Anal. Chem. 1997, 69, 344-348
Capillary Electrokinetic Chromatography Employing p-(Carboxyethyl)calix[n]arenes as Running Buffer Additives Sixun Sun, Michael J. Sepaniak,* Jian-She Wang,† and C. David Gutsche†
Department of Chemistry, University of Tennessee, Knoxville Tennessee 37996-1600
Calixarenes, a class of macrocyclic phenolic compounds with a basket-like shape, are used as capillary electrophoresis reagents for separations of native and substituted polycyclic aromatic hydrocarbons. The p-(carboxyethyl)calix[n]arenes reported herein are a series of charged, moderately water soluble macrocyclic molecules that can form complexes with neutral molecules. Electrokinetic chromatographic separations are based on the differential distribution of molecules between a running buffer phase, which is transported by electroosmotic flow, and an electrophoretically mediated calixarene. The size of the calixarene influences separation performance, illustrating the importance of cavity size and geometry in the complexation process. p-(Carboxyethyl)calix[7]arene provides the best efficiency (>105 plates/m) and selectivity in these studies. The influences of pH, organic solvent, and field strength on elution range, capacity factors, efficiency, and selectivity are also reported. In general, capacity factors are rather low, but the high charge-tomass ratios of certain calixarenes produce relatively wide elution ranges. Molecular modeling data and solubility data are used to interpret the observed selectivity. The addition of macrocyclic reagents to capillary electrophoresis (CE) running buffers has been shown to impart unique selectivity in separations of numerous structurally similar solutes. The interaction of the solute with the macrocyclic compound often involves the formation of an inclusion complex. In this regard, cyclodextrins (CDs) have proven to be extremely useful reagents, with many examples of chiral and achiral separations in conventional CE.1-5 Cyclodextrins have also been effectively employed in capillary electrokinetic chromatography when used as running buffer components in micellar electrokinetic capillary chromatography (MECC)5,6 and in the development of a dual (neutral and charged) CD phase form of electrokinetic chromatography.7 Capillary electrokinetic chromatography separations are based on
the differential distribution of solutes between a running buffer phase that is transported with electroosmotic flow (EOF) and an electrophoretically mediated secondary phase (e.g., charged micelles) or running buffer additive. Other examples of macrocyclic compounds used as CE running buffer additives include crown ethers8 and antibiotics.9 Recently, a class of macrocyclic phenolic and resorcinolic compounds called calixarenes have been demonstrated as CE reagents. The calixarene structure is often described as basketlike, although relatively free rotation around the methylene groups that link the phenol or resorcinol groups in the macrocycle yields numerous calixarene conformations.10 Shohat and Grushka studied the effects of p-sulfonatocalix[6]arene on the migration of certain chlorinated phenols and benzenediols.11 Bachmann and co-workers utilized calix[4]resorcarene tetraanions to separate polycyclic aromatic hydrocarbons (PAHs).12 In that work, an alkyl (C11)-substituted calix[4]resorcarene was shown to be the most useful calixarene and exhibited elution characteristics that resemble those of reversed phase HPLC. Certain O-alkyl-substituted p-sulfonatocalix[6]arenes have been shown to form micelles.13 Thus, the separations observed by Bachmann could possibly involve micelle solubilization rather than inclusion complexation. Both of these cited examples involve separations of neutrals via differential association of the solute with a highly charged calixarene. Thus, resolution (Rs) and capacity factor (k ′, amount of solute associated with calixarene/amount in running buffer) are properly described by the electrokinetic chromatography relationships (eqs 1 and 2) first derived by Terabe et al. for MECC.14 In these equations, N is plate count, given by column
Rs )
1 - t0/tcal xN R - 1 k ′2 4 R 1 + k ′2 1 - (t0/tcal)k ′1 k′)
tr - t0 t0(1 - tr/tcal)
(1)
(2)
length to the detection zone/plate height (L/H), and R is the †
Permanent address: Department of Chemistry, Texas Christian University, Fort Worth, TX 76129. (1) Terabe, S.; Ozaki, H.; Otsuka, K.; Ando, T. J. Chromatogr. 1985, 332, 211217. (2) Fanali, S. J. Chromatogr. 1991, 545, 437-444. (3) Sepaniak, M. J.; Cole, R. O.; Clark, B. K. J. Liq. Chromatogr. 1992, 15, 1023-1040. (4) Copper, C. L.; Davis, J. B.; Cole, R. O.; Sepaniak, M. J. Electrophoresis 1994, 15, 785-792. (5) Novotny, M.; Soini, H.; Stefansson, M. Anal. Chem. 1994, 66, 646A-655A. (6) Copper, C. L.; Sepaniak, M. J. Anal. Chem. 1994, 66, 147-154. (7) Sepaniak, M. J.; Copper, C. L.; Whitaker, K. W.; Anigbogu, V. C. Anal. Chem. 1995, 67, 2037-2044.
