Anal. Chem. 1994,66, 4278-4282
Foam Flotation Enrichment of Enantiomers Daniel W. Armstrong,* Eve Y. Zhou, Shushi Chen, Kang Le, and Yubing Tang Department of Chemistry, University of Missouri- Rolla, Rolla Missouri 6540 1
Adsorptive bubble separation methods are known to be useful for processing large amounts of material at a relatively low cost. These techniques have been used to enrich components from both heterogeneous and homogeneous solutions. There is a need for economical process-scale enantiomeric separations. Thus far there has been little evidence to support the feasibility of using an adsorptive bubble process to enrich enantiomers. We demonstrate that foam-forming chiral collectors can be used in conjunction with an inexpensive glass device to enantiomerically enrich some pharmaceutically important compounds as well as derivatized and underivatized amino acids. Factors that appeared to affect this waterbased separation include the following: (a) column length, (b) column geometry and packing, (c) gas flow rate, (d) concentration of the collectorand the racemate, (e) nature of the collector, (f) temperature, (g) pH, (h) reflux time, (i) foam dryness, and (i) the presence of other materials in the sample (e.g., miscible organic solvents, salts, etc.). The chiral collectors used in this study are known to be able to associate with analytes via ligand exchange interactions, hydrophobic inclusion complexation, and hydrogen-bondinginteractions among others. Separations via “adsorptive bubble techniques” (e.g., flotation, foam flotation, froth flotation, floc flotation, floc-foam flotation, etc.) have been used for many years.’ These methods often are used to process large amounts of material (high volume) at a relatively low cost. The separations can be accomplished in either batch or continuous processes. Flotation is an important technique for concentrating sulfide ores2 Most of the flotation methods used by the mining industry involve treating slurries of finely ground material with a collector (i.e., a surfactant that adsorbs or binds to the component of interest thereby giving it a hydrophobic surface). The hydrophobic particles tend to adhere to and be carried by gas bubbles, thereby separating them from other material that does not adhere to the b~bbles.2-~ More recently flotation methods have been used in conjunction with homogeneous solutions, that is, to enrich certain dissolved solutes. The species most often enriched by this method are metal cations or ~helates.~-’OThis process is often referred to as ion flotation and has been reviewed p r e v i ~ u s l y . ~ ~ - ~ ~ (1) Gaudin, A M. Flotation; McGraw H i t New York. 1932. (2)Rossberg, IC S.E., Ed. Flotation of Sulfide Minerals; Elsevier: Amsterdam, 1985. (3) Castro, S. H., Alvared, J. Froth Flotation. In Deoelopments in Mineral Processing; Fuerstenau, D. W., Ed.; Elsevier: Amsterdam, 1988. (4)b e y , M. W., Ed. Flotation Agents and Processes, Technology andA$plications; Noyes Data C o p . , Park Ridge, NJ, 1980. (5)Walling, C.; Ruff, E. E.; Thomton, J. L.J. Chem. Phys. 1957,67, 486. (6)Wace, P. E.:Banfield, D. C. Chem. Process Eng. 1966,47,70. (7)Jome, J.; Rubin, E. Sep. Sci. 1969,4,313. (8)Huang, R C. H.; Talbot, F. D. Can. J. Chem. Eng. 1973,51,709.
