Measurement of Partition Coefficients in Waterless Biphasic Liquid

Alain Berthod,* Anne Isabelle Mallet, and Madeleine Bully. Laboratoire des Sciences Analytiques, Université Claude Bernard, Lyon 1, UA CNRS 435 (J. M...
0 downloads 0 Views 196KB Size
Anal. Chem. 1996, 68, 431-436

Measurement of Partition Coefficients in Waterless Biphasic Liquid Systems by Countercurrent Chromatography Alain Berthod,* Anne Isabelle Mallet, and Madeleine Bully

Laboratoire des Sciences Analytiques, Universite´ Claude Bernard, Lyon 1, UA CNRS 435 (J. M. Mermet), 69622 Villeurbanne cedex, France

Countercurrent chromatography (CCC) is a chromatographic separation technique that uses a liquid as a stationary phase. Centrifugal forces are used to immobilize the liquid stationary phase when the liquid mobile phase is pushed through it. In CCC, the solutes are separated according to their liquid-liquid partition coefficients. The solutes studied were the alkylbenzene homologues from benzene to hexylbenzene and some polyaromatic hydrocarbons (PAHs) from naphthalene to coronene. Their liquid-liquid partition coefficients were measured in the five waterless biphasic systems formed by heptane, as the apolar liquid phase of the five biphasic systems, and four dipolar aprotic solvents, dimethyl sulfoxide, dimethylformamide, furfural, and N-methylpyrrolidone, and the polar proton-donor solvent methanol. The coefficients were compared to the corresponding capacity factors obtained by classical liquid chromatography on octadecyl-bonded silica. For the five biphasic solvent systems studied, linear relationships were found between the partition coefficients and the sp3 and sp2 hybridized carbon atom number for the alkylbenzene and PAH series, respectively. The sp2 and sp3 transfer energies were estimated, and their ratio was used to quantify the solvent selectivity toward aromatic extraction. Countercurrent chromatography (CCC) is a chromatographic separation technique that uses a liquid stationary phase.1-3 The mobile phase and the stationary phase form a biphasic liquid system. The very name of the technique was chosen to mean that the technique was a hybrid between the countercurrent distribution method (Craig machines) and liquid chromatography.1 It may be misleading: there is no countercurrent liquid circulation in CCC. The technique name was coined by Yoishiro Ito, who invented the technique and developed it.1,5,6 He published more than 160 papers using the CCC acronym, well establishing it. The advantages of having a liquid stationary phase in chromatography are (i) a high loading capability, (ii) a very simple (1) Ito, Y. Adv. Chromatogr. 1984, 60,181-226. (2) Mandava, N. B., Ito, Y., Eds. Countercurrent Chromatography, Theory and Practise; Chromatographic Science Series 44; M. Dekker: New York, 1988. (3) Conway, W. D. Countercurrent Chromatography; VCH Publishers Inc.: Weinheim, 1990. (4) Foucault, A., Ed. Centrifugal Partition Chromatography; Chromatographic Science Series 68; M. Dekker: New York, 1995. (5) Ito, Y.; Weinstein, M. A.; Aoki, I.; Harada, R.; Kimura, E.; Nunogaki, K. Nature 1966, 212, 985-987. (6) Ito, Y.; Conway, W. D. High Speed Countercurrent Chromatography; J. Wiley & Sons Inc.: New York, 1995. 0003-2700/96/0368-0431$12.00/0

© 1996 American Chemical Society

solute retention mechanism, (iii) that either phase of the biphasic system can be used as a mobile phase, (iv) no irreversible solute adsorption, (v) no pH problem, and (vi) less biological solute denaturation. The high loadability is possible because the solutes reach the volume of the liquid stationary phase and not just the surface of the solid phase as in classical liquid chromatography (LC). In CCC, the solutes are retained according to their liquid-liquid partition coefficient:

VR ) VM + PVS

(1)

where P is the solute partition coefficient expressed as the ratio of the solute concentration in the stationary phase to the solute concentration in the mobile phase. The subscripts R, M, and S refer to the retention, mobile phase, and stationary phase volumes, respectively. The dual-mode use of CCC, i.e., the stationary phase becomes the mobile phase and vice versa, precludes any irreversible adsorption inside the CCC column. It is well described in the literature.3,4,7,8 The CCC columns are made of perfluorinated polymers that can withstand extreme pHs, unlike the classical silica skeleton of LC packings.9 Aqueous two-phase systems were used for protein or cell separations without denaturation.10 The direct and accurate measurement of liquid-liquid partition coefficients is another important capability of CCC.11 Octanolwater partition coefficients as high as 20 000 (log P ) 4.3) were determined by CCC.12,13 The dual-mode reversal of the phase roles of octanol and water was proven useful in such determinations.14 The retention volume of the solute is directly related to its liquid-liquid partition coefficient (eq 1). The vast majority of the biphasic liquid systems used in CCC contain water as one solvent. In this work, the capability of CCC to work with waterless biphasic liquid systems, which we introduced for fullerene (7) Gluck, S. J.; Martin, E. J. J. Liq. Chromatogr. 1990, 13, 3559-3570. (8) Berthod, A.; Chang, C. D.; Armstrong, D. W. In Centrifugal Partition Chromatography; Foucault, A., Ed.; Chromatographic Science Series 68; M. Dekker: New York, 1995; Chapter 1, pp 1-24. (9) Berthod, A. J. Chromatogr. 1991, 549, 1-28. (10) Van Alstine, J. M.; Synder, R. S.; Karr, L. J.; Harris, J. M. J. Liq. Chromatogr. 1985, 8, 2293-2313. (11) Berthod, A. In Centrifugal Partition Chromatography; Foucault, A., Ed.; Chromatographic Science Series 68; M. Dekker: New York, 1995; Chapter 7, pp 167-198. (12) Berthod, A.; Dalaine, V. Analusis 1992, 20, 325-331. (13) Berthod, A.; Menges, R. A.; Armstrong, D. W. J. Liq. Chromatogr. 1992, 15, 2769-2779. (14) Gluck, S. J.; Martin, E.; Benko, M. H. in Centrifugal Partition Chromatography; Foucault, A., Ed.; Chromatographic Science Series 68; 1995; M. Dekker: New York, Chapter 8, pp 199-218.

Analytical Chemistry, Vol. 68, No. 3, February 1, 1996 431

Table 1. Physicochemical Properties of the Solvents Used

solvent

MW

bp (°C)

d (g/cm3)

dipole moment (D)

dielectric const

Hildebrandt δ

polarity Snyder 

Reichardt ET

DMF DMSO furfural heptane methanol NMP

73 78 96 100.2 32 99

153 189 162 98.4 64.5 202

0.948 1.095 1.154 0.679 0.791 1.028

3.86 4.3 3.5 0 2.87 4.1

36.7 48.7 38 1.92 32.7 32.2

24.2 24 23.6 15.2 29.3 22.4

6.4 7.2 3.8 0.1 5.1 6.7

40.4 44.4 50 1.2 76.2 35.5

Mutual Solubility at 20 °C wt %

mol %

a

solvent

solvent in heptane

heptane in solvent

solvent in heptane

heptane in solvent

solvent in heptane

DMF DMSO furfural heptane methanol NMP

6.9 1.6 4a 100 10.4 12a

6.3 0.2 4a 100 2 11a

5.1 1.0 2.4 100 9.6 8.3

8.2 0.3 6.6 100 2.2 11.4

48 11 28 679 76 85

56 2.2 45 679 15 107

25 °C.