(8) Kuhn, R.; Erni, F.; Bereuter, T.; Hausler, J. Anal. Chem. 1992, 64, 28152820. (9) Armstrong, D. W.; Rundlett, K.; Reid, G. L. Anal. Chem. 1994, 66, 16901695. (10) Gutsche, C. D. In “Calixarenes” Monographs in Supramolecular Chemistry; Stoddart, J. F., Ed.; Royal Society of Chemistry: Cambridge, U.K., 1989. (11) Shohat, D.; Grushka, E. Anal. Chem. 1994, 66, 747-750. (12) Bachmann, K.; Bazzanella, A.; Haag, I.; Han, K.; Arnecke, R.; Bohmer, V.; Vogt, W. Anal. Chem. 1995, 67, 1722-1726. (13) Shinkai, S.; Mori, S.; Koreishi, H.; Tsubaki, T.; Manabe, O. J. Am. Chem. Soc. 1986, 108, 2409. (14) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 57, 834-841.
344 Analytical Chemistry, Vol. 69, No. 3, February 1, 1997
S0003-2700(96)00602-6 CCC: $14.00
© 1997 American Chemical Society
Figure 1. Structure of p-(carboxyethyl)calix[4]arene.
selectivity factor (k ′2/k ′1). The time t0 is the effective retention time of an unretained solute, tcal is the effective retention time of the charged calixarene, and tr is the retention time of the solute. As with the MECC technique,14 Rs is optimal for relatively low values of k ′ (note last two terms in eq 1), a result of a finite elution range which is bounded by t0 and tcal. An advantage cited in ref 12 and illustrated herein is that calixarenes can have a relatively large charge-to-mass ratio; thus, the observed elution range can be larger than that normally observed in MECC. This report concerns an evaluation of several p-(carboxyethyl)calix[n]arenes as reagents for CE. The calixarenes vary in size (n ) 4-8); p-(carboxyethyl)calix[4]arene is depicted in Figure 1. Moderate size PAHs and polar-substituted PAHs are used as test solutes. The effects that experimental parameters such as calixarene size, pH, organic solvent, and field strength exert on mobility, elution range, migration of the test analytes, and efficiency are reported. Molecular modeling depictions and interaction energy data for calixarene-analyte complexes and solubility data are used to interpret the observed separation behavior. EXPERIMENTAL SECTION Apparatus. Separations were performed using a laboratorybuilt CE system. The system consists of a Hippotronics (Brewster, NY) Model 840A high-voltage power supply and fused silica capillary (50 µm i.d., 365 µm o.d., typically 45 cm total length, 37.5 cm to the detection zone) from Polymicro Technologies, Inc. (Phoenix, AZ). Detection for the studies was performed using a Linear (Reno, NV) Model 204 UV-visible detector operated at 340 nm. A Kipp and Zonen (Delft, Holland) Model BD40 chart recorder was used for collection of all data. Reagents. The p-(carboxyethyl)calix[n]arenes were synthesized as previously described15,16 from p-tert-butylcalix[n]arenes17 via the quinonemethide route, which involves de-tert-butylation, aminomethylation with a dialkylamine and formaldehyde, quaternization with methyl iodide, and treatment with diethyl sodiomalonate, followed by hydrolysis and concomitant decarboxylation. Pyrene (Py) and 9-nitroanthracene (9NO2-An) were purchased from Aldrich (Milwaukee, WI). 1-Aminopyrene (1NH2-Py) was purchased from Sigma (St. Louis, MO). 2-Aminoanthracene (15) Gutsche, C. D.; Nam, K. C. J. Am. Chem. Soc. 1988, 110, 6153-6162. (16) Gutsche, C. D.; Alam, I. Tetrahedron 1988, 44, 4689. (17) Gutsche, C. D.; Iqbal, M. Org. Synth. 1990, 68, 234. Gutsche, C. D.; Dhawan, B.; Leonis, M.; Stewart, D. Org. Synth. 1990, 68, 238. Munch, J. H.; Gutsche, C. D. Org. Synth. 1990, 68, 243.