4278 Analytical Chemistry, Vol. 66, No. 23, December 1 , 1994
Theoretically, bubble flotation methods could separate or enrich any dissolved substance, provided an appropriate collector with adequate selectivity can be found. During the course of our studies on chiral recognition, it was noticed that aqueous solutions of some molecules used as chiral selectors produced foam upon agitation. This raised the possibility of enantiomerically enriching compounds via an absorptive bubble process. Some chiral selectors were used in previous water-based enantioseparation method^.'^-^^ To our knowledge, these are the first reported enantiomeric enrichments achieved with an adsorptive bubble type of separation technique. EXPERIMENTAL SECTION
Materials. Heptakis(2,&di-@methyl)-~-cyclodextrin was purchased from Sigma (St. Louis, MO), permethyl$cyclodextrin was obtained from Cyclolab R&D Laboratory (Budapest, Hungary), and hydroxypropyl-B-cyclodextrinwas obtained from Consortium Fur Elektrochemische Industrie GMBH. As has been noted before, no commerical derivatized cyclodextrins are pure; they are composed of mixtures of homologues and is0mers.2~Wadarin, D- and L-N-t-BOC-phenylalanine, D- and L-N-t-BOC-chlorophenylalanine, dansyl-D,L-tryptophan,dansyl-D,L-phenylalanine,N-heptylL-hydroxyproline,vancomycin, and digitonin were purchased from Sigma. SQ 30,840 and SQ 31,236 were obtained from Squibb (New Brunswick, NJ). All solvents were obtained from Fisher Scientific (Fair Lawn, NJ). The silica cartridges for solid phase extraction were purchased from Alltech (Deerfield, IL). All HPLC columns were obtained from Advanced Separation Technology Inc. (Whippany, NJ). Pyrex gas dispersion tubes and other glassware were obtained from Fisher Scientific. (9)Kubota, IC;Hayashi, S. Can. J. Chem. Eng. 1977,55,286. (10)Walkowiak, W. Sep. Sci. Technol. 1991,26,559. (11)Rusanov, A I.; Levicer, S. A; Zarov, V. T. PooercknostnageRadelene Vercesto, E d . Chimija: Leningrad, 1981. (12)Golman, A M. Ionnaja Flotacija; Nedra: Moscow, 1982. (13)Grieves, R B. In Treatise on Analytical Chemisty; 2nd ed.; Elving, P. F., Ed.; Wiley-Interscience: New York, 1982;Vol. 5. (14)Wilson, D. J.; Clarke, A N. Topics in Foam Flotation; Dekker: New York, 1983. (15)Davankov, V. A;Kurganov, A A; Bocklov, A S. Advances in Chromatography; Giddmgs, J. C., Grushka, E., Cazes, J., Brown, P. R, Eds.; Marcel Dekker: New York, 1983;Vol. 22,p 71. (16)Hare, P. E.; Gil-Av, E. J. Chromatogr. 1976, 122,205. (17)Lepage, J.; Lindner, W.; Davies, G.; Karger, B. Anal. Chem. 1979,51,433. (18)Armstrong, D. W.;DeMond, W. J. Chromatogr. Sci. 1984,22,411. (19)Armstrong, D. W.; Faulkner, J. R, Jr.; Han, S. M. J. Chromatogr. 1988, 452,323. (20)Snopek, J.; Jelinek, I.; Smolkova-Keulemansova, E. J. Chromatogr. 1988, 452,571. (21)Otsuka, IC;Terabe, S.J. Chromatogr. 1990,515,221. (22)Pawlowska, M. Chirality 1991,3, 136. (23)Armstrong, D. W.; Rundlett, IC;Reid, G. L., 111Anal. Chem. 1994,66,1990. (24)Armstrong, D. W.; Rundlett, IC; Chen, J. R Chirality 1994,6, 496. (25)Armstrong, D. W.;Li, W.; Chang, C.-D., Pitha, J. Anal. Chem. 1990, 62, 914. 0003-2700/94/0366-4278$04.50/0 0 1994 American Chemical Society
Figure 1. Schematic showing the glass foaming device used in this work. The detachable column fits into the sample chamber. Air or nitrogen gas enters the bottom of the sample chamber through a fine glass frit. The foam produced in the sample chamber passes into the column. The velocity with which the foam travels up the column is controlled by the gas flow rate and also is affected by any packing material in the column. Many of the bubbles that make up the foam break, thereby forming a thin layer of liquid that flows down the inner surfaces of the column toward the sample chamber.
