Anal. Chem. 2003, 75, 5886-5894
Elution-Extrusion Countercurrent Chromatography. Use of the Liquid Nature of the Stationary Phase To Extend the Hydrophobicity Window Alain Berthod,*,† Maria Jose Ruiz-Angel,‡ and Samuel Carda-Broch§
Laboratoire des Sciences Analytiques, Universite´ Claude-Bernard-Lyon 1, 69622 Villeurbanne, France, Departmento de Quimica Analitica, Universidad de Valencia, 46100 Burjassot, Spain, and Area de Quimica Analitica, Universidad Jaime I, 12006 Castello´ n, Spain
Countercurrent chromatography (CCC) is a liquid chromatography technique with a liquid stationary phase. Taking advantage of the liquid nature of the stationary phase, it is possible to perform unique operations not possible in classical liquid chromatography with a solid stationary phase. It is easy to avoid any solute-irreversible absorption in the CCC column. If the retention volumes of solutes become too high, the dual mode will be used. The roles of the phases are reversed. The stationary phase becomes the mobile phase, and the CCC column is started again. The solutes elute rapidly in what was previously the stationary phase. The theoretical basis of the dualmode method is recalled. The dual-mode method is a discontinuous method. The separation should be stopped when the phase switch is performed. The elution-extrusion procedure is another way to avoid any irreversible adsorption of solutes in the column. The method uses the fact that the liquid volumes occupied by the solutes highly retained inside the column can be orders of magnitude lower than the mobile-phase volume that would be needed to elute them. The elution-extrusion method also has two steps: the first step is a regular CCC chromatogram. Next, the stationary phase containing the partially separated hydrophobic solutes is extruded out of the column in a continuous way using the liquid stationary phase. The theory of the process is developed and compared to the dual-mode theory. Alkylbenzene homologues are experimentally used as model compounds with the heptane/ methanol/water biphasic liquid system to establish the theoretical treatment and compare the performance of two types, hydrodynamic and hydrostatic, of CCC columns. It is shown that the method can dramatically boost the separation power of the CCC technique. An apparent efficiency higher than 20 000 plates was obtained for extruded octylbenzene and a 160-mL hydrodynamic CCC * Corresponding author. E-mail:
[email protected]. † Universite´ Claude-Bernard-Lyon 1. ‡ Universidad de Valencia. § Universitad Jaime I.
5886 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
column with less than 500 plates when conventionally used. Countercurrent chromatography (CCC) is a liquid chromatography technique that uses a liquid stationary phase.1-3 Biphasic liquid systems are used to separate solutes. The big advantage of a liquid stationary phase is that the solutes have access to the whole volume of the phase. CCC is fundamentally a preparative separation technique.4 The problem is to retain the liquid stationary phase when the liquid mobile phase is pushed through it. In all modern CCC machines, centrifugal fields are used for this purpose. CCC is extremely useful for the separation, purification, and isolation of natural products.5-8 These mixtures contain a high number of different compounds with a broad range of hydrophobicity. Most often only one compound needs to be separated from the others. The separation and the liquid system is then adapted for this purification. In the case of the high-throughput screening technology used in the pharmaceutical industry with natural product drug discovery programs, the method uses extracts that should be purified of interfering polar (polyphenols) or apolar (fatty acids or esters) compounds. In this case and with an isocratic elution method, the retention times or volumes of the different solutes will be extremely different. Gradients of organic (1) Berthod, A., Ed. Countercurrent Chromatography, The Support-Free Liquid Stationary Phase; Comprehensive Analytical Chemistry XXXVIII; Elsevier: Amsterdam, 2002. (2) Ito, Y., Conway, W. D., Eds. High-Speed Countercurrent Chromatography; Chemical Analysis 132; Wiley: New York, 1996. (3) Conway, W. D. Countercurrent Chromatography, Apparatus, Theory & Applications; VCH Publishers: Weinheim, 1990. (4) Berthod, A.; Billardello, B. In Advances in Chromatography; Brown, P., Grushka, E., Eds.; Marcel Dekker: New York, 2000; Vol. 40, Chapter 10, pp 503-538. (5) Maillard, M.; Martson, A.; Hostettmann, K. In High-Speed Countercurrent Chromatography; Ito, Y., Conway, W. D., Eds.; Chemical Analysis 132; Wiley: New York, 1996; pp 179-223. (6) Gunawardana, G.; McAlpine, J. In Countercurrent Chromatography; Menet, J. M., Thiebaut, D., Eds.; Chromatography Science Series 82; Marcel Dekker: New York, 1999; pp 249-271. (7) Marston, A.; Hostettmann, K. J. Chromatogr., A 1994, 658, 315-331. (8) Zhang, T. In Countercurrent Chromatography, The Support-Free Liquid Stationary Phase; Berthod, A., Ed.; Comprehensive Analytical Chemistry XXXVIII; Elsevier: Amsterdam, 2002; pp 201-260. 10.1021/ac030208d CCC: $25.00
© 2003 American Chemical Society Published on Web 09/30/2003
