure 3).
∆Goex,A ) -6.6 + 4.5n
(21)
∆Goex,AW ) -7.8 + 5.3n
(22)
The correlation coefficients for the linear correlations in eqs 21 and 22 were, respectively, equal to 0.989 and 0.979. ∆Gex,AorAW < 0 for methanol and ∆Gex,AorAW > 0 for ethanol, propanol, butanol, pentanol. This result showed that when the carbon atom number of the hydrophobic chain of the alcohol molecule increased, both (i) the C70 solvation by free alcohol and alcohol/water clusters was enhanced and thus, (ii) the equilibria 17 and 18 were displaced toward the free C60 form which can interact more easily with the
Figure 5. Plots of Ln KCX,A against -∆H0CX,A (kJ/mol) for X ) 60 (A) and X ) 70 (B).
RP18 stationary phase than the C70 molecule can. This result confirmed the reversed elution order observed for C60 and C70 when methanol was changed into ethanol, propanol, butanol, pentanol and the increased selectivity between C60 and C70 when the alcohol moves from ethanol to pentanol. For methanol, C60 was always eluted before C70. For ethanol, propanol, butanol, Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1305
pentanol, the reverse was observed (Figure 4). This corroborated the fact that the mobile phase was dominant in governing selectivity changes of the apolar solute.6 The ln KCX,A plots (respectively lnKCX,AW) against -∆HoCX,A (respectively -∆HoCX,AW) for the five organic modifiers and X ) 60 and X ) 70 were drawn. The correlation coefficients for the fits were at least equal to 0.979 and a Fisher-Schnedecor test (F-test) demonstrated, with a probability of 0.990, the linear character of the regressions. Figure 5 shows lnKCX,A plotted as a function of -∆HoCX,A for C60 and C70. The respective degree of correlation, r2 ) 0.995 and r2 ) 0.988, can be considered adequate to verify enthalpy-entropy compensation. Thus, the type of interaction was found to be the same for the five organic modifiers. For a set of organic modifiers where there is enthalpy-entropy compensation, the slope of lnK vs -∆H° will be the same for the same type of reaction.34 For the two studied physicochemical processes, the relative difference in the slope values obtained for C60 and C70 was less than 1%. For example, for ln KCX,A plotted as a function of -∆HoCX,A for C60 and C70 (Figure 5) the respective slope values were 0.498 and 0.494. This indicates that the type of interaction was the same for C60 and C70. The compensation temperature β was determined for the solute solvation with the free alcohol (βCX,A) and with the
alcohol water cluster (βCX,AW). The values obtained decreased as follows:
(34) Tomasella, F. P.; Fett, J.; Clive Love, L. J. Anal. Chem. 1991, 63, 474.
AC990842K
1306
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
βCX,AW < βCX,A
(23)
These variations confirmed the previous results25 and indicate that the fullerene solvation by the free alcohol contributed to the retention mechanism more significantly than the CX solvation by the methanol/water clusters. In summary, the concept of an alcohol/water cluster was applied to the retention mechanism of C60 and C70 fullerenes on an RP18 stationary phase. The fullerene solvation energies were determined. The two linear relations obtained between the Gibbs free energy changes of the free alcohol and alcohol/water clusters exchange processes and the carbon atom number of the hydrophobic chain of the alcohol molecule were consistent with a reversed elution order observed for C60 and C70 when methanol was changed into ethanol, propanol, butanol, pentanol. ACKNOWLEDGMENT We thank Mireille Thomassin for her technical assistance. Received for review July 27, 1999. Accepted December 14, 1999.