The foaming device used in this study is shown in Figure 1. This was constructed (along with others) by our glass blower according to our instructions. It consists of three basic parts: (1) a sample chamber, which holds an aqueous or buffer solution containing the racemate and the chiral collector, (2) the gas introduction section, which is separated from the sample chamber by a fine porous frit, and (3) a detachable foam column (25 cm or 40 cm x 1.6 cm id.) that may contain 7 mm diameter glass bead packing material. Several of these devices can be operated in series if desired. Methods. Stock solutions of compounds studied were made into stock solutions in ethanol (or in acetonitrile for those analytes analyzed using vancomycin) at a concentration of 12 mg/mL). Solutions of the chiral collector were made in distilled water or buffer at concentrations ranging from 0.01%to 1%. One milliliter of the racemic mixture solution and 19 mL of chiral collector solution were placed irk0 the foaming chamber, and the foaming column was attached to the sample chamber via a ground-glass joint. Air was used to generate bubbles from the bottom of the chamber. The height of the foam in the column was controlled by adjusting the air flow rate with a h e control valve. In general, dry foam formed after a period of controlled reflux gives the best results. The glass beads also seem to help produce a h e r , slower rising foam. The foam bubbles tend to break on the way up the column thereby producing a thin film of liquid that flows down the column (Le., opposite to the foam migration). The aidow rate was adjusted so that a steady countercurrent condition was produced (somewhat analogous to that of a fractional distillation column). At this point no foam passes to the receiver (Figure 1). A foam fraction was collected either by slightly increasing the
flow rate to force some foam into the receiver or by using a syringe connected to a long thin Teflon tube on the needle to reach the top layer of the foam in the column. In some experiments, 3 mL of a foam fraction and 12 mL of fresh collector solution were used for a series of consecutive runs. Most enantiomeric ratios in the foam fractions were determined by HPLC. A Shimadzu LC 6A liquid chromatography with a variablewavelength detector and a C-R3A Chromatopac data system were used for all data analysis. The chromatographic conditions for the separation of all compounds studied are listed in Table 1. The UV detection wavelength was 254 nm for all analytes except for the N-t-BOC amino acids, which was 225 nm. The warfarin peaks overlapped with the collector, thus solid phase extraction on a silica cartridge was employed to remove the collector before HPLC injection. The silica cartridge was washed with 3-5 mL of diethyl ether before the warfarin fraction was loaded, and then warfarin was eluted from the cartridge with diethyl ether and collected for analysis. Foam fractions of other compounds were injected directly on the chiral stationary phase containing LC column. The enantiomeric purity of dansyl-tryptophan and dansyl-phenylalanine were determined directly on a Waters Quanta 4000 CE apparatus equipped with a 254 nm lamp and a 50 pm (i.dJ x 32 cm (24 cm to detector) fused silica capillary. The run buffer contained 0.1 M, pH 6 phosphate buffer plus 2 mM vancomycin as the chiral selector.24 General conditions for the alkylhydroxyproline and digitonin chiral collector experimentswere as follows: 5 mL of 0.02%(unless indicated otherwise) surfactant solution (E-heptyl-L-hydroxyproline, for example) in water was mixed with 0.25 mL of stock solutions of leucine (2 mg/mL), methionine (2.32 mg/mL), phenylalanine (2.6 mg/mL), and tryptophan (2.8 mg/mL). Approximately 4.4 mg of cupric acetate was added to the solution. When foaming was completed, the residual solution was analyzed as outlined below. All amino acids were derivatized with (6aminoquinolyl)N-hydroxysuccinimidyl carbamate (AQC) and separated via HPLC as indicated in Table I. This achiral derivatization procedure for amino acids and the enantioselective separation method has been described previou~ly.~~ RESULTS AND DISCUSSION
The enrichment of enantiomers using foaming agents (collectors) that also are chiral selectors is examined in this work. It is understood that there are other possible approaches that could produce enantiomeric enrichments as well. For example, an adsorptive bubble method might be feasible if the racemate itself was surface active and a nonfoaming chiral selector was employed. In the former case (which is the focus of this paper), the physicochemical properties of the chiral collector are all important. It must be (a) at least partially water soluble, (l~) able to create a foam, (c) able to enantioselectively bind the analyte, and (d) able to maintain some surface activity when associating with at least one of the two enantiomers. Enantioselectivity could result either from a difference in the association energy between the chiral collector and the two enantiomers, from a difference in the surface adsorption of the two diastereomeric (collector plus enantiomer) complexes, or from some combination of these two factors. Examples of chiral collectors that meet the aforementioned physicochemical criteria include some derivatized cyclodextrins, alkylated amino acids, digitonin, and some antibiotics. Enantiomeric enrichment conditions and results using these collectors are given in Tables 2 and 3. Table 2 summarizes the data for Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