modifier composition are used in classical HPLC to reduce the experiment duration. In CCC, the use of a biphasic liquid system to separate solutes requires knowledge of the liquid-liquid equilibriums and the solute partitioning theory. It should always be kept in mind that any change of the mobile-phase composition is likely to change the stationary-phase composition or volume as well: gradients of mobile-phase composition are difficult to use.9-11 Early on, the proposed solution was to use the liquid nature of the stationary phase in CCC. The first possibility is to interchange the role of the liquid phases during the experiment: the mobile phase becomes the stationary phase and vice versa. This method was dubbed dual-mode CCC, and it is commonly used in many purification processes.3,12-14 This work describes it deriving the theoretical equations that allow us to predict its efficacy. The drawback of the dual-mode method is that it is discontinuous, which could render automation difficult and lengthy. The CCC machine must be reequilibrated after every experiment. The second possibility is to extrude the column content to recover it. This solution was first proposed by Conway, who called it column extrusion.15 He proposed to use compressed nitrogen or air to extrude the whole CCC column content. He designed backward extrusion for very highly retained solutes, still located at the column head, and forward extrusion for solutes closer to the column end.15 The column extrusion method was used in several works.3,12,16-18 It was never fully studied theoretically. This is the goal of this work. We propose to call the method “elutionextrusion” to point out the fact that it is also a two-step method. The elution-extrusion method is also compared with the dualmode method. EXPERIMENTAL SECTION Chemicals. Benzene, toluene, ethylbenzene, butylbenzene, and octylbenzene were obtained from Sigma-Aldrich (SaintQuentin, Fallavier, France). Methanol and heptane were from SDS (Peypin, France). Heptane rather than hexane was used due to its lower toxicity. Water was deionized and distilled before use. Apparatuses. Two CCC apparatuses were used. The first was a hydrodynamic apparatus of the Ito J type with three coils.19 This coil-planet-centrifuge machine was the model CPHV 2000 sold (9) Foucault, A. In Centrifugal Partition Chromatography; Foucault, A., Ed.; Chromatography Science Series 68; Marcel Dekker: New York, 1995; pp 71-97. (10) Foucault, A.; Chevolot, L. J. Chromatogr., A 1998, 808, 3-22. (11) Renault, J. H.; Nuzillard, J. M.; Intes, O.; Maciuk, A. In Countercurrent Chromatography, The Support-Free Liquid Stationary Phase; Berthod, A., Ed.; Comprehensive Analytical Chemistry XXXVIII; Elsevier: Amsterdam, 2002; pp 49-83. (12) Berthod, A.; Brown, L.; Leita˜o, G. G.; Leita˜o, S. G. In Countercurrent Chromatography, The Support-Free Liquid Stationary Phase; Berthod, A., Ed.; Comprehensive Analytical Chemistry XXXVIII; Elsevier: Amsterdam, 2002; pp 21-47. (13) Foucault, A. P. Anal. Chem. 1991, 63, 569A-579A. (14) Agnely, M.; Thiebaut, D. J. Chromatogr., A 1997, 790, 17-30. (15) Conway, W. D. in ref 3, pp 349-351. (16) Ingkaninan, K.; Hazekamp, A.; Hoek, A. C.; Balconi, S.; Verpoorte, R. J. Liq. Chromatogr. Relat. Technol. 2000, 23, 2195-2208. (17) Armbruster, J. A.; Borris, R. P.; Jimenez, Q.; Zamora, N.; Tamayo-Castillo, G.; Harris, G. H. J. Liq. Chromatogr. Relat. Technol. 2001, 24, 1827-1840. (18) Hazekamp, A.; Verpoorte, R.; Panthong, A. J. Ethnopharmacol. 2001, 78, 45-49. (19) Ito, Y.; Sandlin, J.; Bowers, W. G. J. Chromatogr. 1982, 244, 247-259.
from 1988 to 1993 by the Socie´te´ Franc¸ aise de Chromato Colonne (Neuilly-Plaisance, France). It contains three multilayer coils connected in series and spinning with a planetary motion around a central axis. Each spool bears a gear that meshes into the central stationary gear. The connecting tubing is twisted by the spool rotation. It is unwound passing through antitwisting connectors that rotate in the opposite direction. With such a gear arrangement, if the rotor makes one full rotation, each coiled spool makes two rotations. Each spool was filled with 133 turns of 1/8-ft or 3-mm PTFE tubing (i.d. 1/16 ft or 1.6 mm), length 29 m, coiled in 7 layers of 19 turns. The Ito β value is the ratio of the coil radius, r, to the spool revolution radius, R. The β ratio was 0.37 for the inner first layer with r ) 2.2 cm and R ) 6 cm. It was 0.75 for the most outer visible layer with r ) 4.5 cm and R ) 6 cm. The average β value for this CCC apparatus was 0.56. The internal volume of one coiled spool was ∼55 mL. The three-coil apparatus had a total internal volume, VT, of 175 mL. The total PTFE tube length was 87 m, with a total of 400 turns. All rotation speeds given in text correspond to rotor rotation. The second CCC machine was a hydrostatic apparatus, model HPCPC built by Sanki Engineering Ltd. (Ever Seiko, Tokyo, Japan), sold by J. M. Science (Grand Island, NY). This hydrostatic machine has a horizontal rotor (radius 7.8 cm) containing 1060 channels of ∼90 µL interconnected by ducts. Two rotary seals are used for the inlet and the outlet of the mobile phase. The total internal volume was 101 mL. A Shimadzu pump (model LC6A, Kyoto, Japan) was used to fill the CCC apparatus with the stationary phase and to push the mobile phase. Solute monitoring was performed with a Shimadzu UV detector (model SPD-6A), usually at 254 nm, using a preparative cell (8-µL volume, 1-mm path length). The signal was recorded by a Shimadzu integrator model CR5-A. Procedure. The heptane/water/methanol system is very interesting in CCC. For all ternary compositions with more that 1% v/v water, two liquid phases form, of which one is practically pure heptane.3 This means that a biphasic liquid system made by mixing a volume of methanol/water 90/10% v/v solution with any volume of pure heptane is made of pure heptane and a methanol/ water solution saturated by less than 5% v/v heptane.3 The CCC machine is first filled with the light stationary phase (heptane). The rotor is started at the desired speed. A cylinder is placed at the column outlet, and the mobile phase (the denser methanol/water phase) is pumped at the selected flow rate. The stationary phase is displaced by the mobile phase and collected in the cylinder. When the column is equilibrated, the mobile phase exits the CCC machine and two phases appear in the cylinder. The volume of heptane (the stationary phase) collected corresponds to the volume of mobile phase, VM, in the equilibrated column. The volume of stationary phase is VS ) VC - VM, with the column volume, VC. The CCC column is equilibrated; the solute mixture can be injected. THEORY Classical Elution of Solutes. The first step of both the dual-mode method and the elution-extrusion method is a classical elution. The retention volume, VR, of a solute is very simply
VR ) VM + KDVS Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
(1) 5887