separation,15,16 was further investigated. In the petroleum industry, such systems are commonly used in the refining process of crude oils and distillates. The partition coefficients of polyaromatic hydrocarbons (PAHs) and alkylbenzene solutes in five different waterless liquid systems were measured by CCC. The unique selectivities obtained with these five liquid systems were compared to the selectivities obtained by reversed phase LC with a classical octadecyl (C18)-bonded silica stationary phase and two different hydroorganic mobile phase compositions. EXPERIMENTAL SECTION Countercurrent Apparatus. The CCC apparatus was the Model CPHV 2000 from Socie´te´ Franc¸ aise de Chromato Colonne (SFCC, Shandon, 95612 Cergy-Pontoise, France). It is a coil-planet centrifuge device, first designed by Ito.17 It was fully described in a recent article.18 The apparatus volume, VC, was 153 mL, with a total PTFE tube length of 76 m coiled on three spools. Each spool was coiled with 95 turns (five layers of 19 turns) of 1/8 in. (1.6 mm i.d.) PTFE tubing. The Ito β value is the ratio of the coil radius, r, to the spool revolution radius, R. The β ratio was 0.61 for the first inner diameter with r ) 3.7 cm and R ) 6 cm. It was 0.82 for the outermost visible layer with r ) 4.9 cm and R ) 6 cm. The average β value for this CCC device was 0.72. The internal volume of one spool was 51 mL. The spools are interconnected with 1/16 in. PTFE tubes (0.5 mm i.d.). The whole system was housed in an air-thermostated box. The temperature was regulated at 22 °C ( 0.5 °C. Classical Liquid Chromatography. A classical reversed phase liquid chromatograph (RPLC) was used. It comprised a Shimadzu LC-10AS pump (Touzart & Matignon, Vitry, France), a Rheodyne 4125 injection valve, a Shimadzu SPD-6A UV detector (254 nm), and a Shimadzu CR-6A integrator (Touzart & Matignon). The column was a 15 cm × 4.6 mm i.d. column filled with (15) Berthod, A.; (16) Gasper, M. Chromatogr. (17) Ito, Y.; Oka, (18) Berthod, A.;

432

g/L

heptane in solvent

Talabardon, K. Analusis 1995, 23, 174-179. P.; Berthod, A.; Talabardon, K.; Armstrong, D. W. J. Liq. 1995, 18, 1019-1034. H.; Slemp, J. L. J. Chromatogr. 1989, 475, 219-227. Bully, M. Anal. Chem. 1991, 63, 2508-2512.

Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

5 µm Nucleosil silica particles bonded by octadecyl (C18) moiety (Life-Shandon, Cergy, France). The mobile phase was pure methanol or methanol-water (95:5 v/v) at 1 mL/min flow rate. Chemicals. Benzene and the alkylbenzenes toluene, ethylbenzene, propylbenzene, butylbenzene, and hexylbenzene were obtained from Merck (Shuchardt, Germany) and Aldrich (Steinheim, Germany). The PAHs naphthalene, anthracene, pyrene, benzo-R-pyrene, and coronene were obtained from Sigma and Fluka (L’Isles d’Abeau Chesnes, France). The solvents, whose relevant physicochemical properties are listed in Table 1, were obtained from Fluka, Aldrich, and Laurylab (Chassieu, France). Dimethylformamide (DMF), dimethyl sulfoxide (DMSO), Nmethylpyrrolidone (NMP), and methanol were used as received. Heptane was used instead of hexane because its toxicity is lower. CCC Procedure. The mobile phase-stationary phase equilibrium inside the CCC machine needs to be prepared for each biphasic liquid system. The procedure described for the DMSOheptane liquid system can be transposed for the other biphasic systems. Heptane will be the stationary phase and DMSO the mobile phase. The CCC apparatus is first filled with the stationary phase, heptane-saturated by DMSO, using a Model LC6A Shimadzu pump. This takes about 20 min at 8 mL/min. Next, the rotor is started and the rotation allowed to stabilize at the desired speed, usually 800 rpm. DMSO, the denser liquid phase, is pushed at the desired flow rate entering through the “head” of the machine.2-4 As long as the apparatus is not equilibrated, the heptane stationary phase is displaced, exiting by the “tail” of the apparatus. The heptane phase is collected in a graduated cylinder. Once the DMSO phase appears at the CCC device tail, two layers are seen in the graduated cylinder. The mobile phase-stationary phase equilibrium is reached. The CCC “column” is ready. The volume of the heptane phase displaced gives the mobile phase volume, VM, inside the machine. The stationary phase volume, VS, is calculated with the column volume, VS ) VC - VM. The stationary phase retention factor, Sf, is expressed as a percentage of the apparatus volume:

Sf ) (VC - VM)/VC ) VS/VC

(2)