4279
Table 1. Analytical Separatlon Conditions for Enantiomeric Excess Determinations
racemate
structure
column
mobile phase' or run buffer acetonitrile/methanol/acetic acid/triethylamine
warfarin
Cyclobond I*
SQ 30,840
Cyclobond I RSF aqueous buffe@/acetonitrile (80/20)
SQ 31,236
Cyclobond I Spd
(90/10/0.004/0.004)
aqueous buffer%/acetonitrile(85/15)
N-t-Boc-chlorophenylalanine
CI
Cyclobond I RSF aqueous bufferk/acetonitrile (95/5)
N-t-Boc-phenylalanine
CO2H o C H 2 - C H - N HI - C O 2 - C ( C H &
Cyclobond I RSP aqueous buffe@/acetonitrile (93/7)
leucine"
COzH &CH~-CH-NH-CO~-C(CH~)P I
Cyclobond I*
acetonitrile/methanol/acetic acid/triethylamine
Cyclobond I*
acetonitrile/methanol/acetic acid/triethylamine
phenylalaninen
Cyclobond Ib
acetonitrile/methanol/acetic acidhiethylamine
tryptophann
Cyclobond I*
acetonitrile/methanol/acetic acid/triethylamine
fused silicae
0.1 M phosphate buffer (PH 6) containing 3 mM
fused silicae
0.1 M phosphate buffer (PH 6) containing 3 mM
methioninen
COOH
I
H2N-CH-CH2-CH(CH& COOH
I
H2N-CH-CH2-CH2-S-CH3
(90/10/0.4/2) (96/4/0.6/1)
(90/10/0.4/1.2) (96/4/0.6/1.2)
'COOH
dansyl-phenylalanine
vancomycinz4
vancomycinz4
Before analysis these amino acids were derivatized with AQC; fluorescent tagging agent. See ref 25 for the complete rocedure. * Cyclobond I refers to a 25 cm (4.6 cm i.d.) native cyclodextrin HPLC column. Cyclobond I RSP refers to a 25 cm (4.6 cm i.& ruc-hydroxypropyl;Bcyclodextrin HPLC column. d Cyclobond SP refers to a 25 cm (4.6 cm i.d.) Q-hydroxypropyl-,kyclodextrin HPLC column. e This column is a 50
P
pm (id.) x 32 cm (24 cm to detector) fused silica capillary used in capilly electrophoresis. 'Mobile phase compositions are given as volume ratios. gThe buffer consisted of 1%triethylammonium acetate (ad, pH 4.1. The buffer consisted of 1%triethylammonium acetate (ad, pH 7.1.
different derivatized cyclodextrin collectors while Table 3 does the same for all other chiral collectors. Note that the alkylproline collectors (Table 3) were used in a "ligand exchange" f ~ r m a t . ' ~ - ' ~ Hence, copperuI) salts were dissolved in solution before foaming (see Experimental Section). A number of things are evident from the data (Tables 2 and 3). First and foremost, it is clearly demonstrated that enantiomeric enrichments are possible using adsorptive bubble processes. While enantiomeric enrichments after a single pass through the column at room temperature (ee's = 4-30) are not extraordinary, they are easily enhanced by modifying certain experimental conditions. As can be seen in Table 2 for N-t-BOC-chlorophenylalanine and N-t-BOC-phenylalanine, lowering the temperature of the foam flotation column and sample reservoir to 4 "C results in 4280 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
a significant increase in the enantiomeric excess. Increasing the length of the foam column from 25 to 40 cm also appeared to enhance the effectiveness of the separation (Table 2 and Figure 2). The results for warfarin were particularly interesting. This was because enrichments were produced with two different cyclodextrin derivatives but they gave opposite enantiomeric selectivity (Figure 3 and Table 2). This was the only case where an enantioselective reversal was observed. Enantiomeric enrichments also can be enhanced by repeating the foam fractionation process with a series of identical devices (see Experimental Section for details). Results for two different compounds (when hydroxypropyl+cyclodextrin was used as the chiral collector) are shown in Tables 4 and 5. It appears that there are some similarities between the operation of a foam fractionation
Table 2. Foam Flotation Conditions and Enantlomerlc Enrichments Using Derlvatized Cyclodextrin Collectors
racemates warfarin SQ 30,840 SQ 31,236 N-t-BOCchlorophenylalanine
N-t-BOC-phenylalanine
collectors0 PM-p-CD DM-B-CD Od-B-CD HP-p-CD HP-b-CD PM-p-CD PM-p-CD PM-p-CD PM-P-CD PM-p-CD DM-p-CD
collector conc (g/lOO mL)
temp (“C)
foaming column ht (cm)
no. of passes through column
enantiomeric ratiob
eec
1 1 1 1 1 0.05 0.05 0.05 0.05 0.05 0.05
23 23 23 23 23 23 4 23 4 4 4
25 25 25 25 25 25 25 25 25 40 25
1
60/40 44/56 56/44 66/34 46/54 47/53 37/63 41/59 18/82 12/88 48/52
20 12 12 32 8 6 26 18 64 76 4
1 1 1
1 1
1 1 1
1 1
(%I
“ Abbreviations: PM-D-CD = he takis-2,3,&tri-Omethyl-j3qclodextrin;DM-P-CD = heptakis-2,W-Ometh 1Bcyclodextrin; Oct-p-CD = a mixture of qclodextrin monooctyl and Joctyl esters; HP-P-CD = hydroxypropyl- qclodextrin. * This is the p e d L e a ratio of first peak/second peak. e dantiomeric excess (ee) is defined as follows: ee = [(El - &)/(E1 + Ez ] x loo%,where E1 is the amount of the enantiomer present in higher concentration and E2 is the amount of the enantiomer present in lower concentration.