with KD, the solute distribution constant (ratio of the solute concentration in the stationary phase to concentration in the mobile phase at equilibrium). Since KD is responsible for the solute partitioning between phases, it can also be termed partition ratio, but IUPAC does not recommend calling it partition coefficient.20 The solutes are sorted proportionally by increasing distribution constants. Inside the CCC “column”, all solutes move at a speed also connected to their distribution constant. When a volume VCM of mobile phase passed through the column (CM for classical mode), the noneluted solutes are located inside the column at a “distance” xi from the column head:
xi ) LVCM/VRi
(2)
with L, the column “length”. xi could be expressed in volume units if VC replaces L. As the volume VCM increases, the solute i elutes from the column when xi becomes higher than L, which is when VCM becomes higher than VRi. Defining a selectivity criterion as the ratio of the retention volume difference between two solutes over their distribution constant difference, dVR/dKD,12 it can trivially be found that
dVR /dKD ) VS
(3)
Equation 3 clearly shows that the separation between peaks will be as high as the volume of liquid stationary phase retained by the CCC machine can be.12 The resolution factor, Rs, is expressed by
Rs )
VR2 - VR1 (W2 + W1)/2
(4)
with W the peak width at base. Peak widths are related to band broadening. Band Broadening and the Craig Machine. It is important to recall the basis of the chromatographic separation process using the Craig machine model since only solute partitioning is to be considered. In CCC, as in the Craig model, the only parameter responsible for solute separation is the solute distribution constant, KD, the ratio of the solute concentration in the stationary phase to concentration in the mobile phase at equilibrium. The Craig machine is made of N cells or zones in which perfect partition equilibrium is obtained between the two liquid phases. Next, the content of the mobile phase in cell x is passed to cell x + 1 with fresh mobile phase entering cell 1 and monitoring mobile phase exiting cell N. A computer program modeling the Craig machine was developed recently for CCC.21 The program was written in Delphi 3 (visual Pascal language for Windows) and called CHESS for chromatographic experiment simulation system. It is freely available on request.22 It gives the solute concentrations in the two (20) Rice, N. M.; Irving, H. M. N. H.; Leonard, M. A. Pure Appl. Chem. 1993, 65, 2373-2396. (21) Sutherland, I. A.; De Folter, J.; Wood, P. J. Liq. Chromatogr. Relat. Technol. 2003, 26, 1296-1310. (22) De Folter, J. Chromatographic Experiment Simulation System; Brunel University, contact Internet:
[email protected] or berthod@ univ-lyon1.fr
5888
Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
Figure 1. Concentration profile in the two liquid phases of a 100zone Craig machine after 175 mobile-phase transfers. One hundred cells of 2.5 mL () 1 volume unit, dotted area), total machine volume, 250 volume units; stationary-phase volume, 150 volume units; mobilephase volume, 100 volume units. Dashed line (and negative), concentration in stationary phase; full line, concentration in mobile phase. Solute 1 (thin lines), KD ) 2; solute 2 (boldface lines), KD ) 0.5. solute 1 maximum position cell 44; solute 1 peak width at base, 26 volume units. Solute 2 half peak width at base, 20 volume units inside the column and ∼30 volume units outside.
liquid phases inside the column and in the mobile phase outside the column. Figure 1 shows the concentration profiles obtained when two solutes with respective KD constants of 0.5 and 2 were introduced at the same initial concentration in a machine with 100 cells of 2.5 volume units each. After 175 mobile-phase transfers, solute 1 is eluting. A total of 50% left the column and the other 50% is still inside the column. Assuming that a theoretical plate of a countercurrent chromatograph can be view as a cell of the Craig machine, the CHESS program can be used to predict accurately real CCC chromatograms.21 What was important to understand was the evolution of the band broadening inside the column. The program shows that a band injected in one cell in the column moves at a speed proportional to the solute distribution constant (and, of course, the flow rate) (eq 2). As the band moves it broadens. Figure 2 shows the concentration profiles for two solutes with distribution constants of 5 and 10 after 850 mobile-phase unit volumes were passed in the same 100-cell Craig machine as in Figure 1. Comparing Figures 1 and 2, it is apparent that the difference in band broadening occurs outside the column, not inside. Assuming that a 100-cell Craig machine will produce classical chromatograms with 100 theoretical plate Gaussian peaks, W, the peak width at base of eluted peaks is simply
W ) 4σ ) 4VR /xN
(5)
with σ the standard deviation of the Gaussian peak and N the column theoretical plate number (taken as the number of cells of the Craig machine). The band-broadening process is completely different inside the column. Figures 1 and 2 show that the bandwidths of the peaks
Table 1. Chromatographic Parameters Obtained in the Separation of a Hypothetical Mixturea classical reversed-phase elution
dual mode
compd
KD
VR (mL)
W (mL)
Rs
VR (mL)
W (mL)
1 2 3 4 5 6
0.2 0.8 2 6 10 14
40 100 220 620 1020 1420
9.2 23 51 143 236 328
3.7 3.25 4.1 2.1 1.4
40 100 195 152 143 139
9.2 23 30 18 14 12
elution-extrusion Rs 3.7 1.8 0.54 0.29 1.6b
VR (mL)
W (mL)
40 100 179 225 235 239
9.2 23 21 13 10 8
Rs 3.7 3.6c 2.7 0.90 0.50
a K , apolar/polar solute distribution ratio; V , total solute retention volume; W, theoretical peak width at peak base; R , resolution factor; machine D R s volume, VC ) 120 mL; apolar stationary-phase volume, VS ) 100 mL; polar mobile-phase volume, VM ) 20 mL; classical elution for dual mode and b elution-extrusion, VCM ) 130 mL, eluting solutes 1 and 2. Efficiency N ) 300 plates. Resolution between peak 2 (reversed-phase mode) and peak 6 (normal-phase mode). c Resolution between peak 2 (reversed-phase mode) and peak 3 (extruded).