Sf values higher than 50% are desirable to obtain a good resolution.4,9,11 The mixture of solutes to be separated is injected using an in-line Rheodyne 7125 valve with a 500 µL sample loop. All experiments were done at an 800 rpm rotation speed and a 1.5 mL/min mobile phase flow rate. A Model SPD6A Shimadzu UV detector was used with a preparative 6 µL cell (path length, 1 mm). The Particular Case of the Heptane-Furfural System. The furfural received was brown colored. It became colorless upon distillation. However, it was only stable under nitrogen atmosphere. Upon exposure to air, a yellow color developed in minutes. In the CCC machine, the pale yellow entering furfural exited the centrifuge with a brown color. The vigorous mixing and gas diffusion through the PTFE tubing were the probable causes of the observed color change. The furfural color precluded the use of any spectrophotometric detector. A Cunow DDL 21 evaporative light scattering detector (ELSD) was used at 90 °C to monitor continuously the furfural mobile phase when the PAHs were studied. The alkylbenzenes were too volatile to be detected with the ELSD. Their heptane-furfural partition coefficients were measured with the furfural stationary phase. The heptane mobile phase was collected with a Model Frac-100 fraction collector (Pharmacia LKB, Dublin, Ireland). Each 4 mL fraction was analyzed by gas chromatography using a Model GC-17A Shimadzu gas chromatograph with a 10 m × 0.53 mm i.d. macrobore open tubular CP-Sil 5CB column (ChromPack, Les Ulis, France) (film thickness, 5 µm). The carrier gas was helium with flame ionization detection. The CCC chromatogram was reconstructed using the GC alkylbenzene peaks normalized with the furfural peak area. RESULTS AND DISCUSSION Choice of the Liquid Systems. Heptane was selected for the apolar liquid phase. It forms biphasic liquid systems with many polar and intermediate polarity solvents with relatively low mutual solubility. Table 1 lists the polarity values of the solvents in three different scales. The Hildebrandt scale19 uses the work necessary to separate two solvent molecules. The parameter, δ, is expressed by

δ ) x(∆HV - RT)/Vm

(3)

where ∆HV is the solvent molar heat of vaporization, R is the perfect gas constant, T is the absolute temperature, and VM is the solvent molar volume. The Snyder scale,20 parameter o, is based on the eluting power of the solvent used as a mobile phase in thin-layer chromatography on alumina plates. The Reichardt scale,21 parameter ET, is based on the transition energy for the longest wavelength solvatochromic absorption band of a pyridinium N-phenoxide betaine dye. Table 1 shows that mutual solubility and solvent polarity are not linked. Heptane is much more soluble in polar methanol than in less polar furfural. The selected solvents, except methanol, are dipolar non-proton-donor solvents. They are routinely used (19) Hildebrand, J. H.; Prausnitz, J. M.; Scott, R. L. Regular and Related Solutions; Van Nostrand Reinhold: Princeton, NJ, 1970. (20) Snyder, R. L. Principle of Adsorption Chromatography; M. Dekker: New York, 1968. (21) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH Publishers Inc.: Weinhein, 1988.

Table 2. Stationary Phase Retention Volumesa heptane stationary phase flow: head f tail solvent system heptane-DMF heptane-DMSO heptane-furfural heptane-NMP heptane-methanol

solvent stationary phase flow: tail f head

∆d Sf VS Sf VS (g/cm3) (%) (mL of heptane) (%) (mL of solvent) 0.23 0.40 0.44 0.27 0.09

80 79 82 76 49

125 123 128 118 77

90 89 91 83 30

140 139 142 129 47

a ∆d is the liquid phase density difference. Mobile phase flow rate, 1.5 mL/min; rotation speed, 800 rpm; VC ) 156 mL.