(2NH2-An) was purchased from Pfaltz and Bauer (Stamford, CT). Stock solutions of each solute (1 × 10-2 M) were made in acetonitrile and then diluted in running buffer. The running buffer consisted of 10 mM sodium phosphate, 6 mM sodium tetraborate, and 30-42% (v/v) acetonitrile. pH values were adjusted by using concentrated NaOH or H3PO4 solution. Weighed amounts of p-(carboxyethyl)calix[n]arenes were dissolved in 3.0 mL of running buffer and ultrasonicated for 10 min, yielding calixarene concentrations of 2.0-4.0 mM. Procedures. A region of the capillary polyimide coating was removed to generate a transparent detection “window”. The capillaries were initially conditioned with 0.1 N NaOH. At the start of each day, and whenever the running buffer was changed, the capillaries were rinsed with 0.01 N NaOH, H2O, and running buffer. Samples were injected hydrostatically by elevating the inlet end of the capillary 10 cm and immersing it in the sample for 5-10 s. Separations were performed at 15 kV, except for data contained in the modified Van Deemter plot. The void time (t0) of each run was marked by a solvent disturbance. The value of tcal was estimated from injections of the calixarene. In some instances, secondary calixarene peaks indicated that the injected calixarene was contaminated with higher or lower forms of the macrocycle. Peak identification was based on peak height increase from spiking of one particular solute. The solubilities of the PAH analytes were determined from absorbance measurements performed using a Hewlett-Packard 8452A UV-visible diode array spectrophotometer (Wilmington, DE). Saturated solutions were prepared by adding 15 mg of analyte to 10 mL of running buffer (pH 8, 30% acetonitrile), ultrasonicating for 20 min, and then filtering. The absorbance of the saturated solution and standard solutions of the corresponding PAH were measured at λmax. Solubilities were then calculated on the basis of Beer’s law. Plots presented herein were generated using KaleidaGraph for Macintosh. Molecular modeling studies were performed using SYBYL 6.2 molecular modeling software developed by Tripos Associates, Inc. (St. Louis, MO), and computations were performed on an Evans and Sutherland ESV-3 workstation (Salt Lake City, UT). Molecular models of the p-(carboxyethyl)calix[n]arenes and PAH analytes were constructed using the SKETCH routine of the software and were minimized using GasteigerHuckel charges and 100 000 iterations. Under the pH conditions for our studies, all the (carboxyethyl) groups should be deprotonated, and one of the hydroxyl group could be deprotonated.10 Structures with and without one deprotonated hydroxyl group were constructed and minimized by using the same routine as mentioned above. The protonated and deprotonated forms exhibited the same general macrocyclic shapes. The structure of p-(carboxyethyl)calix[7]arene was also solvated using the SYBYL droplet method. Again, it was determined that solvated and unsolvated calixarene exhibited the same general shape. The SYBYL DOCKING program was used to calculate the energy of selected calixarene-analyte complexes and to create depictions of the complexes. This was accomplished by treating the calixarene as a site and the PAH analyte as a ligand. The SYBYL MINIMIZE-DOCK mode (using 100 iterations) was used to optimize the ligand-site configuration and to determine the energy of the ligand-site complex. This operation was performed 25 times. Under these conditions, the energy converged (after about 20 operations) on energy values that differed by less than 1%. Analytical Chemistry, Vol. 69, No. 3, February 1, 1997
345
Table 1. Retention Time, Mobility (µ, cm2 s-1 V-1), k ′, and r Constants of 2NH2-An, 1NH2-Py, 9NO2-An, and Py for Running Buffers (pH 8, 30% (v/v) Acetonitrile) Containing 4 mM p-(Carboxyethyl)calix[n]arenes (Except Calix[8]arene < 2 mM) calix[4]- calix[5]- calix[7]arene arene arene 2NH2-An
Figure 2. Modified Van Deemter plots for 2NH2-An ([) and Py (b) using a running buffer (see Experimental Section) with pH adjusted to 8, 42% (v/v) acetonitrile, and 4 mM p-(carboxyethyl)calix[7]arene.