Figure 2. Concentration profile in the two liquid phases of the Figure 1 Craig machine after 850 mobile-phase transfers. Total machine volume, 250 volume units. Dotted area, phase composition inside the machine; stationary-phase volume, 150 volume units; mobile-phase volume, 100 volume units. Dashed line (and negative), concentration in stationary phase; full line, concentration in mobile phase. Solute 1 (thin lines), KD ) 10; solute 2 (boldface lines), KD ) 5. Solute 1 maximum position, cell 53; solute 1 peak width at base, 29 volume units. Solute 2 half peak width at base, 20 volume units inside the column and ∼160 volume units outside.
with KD ) 2 (Figure 1) and KD ) 10 (Figure 2) have a similar magnitude when the peaks cross the same cells inside the column. The two peak widths are respectively (Figure 1) 26 in cell 44 for KD ) 2 after 175 transfers and 30 in cell 53 for KD ) 10 after 850 transfers. In first approximation, the bandwidth at base, inside the column, is close to 4 times the square root of the cell number corresponding to the band position for any distribution constant or phase ratio. Of course, the solute distribution constant and the phase ratio are related to the residence time of the solute inside the column. Assuming Gaussian profiles, it means that WCM, the bandwidth inside the column, can be expressed by
WCM ) 4xVCM/NVRi VC
(6)
with the condition VCM < VRi. Resolution Factor in Classical Elution CCC. With the classical assumptions of Gaussian peaks with σi, the standard deviation of peak i, and a constant efficiency, N (plate number), the peak width at base can be expressed by eq 5. Using eqs 4
Figure 3. Solute motion in classical elution mode in CCC. (A) Injection of a mixture of four solutes; (B) the more polar solute 1 is close to leaving the column; (C) the less polar solutes 3 and 4 start to separate, and solute 2 is exiting.
and 5, the resolution factor equation in classical elution CCC becomes
Rs )
KD2 - KD1 xN 4 VM (KD2 + KD1) + VS 2
(7)
When the distribution constants of solutes are very high, not only does the resolution factor decrease rapidly (eq 7) but also the mobile-phase volume needed to elute the solute can be very large (eq 1); Figure 3 illustrates classical elution in CCC with injection (Figure 3A) and progressive elution of the solutes (Figure 3B and C). Table 1 lists the retention volumes, peak width at base, and resolution factors for six hypothetical solutes having a wide range of hydrophobicity: 0.2 < KD < 14. Solutes 3-6 have polar/ apolar distribution ratios increasing by four units so that their corresponding retention volumes increase by a constant 400-mL volume in the experimental conditions selected. The resolution factors between peaks 3-4, 4-5, and 5-6 decrease respectively from 4.1, 2.1, down to 1.4 (Table 1, classical elution in reversed phase). Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
5889
The minus sign points out that there is an inversion in the solute elution order (Figure 4E). The solute with the higher KD ratio will elute first. The classical elution would produce the solute order 1, 2, 3, 4 (Figures 3 and 4D). The dual-mode method will produce the solute elution order 1, 2 (Figure 3), and, after reversing the phase role, 4 and 3 (Figure 4E). Equation 9 also shows that the volume difference between peaks will decrease quadratically with increasing KD ratio. Equation 4 can still be used to express the resolution factor. The peak width at base, W ′i, obtained after reversing the phase role is a combination of the band broadening produced in the forward and reverse steps. Due to the additivity of the variances, we can form
W ′i ) 4x(σCM2 + σ ′i2 )
(10)
It was determined that the band broadening inside the column could be expressed by eq 6 for any phase ratio and distribution constant, and then the band broadening due to the forward mode is exactly equal to that due to the reverse mode and eq 6 gives Figure 4. Different elution modes in CCC. (D) Classical mode following Figure 3 illustrations; (E) dual-mode method; the phase roles are reversed; (D) elution-extrusion method; the whole content of the column is extruded out of the column.