in the petroleum industry for aromatic extraction and distillate purification. Methanol was added as a polar proton-donor solvent often used in CCC separations. Liquid Phase Retention. The stationary phase retention is the main problem in CCC. Obviously, some liquid stationary phase retention is necessary to obtain some compound separation by CCC. It was shown that numerous physicochemical parameters such as density,22 viscosity, mutual solubility, and interfacial tension2-4,23 were related to the liquid retention by hydrodynamic CCC devices with coiled tubes. Also, the liquid stationary phase retention increases with the centrifuge rotation speed and decreases with the flow rate.1,24-25 Table 2 lists the retention factors obtained with the five biphasic waterless systems at the 800 rpm and 1.5 mL/min operating conditions. A great advantage of CCC is that either liquid can be the mobile phase. With the liquid systems studied, heptane is the lighter liquid (Table 1). If heptane is chosen to be the stationary phase, the denser solvent is the mobile phase. It should enter the CCC apparatus through the head, which is the highest pressure point.1-4 If heptane is the mobile phase, it should enter the CCC apparatus through the tail to move through the denser stationary phase up to the head of the machine. The tail is the lowest pressure point. The measured liquid stationary phase retention by the CCC apparatus used was higher than 75% of the internal volume for all solvents but methanol (Table 2). The difficult liquid retention with the heptane-methanol system is due to the low density difference between the two liquid phases.22 Also, the high solubility of heptane in methanol is not a favoring factor. Yet, the 30% methanol retention is enough to measure the methanolheptane alkylbenzene partition coefficients.18 Solute Partition Coefficients. Table 3 lists the heptanesolvent partition coefficients measured using the retention volume of the compounds injected in the CCC machine (eq 1). Figure 1 shows two typical CCC chromatograms of alkylbenzenes (top) and PAHs (bottom) obtained with the heptane-DMF biphasic liquid system. The experimental conditions are listed in the figure legend. The partition coefficient values listed in Table 3 are the averages of triplicate measurements. The error on the partition coefficient measurement by CCC can be expressed by differentiating eq 1: (22) Berthod, A.; Schmitt, N. Talanta 1993, 40, 1489-1498. (23) Ito, Y. J. Chromatogr. 1984, 301, 377-395. (24) Berthod, A. J. Chromatogr. 1991, 550, 677-693. (25) Fad, M. K. Rev. Inst. Fr. Petrole 1982, 37-2, 225-247.

Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

433

Table 3. Heptane-Solvent Partition Coefficients (24 °C) k′ solvent with heptane furfural NMP

solute

DMF

DMSO

benzene toluene ethylbenzene propylbenzene butylbenzene hexylbenzene

0.59 0.75 0.89 1.05 1.25 1.70

0.98 1.51 2.12 3.16 4.53 9.6

Alkylbenzene Homologues 0.85 0.50 1.11 0.64 1.33 0.74 1.81 0.86 2.23 0.99 3.49 1.23

naphthalene anthracene pyrene benzopyrene coronene

0.32 0.17 0.13 0.065 0.035

0.55 0.34 0.30 0.20 0.14

Polyaromatic Hydrocarbons 0.45 0.21 0.23 0.09 0.17 0.06 0.09 0.02 0.06

methanol

1.05 1.32 1.52 1.82 2.17 3.08 ∼1 ∼1 ∼1

pure methanol

methanol-water ( 95:5 v/v)

0.11 0.15 0.19 0.24 0.30 0.48

0.18 0.25 0.34 0.44 0.57 0.99

0.21 0.42 0.67 1.15 2.10

0.37 0.76 1.17 2.08 3.89

a Partition coefficient obtained with the CCC retention volumes of the solutes (eq 1) and expressed as the ratio of the solute concentration in the heptane phase to the concentration in the solvent phase. k′ values are the RPLC capacity factors of the solutes obtained with a 15 cm column filled with octadecyl-bonded Nucleosil silica phase.

dP ) (∂P/∂VR) dVR + (∂P/∂VS) dVS

(4)

dP ) 1/VS(dVR + (1 - P) dVS)

(5)

which gives

Equation 5 shows that the higher the stationary phase volume, VS, the lower the error on the partition coefficient measurement, dP. It explains the low accuracy of the PAH partition coefficients in the heptane-methanol system. With this system, it was not possible to obtain any PAH separation due to the low stationary phase retention volume. PAHs were injected one by one, producing a peak close to the device volume, VC. Their partition coefficients then are close to unity. Furthermore, the solubility of the higher PAHs (benzopyrene and coronene) is limited in the phases of the heptane-methanol system. This was another factor precluding accurate partition coefficient determination. For the other liquid systems, the stationary phase retention is higher than 80% (Table 2), corresponding to more than 120 mL of stationary phase. The error on the retention volumes is in the milliliter range, producing a partition coefficient absolute error lower than 0.01 unit. It should be noted that partition coefficients are commonly given with an absolute error of 0.1 unit on the log P value. The partition coefficients obtained show that the alkylbenzenes have an affinity for the heptane phase that increases with the alkyl chain length. Oppositely, the affinity of the PAHs for the heptane phase decreases with the carbon atom number. This trend is the reverse of the classical trend observed in RPLC. The retention times of both the alkylbenzenes and the PAHs increase with the carbon atom number. For comparison, the capacity factor of the studied solutes, k′, which corresponds in RPLC to the partition coefficient in CCC, is listed in Table 3 for two mobile phase compositions (pure methanol and 95:5 v/v methanol-water). Comparison of Hydrocarbon Solvent Selectivities. The data listed in Table 3 can be used to compare the solvent selectivities toward alkylbenzene and PAH extraction. This information is of paramount interest in the petroleum industry, which uses the selected solvents in several refining processes.25 434 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