Hydrogen bonding (dynamic) was observed using the VIEW routine of the SYBYL software. The relative interaction energies of the complexes were then determined as the best ligand-site energy minus the energies of the MINIMIZED free ligand (analyte) and site (calixarene). Safety. Some PAHs are known carcinogens, so caution should be exercised when working with them. PAH solutions were prepared under a ventilated hood. Disposable latex gloves were worn while working with these compounds. Care was taken to dispose of waste solutions properly. RESULTS AND DISCUSSION The first term in eq 1 illustrates the importance of high efficiency (large N) in determining Rs. Plate count is often limited solely by axial diffusion in CE, whereas capillary electrokinetic chromatography involves solute transfer between primary (running buffer) and secondary (e.g., micellar or calixarene) phases and, as such, is influenced by resistance to mass transfer. Increasing the field (E) in CE is analogous to increasing mobile phase velocity in HPLC. At low E, separation times are long, and axial diffusion tends to dominate; i.e., there is an inverse relationship between plate height (H) and E. At high E, there is less time for axial diffusion, but problems with slow mass transfer are exacerbated; i.e., there is a direct relationship between H and E. If the field is increased to the point of excessive thermal load, a thermal gradient can form in the capillary, and this can cause efficiency to degrade dramatically.18 A plot of H versus E (essentially a Van Deemter plot) often shows a distinctive optimum field in capillary electrokinetic chromatography.18 This is demonstrated in Figure 2 for the solutes Py and 2NH2-An over a field range of approximately 100-400 V/cm. Both solutes exhibit an optimum at about 250 V/cm (2NH2-An efficiency ∼ 200 000 plates/ m). It is somewhat puzzling that the Py optimum is significantly more distinct. This behavior may be related to the very poor solubility (see below), and hence the low degree of solvation, of Py relative to that of 2NH2-An. This could lead to a lower diffusion coefficient and slower running buffer-calixarene mass transfer (18) Sepaniak, M. J.; Swaile, D. F.; Powell, A. C.; Cole, R. O. In Capillary Electrophoresis Theory and Practice; Grossman, P. D., Colburn, J. C., Eds.; Acadamic Press, Inc.: New York, 1992; Chapter 6.
346 Analytical Chemistry, Vol. 69, No. 3, February 1, 1997
tr (min) k′ R 1NH2-Py tr (min) k′ R 9NO2-An tr (min) k′ R Py tr (min) k′ EOF t0 (min) µeof (10-4) calix[n]arene tcal (min) µcal,net (10-5)
5.3 0.14 1.0 5.3 0.14 1.0 5.3 0.14 1.0 5.3 0.14 5.0 3.8 9.2 20
5.2 0.18 1.7 5.6 0.32 1.3 5.9 0.42 1.3 6.2 0.53 4.6 4.1 18 11
5.7 0.27 1.5 6.1 0.40 1.3 6.4 0.50 1.2 6.7 0.61 4.8 3.9 19 9.7
calix[8]arene 5.0 0.084 1.5 5.2 0.13 1.2 5.3 0.15 1.0 5.3 0.15 4.6 4.1 -120 -1.5
Table 2. Retention Time, Mobility (µ, cm2 s-1 V-1), k ′, and r Constants of 2NH2-An, 1NH2-Py, 9NO2-An, and Py Using 4 mM p-(Carboxyethyl)calix[7]arene (30% (v/v) Acetonitrile) at Different pH Values
2NH2-An 1NH2-Py 9NO2-An Py EOF calix[7]arene
tr (min) k′ R tr (min) k′ R tr (min) k′ R tr (min) k′ t0 (min) µeof (10-4) tcal (min) µcal,net (10-5)
pH ) 8
pH ) 9
pH ) 10
5.7 0.27 1.5 6.1 0.40 1.3 6.4 0.50 1.2 6.7 0.61 4.8 3.9 19 9.7
6.0 0.17 1.4 6.4 0.24 1.2 6.7 0.30 1.2 7.1 0.37 5.1 3.7 -130 -1.4
6.5 0.13 1.6 7.0 0.20 1.2 7.3 0.25 1.3 7.8 0.32 5.7 3.3 -54 -3.4