To solve the very large elution volume problem, the classical elution mode can be performed for a limited time only, using a given volume, VCM, of mobile phase. Equation 2 gives the position of the remaining solutes (with known KDs) inside the column. They will be eluted using the unique CCC fact that the stationary phase is a liquid. Two methods can be used. Dual-Mode Method. In the dual-mode method, the role of the phases is reversed and the solutes still in the column are eluted in what was the stationary phase during the first step (Figure 4D and E). Due to the density difference between the phases, the direction of the flow is also reversed.1,12-14 After a volume, VCM, has been passed in the classical elution mode, the mobile phase is stopped, not the rotor rotation. The pump is rinsed with the other phase. The flowing direction is reversed and what was the stationary phase is now pumped in the CCC column (Figure 4E). The prime sign will be used for all volumes of the new mobile phase. A volume V ′Ri is needed to elute the solute in the opposite direction. It was demonstrated23,24 that this volume depended only on the solute distribution constant and on the volume of mobile phase used in step 1 (classical elution):
V ′Ri ) VCM /KDi
(8)
The selectivity criterion is obtained by differentiating eq 8:
dV ′R /dKD ) -VCM /KD2
(9)
(23) Menges, R. A.; Bertrand, G. L.; Armstrong, D. W. J. Liq. Chromatogr. 1990, 13, 3061-3074. (24) Gluck, S. J.; Martin, E. J. J. Liq. Chromatogr. 1990, 13, 3559-3573.
5890
Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
W ′i ) 4VCx2(VCM/NVRi )
(11)
in which VRi is the total mobile-phase volume that would be needed in the classical mode to elute the solute (eq 1). Unfortunately, the dual-mode resolution equation, obtained combining eqs 4, 8, and 11, is not simple:
R′s )
1 2VC
x
(
NVCM 2
1/KD1 - 1/KD2
x1/VR1 + x1/VR2
)
(12)
It shows that the dual-mode resolution factor increases with the square root of the mobile-phase volume used in the forward (classical) way. Of course increasing the VCM volume will increase the number of solutes eluted during the classical elution step, leaving less solutes inside the CCC column. Since the VRi volumes are necessarily higher than VCM, eq 12 shows that the dual-mode resolution factor will be lower that that obtained directly (eq 6). The highly positive point of the dual-mode method is quickness as illustrated by Table 1. Compound 6, with a KD ratio of 14 needs 1.4 L of mobile phase to be eluted classically; that means that the experiment duration will last almost 8 h at 3 mL/min. Compound 6 is eluted with only 130 mL of mobile phase in the first (classical) step and 9 mL of “stationary” phase in the second step of the dualmode method (duration ∼1 h; Table 1 and Figure 4). As predicted by eq 12, the dual-mode resolution factors are almost 4 times lower than those obtained in classical elution mode: R′s 3-4 ) 1.8, R′s 4-5 ) 0.54, and R′s 5-6 ) 0.29 compared to Rs of 4.1, 2.1, and 1.4, respectively (Table 1). Elution-Extrusion Method. The method name says it all; its first step is a classical elution of part of the solutes (Figure 3). Next, the whole content of the CCC machine, including mobile and stationary phases, and the remaining solutes, is extruded (Figure 4F). Equation 2 gives the positions of the noneluted solutes after the volume VCM of mobile phase has been pumped
Table 2. Chromatographic Parameters Obtained in the Separation of the Table 1 Hypothetical Mixture in the Normal-Phase Modea classical normal-phase elution
dual mode
compd
1/KD
VR (mL)
W (mL)
Rs
VR (mL)
W (mL)
6 5 4 3 2 1
0.07 0.1 0.17 0.5 1.25 5
27 30 37 70 145 520
6.3 6.9 8.5 16.2 33.5 120
0.43 0.87 2.7 3.0 4.9
27 30 37 150 90 60
6.3 6.9 8.5 33.1 23.0 12.2
elution-extrusion Rs 0.43 0.87 2.1 1.7 2.3b
VR (mL)
W (mL)
27 30 37 84 129 158
6.3 6.9 8.5 23 16 8.5
Rs 0.43 0.87 3.0c 2.2 2.4
a 1/K , polar/apolar solute distribution ratio; V , total solute retention volume; W, theoretical peak width at peak base; R , resolution factor; D R s machine volume, VC ) 120 mL; polar stationary-phase volume, VS ) 100 mL; apolar mobile-phase volume, VM ) 20 mL; classical elution for dual b mode and elution-extrusion, VCM ) 50 mL, eluting solutes 6, 5, and 4. Efficiency N ) 300 plates. Resolution between peak 4 (normal-phase mode) and peak 1 (reversed-phase mode). c Resolution between peak 4 (normal-phase mode) and peak 3 (extruded).
in the CCC machine in the classical mode. A volume V ′′Ri is needed to push solute i out of the column:
V ′′Ri ) VC (1 - VCM /VRi)
(13)
The retention volume difference between two adjacent solutes versus the distribution constant difference is
dV ′′R/dKD ) VC VS VCM/V 2Ri
(14)
If eq 14 shows that the solute elution order will be preserved (positive derivative), a quadratic decrease of the solute retention volume difference could be expected for hydrophobic solutes with high KD ratios implying high VRi volumes. This is similar to the dual-mode method (eq 9). The good point is band broadening. Assuming that the extrusion process moves the two phases without further broadening the bands, the peak widths at base inside the column are relatively narrow as expressed by eq 6 and shown in Figure 1. The resolution factor, R′′s, is simply expressed by combining eqs 4, 6, and 13:
Rs )
(x x )
xN V 2 x CM
1 VR1
1 VR2
(15)
R′′s also increases with the square root of the mobile-phase volume passed in the classical mode. Table 1 lists the chromatographic parameters corresponding to the set of six compounds. The maximum retention volume is obtained for solute 6 extruded with 239 mL (130 mL in classical mode plus 109 mL of extruded volume) in ∼80 min at 3 mL/min. The resolution factors are still lower than those obtained with the classical mode. Comparing with the dual-mode method, the retention volumes have the same order of magnitude but a 3-fold enhancement of the resolution factors is observed (Table 1). Adding the fact that the elutionextrusion method does not involve a change in the direction of the flow so that it could be run continuously, it seems to be a clear improvement in the CCC elution of highly retained solutes. Normal-Phase or Reversed-Phase Elution? For historical reasons,25 normal-phase chromatography refers to an apolar (25) Laitinen, H. A., Ewing, G. W., Eds. History of Analytical Chemistry; American Chemical Society: Washington, DC, 1977.