Figure 1. CCC chromatograms. Top, alkylbenzenes; bottom, polyaromatic hydrocarbons. Stationary phase, heptane; mobile phase, DMF 1.5 mL/min flowing from head to tail; rotation speed, 800 rpm; inlet pressure, 1.8 kg/cm2 (25 psi); injection volume, 0.5 mL (solvent DMF); injected mass (top, UV 254 nm, 1.28 AUFS) benzene 40 mg, toluene 49 mg, ethylbenzene 36 mg, propylbenzene 42 mg, pentylbenzene 46 mg, (bottom, UV 254 nm, 2.56 AUFS) benzopyrene 2.5 mg, pyrene 4 mg, anthracene 5 mg, naphthalene 7.5 mg, and benzene 80 mg.

For alkylbenzene homologues, the partition coefficient, P, is related to the free energy of transfer, ∆G°, from one liquid phase

Table 4. ln P versus nC Regression Lines and Transfer Energies k′ a DMF

solvent with heptane furfural NMP

DMSO

slope intercept r2 coefficient ∆G°CH2 (kJ/mol)

0.173 -0.48 0.995 0.42

0.377 0.0 0.999 0.92

Alkylbenzene Homologues 0.232 0.135 0.14 -0.58 0.995 0.995 0.56 0.33

slope intercept r2 coefficient ∆G°Csp2 (kJ/mol) -G°CH2/∆G°Csp2

-0.161 0.49 0.999 -0.39 1.07

-0.096 0.33 0.995 -0.24 3.93

Polyaromatic Hydrocarbons -0.163 -0.227 0.83 0.76 0.999 0.997 -0.40 -0.56 1.42 0.59

a

methanol

pure methanol

methanol-water (95:5 v/v)

0.176 0.07 0.998 0.43

0.238 -2.17 0.996 0.58

0.275 -1.66 0.998 0.67

0.167 -3.21 0.995 0.40 -1.42

0.171 -2.68 0.997 0.42 -1.61

In RPLC, the k′ versus nC lines were analyzed.

Figure 2. Plot of eq 9 lines with ln P ) 0. The hydrocarbon molecules located at the left and above the lines are extracted by the solvent; at the right and below the line they stay in the heptane apolar phase. These lines model the aromatic extraction by solvents at 25 °C.

to the other18,26 by

RT ln P ) nCch2∆G°CH2 + ∆G°Ph

(6)

where nCch2 is the carbon atom number in the alkyl chain, and CH2 and Ph refer to a methylene and a phenyl group, respectively. The plot of ln P vs nCch2 should produce a straight line whose slope and intercept give the methylene and phenyl energies of transfer, respectively. The model of eq 6 was transposed for PAHs:

RT ln P ) nCsp2∆G°Csp2 + A

with

nCsp2 g 6

(7)

in which the subscript Csp2 refers to the number of sp2 hybridized carbon atoms in the PAH molecule, and A is a constant. Table 4 lists the slopes, intercepts, regression coefficients, and transfer energies obtained with the Table 3 data. All the regression coefficients were higher than 0.995, which shows that eq 7 has some pertinence. The ∆G°CH2 and ∆G°Csp2 values obtained from the slopes of the ln P vs nC plots are given in kJ/ mol. The ln k′ vs nC plots were also studied for comparison with RPLC. (26) Guiochon, G. In HPLC: Advances and Perspectives; Horwath, Cs., Ed.; Academic Press: New York, 1980; Vol. 2, pp 1-54.