for the 2NH2-An, which could account for the smaller absolute low- and high-field slopes for that analyte in the plot. Figure 3 and Tables 1 and 2 provide information on the retention characteristics of the four calixarenes for the PAH analytes. The effects of changes in running buffer conditions (pH and v/v% acetonitrile) are also shown. As seen in Figure 3, the best separations are obtained with the intermediate calix[n]arenes (n ) 5 and 7). It appears that the p-(carboxyethyl)calix[4]arene cavity is too small for complexation (very little migration modification is observed) and the large p-(carboxyethyl)calix[8]arene cavity does not provide a suitably “snug” fit (modest migration modification and poor selectivity are observed). For example, the molecular modeling interaction energy (see Experimental Section) for calix[7]arene-2NH2-An is about 35 kcal/mol lower (more stable) than that of calix[4]arene-2NH2-An. Moreover, the latter complex has the 2NH2-An largely outside of the basket-like cavity of the calixarene (see below). Another observation is that capacity factors are relatively low with all the calixarene systems. The largest value (Py in the p-(carboxyethyl)calix[7]arene system) is only 0.61 (see Table 1). The low values for k ′ are probably a result of the limited amount of p-(carboxyethyl)calixarene that can be dissolved in the running
Figure 3. Separation of PAH analytes (1, 2NH2-An; 2, 1NH2-Py, 3, 9NO2-An; 4, Py) using a running buffer with pH adjusted to 8, 30% (v/v) acetonitrile, and (A) 4 mM p-(carboxyethyl)calix[4]arene, (B) 4 mM p-(carboxyethyl)calix[5]arene, (C) 4 mM p-(carboxyethyl)calix[7]arene, or (D) 2 mM p-(carboxyethyl)calix[8]arene. The effects of increasing the v/v% of acetonitrile for the system in C are also shown.
buffer (concentrations used in these studies were near the solubility limit) and only modest complexation constants (102103 in some cases) for PAHs with calixarenes.10 Conversely, p-sulfonatocalixarenes exhibit good water solubility,11 probably due to the ionization of the very acidic hydroxyl groups on those calixarenes.19 However, we did not observe any retention for the PAH analytes with p-sulfonatocalix[6]arene. Apparently, the ionized hydroxyl groups inhibit complexation of the p-sulfonatocalix[6]arene with the relatively hydrophobic PAHs. This same effect may account for the reduction of k ′ values for the PAH analytes using p-(carboxyethyl)calix[7]arene as pH is increased from 8 to 10 (see Table 2 and related discussion below). The last term in eq 1 illustrates the effect of elution range on resolution. In MECC systems that employ sodium dodecyl sulfate as the surfactant, the ratio of t0 to the effective retention time of the micellar phase is typically 0.3.18 Under these conditions, optimal k ′ values are less than 5.14 The ratio can be decreased (elution range increased) by manipulating EOF via pH adjustments or the addition of organic solvents. In these studies, the value of t0/tcal varies greatly with the size of the calixarene and with pH. The elution range increases with calixarene size (and hence charge) as shown in Table 1. To determine tcal for p-(carboxyethyl)calix[8]arene, it was necessary to reverse the polarity of the system, since that calixarene actually has an opposing mobility that is larger than that of EOF. As pH is increased, ionization of the hydroxyl groups of the calixarene results in an increase in mobility (note the effect shown in Table 2 for p-(carboxyethyl)calix[7]arene). Increasing the concentration of organic solvent also reduces EOF and extends the elution range. This is demonstrated with the separations shown below Figure 3C (note that the time of the solvent disturbance increases with v/v% (19) Yoshida, I.; Yamamoto, N.; Sagara, F.; Ishi, K.; Shinkai, S. Bull. Chem. Soc. Jpn. 1992, 65, 1012.