mobile phase used with a polar stationary phase. Reversed-phase liquid chromatography uses an apolar stationary phase associated with a polar mobile phase. All CCC methods described so far can be used in either normal- or reversed-phase mode since any liquid phase of the system selected can be the mobile phase. It should just be pointed out that the partition ratio, KD, is defined as the ratio of the solute concentration in the stationary phase to that in the mobile phase. It means that KD in normal-phase classical elution mode will become 1/KD in reversed-phase classical elution mode. Table 2 presents the retention volume obtained with the six solutes of Table 1 but using an apolar mobile phase (normal-phase mode) for the classical elution and for the first step of the dualmode or elution-extrusion methods. In that case, the elution order is reversed compared to Table 1. The most hydrophobic compound 6 elutes first. It is clear that compounds 6-4 are poorly separated and compounds 3-1 are well separated in classical elution but with large retention volume (∼600 mL or more than 3 h at 3 mL/min). The dual-mode and elution-extrusion method will not change the poor separation of compounds 6-4 since they both use the identical classical mode in their first step. The reduction of the experiment duration is clear for the elution of the three less hydrophobic solutes. With only one-fourth of phase volume (or time), the two methods are able to resolve completely the three solutes (Table 1). Here also, the elution-extrusion method produces better resolution factors compared to the dualmode method. RESULTS AND DISCUSSION A set of five alkylbenzene homologues was used to evaluate the elution-extrusion method and compare the results with those obtained using the dual-mode method. These solutes were selected because their distribution ratios in the heptane/methanol/ water biphasic liquid system were extensively studied by CCC.26 Two CCC machines of different types were tested. One was a Itotype hydrodynamic machine made of coiled Teflon tubing without any rotary seals.19 There is a continuous contact between the mobile and the stationary phases throughout the coiled tubes.1-3,12 The second machine was of the hydrostatic type, the mobilestationary phase contact occurs only in the channels. The connecting ducts contain the mobile phase only.1-3,13 (26) Berthod, A.; Bully, M. Anal. Chem. 1991, 63, 2508-2512.
Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
5891
Table 3. Chromatographic Parameters Corresponding to the Dual-Mode Separation of Figure 5a compound
KD
benzene 2.0 toluene 3.0 octylbenzene 31 butylbenzene 7.7
tR VR VRb W Wc N (min) (mL) (theor) (mL) (theor) plates 50 69 6 23
100 138 6 23
102 138 6.4 25
33 48 12 14
9 17
150 130 4700d 4000d
a K , heptane/methanol solute distribution ratio; V , Figure 5 solute D R retention volume; W, peak width at peak base; machine volume, VC ) 58 mL; heptane stationary phase volume, VS ) 43 mL; methanol/water (90/10% v/v) polar mobile-phase volume, VM ) 15 mL; stationary-phase retention ratio, Sf ) 74%. Classical elution for VCM ) 200 mL; efficiency N ) 150 plates measured on benzene. b Calculated using eq 1 for benzene and toluene and 8 for the dual-mode elution. c Calculated using eq 6 and 10. d Calculated using N ) 16 (VR/W)2 with the V ′Ri + VCM retention volume.
Figure 5. Dual-mode method. (A) Classical mode for 100 min at 2 mL/min methanol/water 90/10% v/v flowing in the head-to-tail direction. Solutes eluted: benzene (80 µg) and toluene (70 µg). (B) Dual mode for 40 min at 1 mL/min heptane flowing in the tail-to-head direction. Solutes eluted: octylbenzene (50 µg) and butylbenzene (60 µg). Hydrodynamic CCC machine mounted with one spool only, VC ) 58 mL; rotor rotation, 800 rpm; injection volume, 1 mL. UV detection at 254 nm. See Table 3 for all other experimental data.