Figure 3. Plot of eq 9 lines for RPLC with ln k′ ) 0. These lines model the hydrocarbon solid phase extraction on C18-bonded phases with methanol and 95% methanol-5% water mobile phases.

To quantify the notion of solvent selectivity, we assume that the results obtained for the linear alkylbenzene series and the PAH series can be extended to any combination of aromatic rings and alkyl side chains. Oversimplifying, any hydrocarbon molecule is considered to contain nCsp3 carbon atoms in the sp3 hybridized state and nCsp2 carbon atoms in the sp2 hybridized state. Possible alkyl ramifications or cycles, multiple side chains, fused aromatic rings, or multiple rings or combinations of these factors are not considered. Heteroatoms that drastically change the molecule polarity are excluded. The ∆G°Ph of eq 6 corresponds to nCsp2 ) 6 in eq 7. For a hydrocarbon molecule, then, eqs 6 and 7 give

RT ln P ) nCsp3∆G°CH2 + nCsp2∆G°Csp2 + A

(8)

Equation 8 shows a linear relationship between nCsp2 and nCsp3 for a constant P value:

nCsp2 ) -nCsp3∆G°CH2/∆G°Csp2 - (A - RT ln P)/∆G°Csp2 (9) The slope of the n vs n lines is the ratio ∆G°CH2/∆G°Csp2 at a constant P value. The intercept depends on ln P. The ∆G°CH2/ ∆G°Csp2 ratio shows how many sp2 hybridized carbon atoms are needed to compensate the heptane partition coefficient increase due to one sp3 hybridized carbon atom (Table 4). This ratio can be used to quantify the heptane-solvent selectivity. Figure 2 shows the ln P ) 0 lines for the four solvents studied; ln P ) 0 Csp2

Csp3

Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

435

(P ) 1) corresponds to an equal partitioning of the molecule between the heptane phase and the solvent phase. The ln P ) 0 line cuts the plane in two parts. The lower right part corresponds to the hydrocarbons with high P values (ln P > 0), i.e., high heptane affinity. The upper left part corresponds to hydrocarbons with P < 1 values (ln P < 0), or high solvent affinity. These hydrocarbons are extracted by the solvent phase. Figure 2 clearly shows that DMSO has a high selectivity for PAHs with low alkyl substituents. NMP extracts a large amount of hydrocarbon molecules. It leaves only the saturated and highly alkylsubstituted aromatic molecules in the heptane phase. The selectivity order for aromatic hydrocarbons is DMSO > furfural > DMF > NMP. Figure 3 shows the ln k′ ) 0 lines for the two mobile phases studied in RPLC. The slopes are negative (Table 4). The conclusion is straightforward: the hydrocarbon affinity for the C18-bonded stationary phase increases with the molecule carbon content regardless of the hybridization state of the carbon atoms. A 5% v/v amount of water in the methanol phase pushes more light hydrocarbon molecules on the apolar stationary phase. Solid phase extraction on short cartridges filled by octadecyl-bonded silica is commonly used to extract easily apolar material from hydroalcoholic solutions.

436

Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

CONCLUSION The capability of the CCC technique to measure easily and accurately liquid-liquid partition coefficients is shown. The usefulness of such coefficients in various chemical fields is known. In the example proposed, the partition coefficients were measured in two-solvent biphasic liquid systems. The addition of small amounts of a third solvent can change dramatically the partition coefficient.2,3,8,11,14-16 The chromatographic selectivity can be adjusted using solvent mixtures, as recently described for fullerene separation by CCC.15,16 The alkylbenzene homologues and PAHs partition coefficients could be measured by CCC with a biphasic liquid system made of more than two solvents. The furfuralDMSO-heptane system could be used, for example. ACKNOWLEDGMENT Funding for this research has been provided by the Centre National de la Recherche Scientifique (CNRS, UA 435) and by TOTAL Raffinage Distribution (Ph.D. Grant E29 920113 AB SH 820540 to M.B.), which we gratefully acknowledge. Received for review April 27, 1995. Accepted October 16, 1995.X AC950409K X

Abstract published in Advance ACS Abstracts, December 15, 1995.