acetonitrile). Capacity factors exhibit an expected reduction with increasing organic modifier as well. For example, the k ′ of 2NH2An is reduced from 0.27 to 0.17 as the v/v% acetonitrile is increased from 30 to 42 (this occurs despite an increase in retention time from 5.7 to 8.6 min). The decrease in EOF with increasing pH reported in Table 2 is counter to what is normally observed.16 An independent experiment was conducted to verify this behavior. EOF was measured for a fresh capillary over the pH range of 6-9 and found to increase by 30% (consistent with the normal trend). The same capillary was taken over the same pH range but with buffers that contained 4 mM p-(carboxyethyl)calix[7]arene. This resulted in a 29% decrease in EOF. The capillary was then rinsed with pH 9 buffer that did not contain calixarene, and the EOF was periodically measured. It required about 20 min of rinsing before the EOF increased to a steady value. It appears that, despite a negative charge, the calixarene interacts with the capillary surface and influences EOF. Separation selectivity in CE systems involving CDs is reported to depend on the physiochemical and geometric characteristics of the solute guest and the CD host.4,6 Hydrophobic interactions of the solute with the apolar cavity of the CD can act in concert with polar and hydrogen bonding interactions with hydroxyl groups on the “lip” of the CD. Similar interactions are expected to influence solute-calixarene complexation. Indeed, the PAH analytes were chosen for these studies to highlight these sorts of interactions. Figure 4 contains molecular modeling depictions of lowest energy complexes of p-(carboxyethyl)calix[7]arene with 2NH2-An and 1NH2-Py. The depictions illustrate inclusion that involves both hydrophobic interactions with the cavity of the basket-like calixarene and possible hydrogen bonding. The fact that 1NH2-Py interacts more strongly with the calixarene than the 2NH2-An (note elution order in Figure 3C) may be due to its Analytical Chemistry, Vol. 69, No. 3, February 1, 1997
347
Figure 4. Molecular modeling depictions of the lowest energy complexes of (A) 2NH2-An and (B) 1NH2-Py with p-(carboxyethyl)calix[7]arene (top views) and p-2NH2-An with (C) p-(carboxyethyl)calix[7]arene and (D) p-(carboxyethyl)calix[4]arene (side views).
snugger fit in the cavity, which may enhance hydrophobic interactions. The molecular modeling interaction energy of p-(carboxyethyl)calix[7]arene-1NH2-Py is a full 3.6 kcal/mol lower (more stable) than that of p-(carboxyethyl)calix[7]arene2NH2-An. Hydrogen bonding was identified in both of these complexes. However, the former complex exhibited four hydrogen bonds, while the latter (less stable) complex exhibited six hydrogen bonds. The closeness of the amino groups of these PAH analytes to calix[7]arene can be seen in Figure 4A,B. It appears that hydrophobic interactions exert a stronger influence on inclusion than hydrogen bonding in this case. The poor fit of 2NH2-An in calix[4]arene (see above) is demonstrated in Figure 4D as well. 9NO2-An can act as a hydrogen bond acceptor. However, depictions of its inclusion complex revealed considerable spacing between the nitro group and the calixarene hydroxyl groups. Py
348 Analytical Chemistry, Vol. 69, No. 3, February 1, 1997
does not have the required substitution to hydrogen bond. Based on this discussion, one would expect that Py and 9NO2-An would elute first, followed by 1NH2-An, and then 1NH2-Py. The early elution of the amino-substituted PAHs relative to the other PAHs (see Figure 3C) can probably be attributed to stronger interactions with the running buffer and/or a cathodic mobility due to partial protonation. Solubilities in running buffer of the amino-PAHs are identical (2.3 × 10-4 M) and considerably larger than those of the 9NO2-An (7.1 × 10-5 M) and Py (6.0 × 10-5 M). While specific interactions with the calixarene contribute to the observed selectivity, solubility in the running buffer is also a strong factor in determining the observed elution order in these studies. In conclusion, p-(carboxyethyl)calix[n]arenes are a series of macrocyclic reagents that can be used in capillary electrokinetic chromatography and offer advantages that include the following: (1) basket-shaped conformation that allows formation of complexes with different molecules and exhibits unique separation selectivity; (2) high charge-to-mass ratio that produces wide elution range; and (3) possibility for the synthesis of analogs which have different functional groups (including chiral groups) for further control over selectivity. As drawbacks, the calixarenes produce a relatively large background in spectrophotometic detection and have demonstrated rather low capacity factors. The latter limitation may be improved via the use of other solvent systems and modifications in the calixarene structure to improve solubility. ACKNOWLEDGMENT This work was sponsored by the Division of Chemical Sciences, Office of Basic Sciences, United States Department of Energy, under Grant DE-FG02-96ER14609 with The University of Tennessee, Knoxville. Support was also contributed by Merck and Co., the University of Tennessee, Knoxville Scholarly Activities Research Incentive Fund, the National Science Foundation, and the Robert A. Welch Foundation. Received for review June 19, 1996. Accepted November 22, 1996.X AC960602U X
Abstract published in Advance ACS Abstracts, January 1, 1997.