Dual-Mode Method. Figure 5A shows an actual chromatogram of benzene and toluene eluted in the classical mode with a methanol/water mobile phase in the head-to-tail direction and a heptane stationary phase.1-3 The hydrodynamic machine was used with only one coil connected (machine volume 58 mL). The numerous spikes seen on the chromatogram are produced by microdroplets of heptane passing through the detector cell. Indeed, a constant and very low “bleed” of stationary phase is commonly observed in hydrodynamic CCC with the heptane/ methanol/water liquid system.26 This “bleed” can be measured since the stationary phase, heptane, is not miscible with the mobile phase, methanol/water (90/10% v/v). It formed a thin layer in a graduated cylinder collecting the column effluent. The stationaryphase leak was estimated to be as low as 10 µL/min (less than 1 mL of heptane was lost during a 1.5-h experiment) at 2 mL/min of mobile-phase flow rate and 800 rpm rotor speed. The Gaussian peaks can still be seen clearly. Figure 5B shows the second chromatogram obtained after the phase roles were reversed. The mobile phase is now the heptane phase flowing at 1 mL/min in the opposite (tail-to-head) direction.1-3 The two more hydrophobic alkylbenzene compounds are eluted in less than 25 mL. With the heptane mobile phase, the noise is much reduced because the flow rate is half that of Figure 5A and methanol, the stationary phase tends to accumulate to the tail side 5892 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
of the machine.1,3 Table 3 lists all the parameters corresponding to Figure 5. The theoretical values were calculated using eqs 8-11 taking the experimental efficiency obtained in Figure 5A as the cell number in the dual-mode step. The dual-mode method has been used for years, and these results fully confirm the works of several other authors.1-3,14,23,24 Similar results were obtained using the hydrostatic machine. Elution-Extrusion Method. Detection was the main problem encountered trying to verify experimentally the validity of the theoretical treatment of the new elution-extrusion method. Since by definition different nonmiscible liquid phases will pass through the detector, a dramatic noise will hinder all peak recognition. It was shown that excellent detection was possible by adding a clarifying agent with a postcolumn addition setup. 2-Propanol was used to smooth UV detection by dissolving the heptane microdroplets in a propanol/methanol/water solution.26 It was not possible to use a clarifying agent because it works by making the mobile and stationary phases mutually soluble, but we wanted to control accurately their respective volumes. After countless experiments done reconstructing chromatograms after collection of the phases and analyses, it was found that the best way to reduce the detector noise was to keep the rotor running and to push the column content with the stationary phase for the extrusion step. When heptane, the stationary phase, was used to extrude the column content, it rapidly pushed the polar phase contained in the column and UV detection was possible as soon as a clear apolar phase passed through the detector. Hydrodynamic Machine. Figure 6A shows an actual chromatogram obtained using the hydrodynamic machine with its three coils connected (column volume 175 mL). Benzene was eluted in the classical elution mode. Its peak width allowed calculating a 530-plate efficiency. This is more than 3 times higher than the efficiency obtained with the benzene peak under the same conditions but with only one coil connected (Figure 5A). As with any other chromatographic technique, the column efficiency is proportional to the column “length”. The 150-plate count of Figure 5A with one coil corresponds to ∼1 theoretical plate/tubing turn. Figure 6A, with 3 coils or 400 tubing turns, corresponds to 1.3 theoretical plates/tubing turn. Also, with three coil connected (Figure 6A), the detector noise was not perfect but not as bad as it was with only one coil connected (Figure 5A). Table 4 lists all the chromatographic parameters corresponding to the Figure 6
Table 4. Chromatographic Parameters Corresponding to the Elution-Extrusion Separation of Figure 6 compound
KD
tR VR VRb W Wc Nd (min) (mL) (theor) (mL) (theor) plates
6A: Hydrodynamic 3-Coil CCC Machinea benzene 2.0 67 268 273 46 toluene 3.0 88 352 346 22.5 ethylbenzene 4.0 95 380 386 19 butylbenzene 7.7 103 412 444 14 octylbenzene 31 109 436 495 12 6B: benzene 2.0 toluene 3.0 butylbenzene 7.7 octylbenzene 31
Hydrostatic CPC Machinee 46 184 183 52 57 228 232 24 63 252 287 13 67 268 312 8
30 26 19 10
530 3900 6300 13000 21000
27 16 8
200 1400 6200 18000
a K , heptane/methanol solute distribution ratio; V , Figure 6 solute D R retention volume; W, peak width at peak base; hydrodynamic 3-coil machine volume, VC ) 175 mL; heptane stationary-phase volume, VS ) 95 mL; methanol/water (90/10% v/v) polar mobile-phase volume, VM ) 80 mL; stationary-phase retention ratio, Sf ) 54%. Classical elution for VCM ) 340 mL at 4 mL/min; efficiency N ) 530 plates measured on benzene; rotor rotation speed, 800 rpm. b Calculated using eq 1 for benzene and eq 13 for the dual-mode elution. c Calculated using eq 6. d Calculated using N ) 16 (V /W)2 with (V ′′ + V R Ri CM) for the retention volume and the experimental peak width at base. e Hydrostatic CPC machine with 1015 channels, VC ) 101 mL; heptane stationary-phase volume, VS ) 80 mL; methanol/water (90/10% v/v) polar mobile-phase volume, VM ) 21 mL; stationary-phase retention ratio, Sf ) 79%. Classical elution for VCM ) 220 mL at 4 mL/min; efficiency N ) 200 plates measured on benzene; rotor rotation speed, 1000 rpm.
Figure 6. Elution-extrusion method. (A) Chromatogram obtained with a hydrodynamic machine (VC ) 175 mL). Flow rate, 4 mL/min polar phase in the head-to-tail direction during the elution step (85 min), switched to heptane used as the extruding agent in the headto-tail direction (35 min), 800 rpm. Peak identification: (1) benzene (80 µg), (2) toluene (70 µg), (3) ethylbenzene (70 µg), (4) butylbenzene (60 µg), and (5) octylbenzene (50 µg); injection volume, 1 mL in heptane. (B) Chromatogram obtained with a hydrostatic CPC machine (VC ) 101 mL). Flow rate, 4 mL/min with polar phase (descending, head-to-tail direction, elution step, 55 min) and heptane (same direction, extrusion step, 20 min), 1000 rpm. Peak identification: (1) benzene (80 µg), (2) toluene (70 µg), (3) butylbenzene (60 µg), and (4) octylbenzene (50 µg); injection volume, 1 mL in heptane. Detection, UV at 254 nm. See Table 4 for all other experimental data.
experiments. In the extrusion phase, the experimental retention volumes and peak widths were systematically lower that the calculated ones. It was found that a large part (∼70% or 55 mL of the initial polar-phase volume but only 31% of the CCC column volume) of the polar phase was not extruded by the heptane mobile phase pushed in the head-to-tail direction as it should.1-3,12,13 It stayed entrapped in the coils by the centrifugal field. The whole column content was extruded when the rotor rotation was stopped and the column effluent was collected in fractions. But the accuracy of the measurements of the peak positions and, especially, the peak widths on reconstructed chromatograms was so low that the results were not exploitable. Such phase entrapment is not observed when a hydrodynamic machine is regularly equilibrated, the mobile phase equilibrating (or displacing) the other phase coil after coil.1-3,13 It is observed that there is only a partial displacement when a liquid phase is used in the “wrong” way in an equilibrated CCC column.
Hydrostatic Machine. Figure 6B shows an actual chromatogram obtained using the hydrostatic CPC machine. Benzene was also eluted in the classical way to evaluate the machine efficiency. Its peak shows a 200-plate efficiency (Table 4). It means that about five channels are actually needed to obtain a single theoretical plate. It is known that the mixing efficiency of hydrostatic machine is lower than that of hydrodynamic machine.9,13,27 The phase retention is much better. Comparing the Sf ) VS/VC values at the same flow rate for the same liquid system, they are 54 and 79% for the hydrodynamic and hydrostatic machines, respectively (Table 4). Figure 6 clearly shows that the last eluted peak is the thinnest, an unusual situation in LC. However, it was also found that the peaks were even thinner than predicted with shorter retention times or volumes (Table 4). The reason is again that the major part (∼90% or 19 mL) of the polar phase remained entrapped in the channels (this is only 19% of the CCC column volume). Such phase entrapment was already observed with CPC machines. It was actually used to work with reduced amount of octanol phase in octanol/water distribution ratio measurements.28 The fact that 70-90% of the initial mobile-phase volume remain entrapped in the CCC column should be taken into account in an exact modeling: it reduces both the extrusion volumes and peak widths (Table 4). Unfortunately this volume is critically dependent on experimental conditions (CCC column type, physicochemical characteristics of the biphasic liquid system, rotor rotation speed, and flow rate). From a practical point of view, the equations given in this work are valid, giving the correct peak positions and peak (27) Marchal, L.; Foucault, A. P.; Patissier, G.; Rosant, J. M.; Legrand, J. In Countercurrent Chromatography, The Support-Free Liquid Stationary Phase; Berthod, A., Ed.; Comprehensive Analytical Chemistry XXXVIII; Elsevier: Amsterdam, 2002; pp 115-157. (28) Berthod, A.; Han, Y. I.; Armstrong, D. W. J. Liq. Chromatogr. 1988, 11, 1441-1456.
Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
5893
widths, only when no polar phase remains unextruded (Figure 4F). Both peak positions and widths are overestimated if some “initial” mobile phase stays inside the CCC column during the extrusion step. The expected peaks elute 1-15% earlier than expected with a thinner shape (Table 4), most often producing a resolution better than expected. Indeed, the extruded solutes have necessarily a low affinity for the entrapped liquid phase since the elution-extrusion method was used because this liquid phase was not able to elute them in the classical CCC mode. A Breakthrough for CCC Separations? Using the elutionextrusion method, it was actually found that the fact that a large part of the initial mobile phase remained entrapped in the machines, when the field stayed on, was a good point: the peaks eluted faster and thinner (Table 4) and the machine was more rapidly reequilibrated and ready for the next run. Indeed, once all remaining solutes were extruded off the column, the heptanephase pump was stopped, the polar-phase pump was resumed, and the column was reequilibrated in minutes. For both the dualmode and the elution-extrusion methods, it is absolutely certain that no solutes can stay in the column. All solutes are eluted either during the classical elution step with the polar liquid phase (less hydrophobic or more polar compounds in our examples) or during the step with the apolar liquid phase (that was the stationary phase in the initial step). The advantage of the elution-extrusion method compared to the dual-mode method is its continuity: the CCC column inlet is just switched from the mobile-phase liquid (elution) to the “stationary”-phase liquid (extrusion) and the chromatogram is obtained with a liquid flow always in the same direction (Figure 6). To reequilibrate the CCC column, the mobile-phase pump is
5894
Analytical Chemistry, Vol. 75, No. 21, November 1, 2003
switched back for few minutes on the column inlet. The other advantage of the method is its resolution power: an apparent efficiency never seen before in CCC can be obtained. Five alkylbenzene homologues were separated with baseline resolution and an apparent efficiency close to 20 000 theoretical plates (Figure 6). Detection is the main problem of the method. UV detectors do not like immiscible liquids flowing in. An evaporative light scattering detector will be the solution with nonvolatile solutes such as tannins, sugars, fatty acids, or proteins. In many cases, a fraction collector associated with fraction analyses will be necessary. Once the peak position is known, the method is very repeatable. ACKNOWLEDGMENT A.B. thanks the French Centre National de la Recherche Scientifique (UMR2496 ERS2007 FRE2394) for continuous support. S.C.-B. and A.B. thank the European Community for the Marie-Curie Fellowship “Improving the Human Research Potential” HPMF-CT-2000-00440 that supported the two-year S.C.-B. sabbatical stay. M.J.R.-A. thanks the Ministerio de Ciencia y Tecnologı´a (Madrid) for a grant supporting her stay in A.B.’s group. Also, A.B. thanks the European Community for the INTAS Grant 000-00782 “Countercurrent Chromatography”.
Received for review May 20, 2003. Accepted August 20, 2003. AC030208D