f
Table 3. Foam Flotation Conditions and Enantiomeric Enrichments Uslng Chlral Collectors Not Related to Cyclodextrin
racemates leucine methionine phenylalanine tryptophan dansyl-tryptophan dansyl-phenylalanine
collectors” HepHProl Dodec-HProl HepHprol HepHProl digitonin vancomycin vancomycin
collector conc &/lo0 mL) 0.4
satd soh 0.2 0.2
satd soh 0.06 0.06
temp (“C)
foaming column ht (cm)
no. of passes through column
enantiomeric ratiob
eec
23 23 23 23 23 4 4
25 25 25 25 25 25 25
1 1 1 1
35/65 46/54 42/58 41/59 47/53 41/59 47/53
30 8 16 18 6 18 6
1 1
1
(%)
0 Abbreviations: HepHProl = n-heptyl-L-hydroxyproline; Dodec-Hprol = n-dodecyl-L-hydroxyproline. This is the peak area ratio of fist peak/ second peak. Enantiomeric excess (ee) is defined as follows: ee = [(El - Ez)/(& + Ez)] x loo%! where E1 is the amount of the enanhomer uresent in higher concentration and EZis the amount of the enantiomer present in lower concentrahon.
---
0
4
8
Y O
ii
TIME,
TIME, MIN Flgure 2. LC chromatogramsshowing the enantiomeric enrichment of D-N-tBOC-phenylalanine(Le., the second peak in both chromatograms) using (A) a 25 or (B) a 40 cm long foam column. The experiment was carried out at 4 “C using permethyl$-cyclodextrin as the chiral collector, under total reflux conditions (see Experimental Section). The analytical LC chiral separations were done on a Cyclobond I-RSP column (see Table 1 for exact Conditions).
column and a fractional distillation column. For example, longer columns (Table 2) and additional fractionation cycles (Tables 4 and 5) give a more highly enriched product. The best results are obtained in both systems when the columns are operated under “reflux” conditions (where there is a countercurrent flow in the column). As shown in Figure 4, allowing the foam system
8
Y O
ii
8
12
M I N
Figure 3. LC chromatograms: (A) separation of racemic warfarin starting material, (B) foam-enriched product using permethyl-pcyclodextrin as a chiral collector, and (C) foam-enriched product using heptakis(2,6-di-Omethyl)-~-cyclodextrinas a chiral collector. Experiments B and C were done at room temperature using a 25 cm long packed foam column (see Experimental Section). The analytical separations were done on a Cyclobond I HPLC column (see Table 1 for exact conditions).
to equilibrate or reflux for a period of time (before sample removal) enhances the enrichment process. However, after this particular system has equilibrated for a period of time (-25-90 min), there is no benefit (i.e., no further enrichments) in extending the process. Figure 4 also shows that the concentration of the chiral collector affects the enantiomeric enrichment. There appears to be an optimum concentration of collector and racemate for every system. The enantiomeric excess decreases if the system is operated with too much or too little collector. The optimum concentrations of collector and racemate must be determined empirically. As yet we have found no recognizible trend or theory Analytical Chemisfry, Vol. 66, No. 23, December 1, 1994
4281
Table 4. Effect of Repeated Passes through a Foam Flotation Column on the Enantiomeric Enrichment of SQ 30,840.
cycle number peakarea
1
2
3
4
5
6
7
8
first peak second peak peakratio
2.31 2.08
2.44 1.55
1.39 0.76
1.22 0.56
1.18 0.55
1.23 0.44
0.82 0.24
0.68 0.16
1.11 1.61
1.83
2.18
2.15
2.80
3.42
4.25
Hydro ropyl-/3qclodextrinwas used as the chiral collector (See Experimen3Section and Table I1 for further experimental details. Table 5. Effect of Repeated Passes through a Foam Flotation Column on the Enantiomeric Enrichment of SQ 31,230.
I
cycle number peak area
1
2
3
4
5
6
first peak second peak peak ratio
3.61 4.26
2.66 3.42
1.97 2.66
1.48 2.16
1.22 2.55
0.64 1.43
1.18
1.28
1.35
1.46
2.09
2.23
Hydroxypropylj3qclodextrin was used as the chiral collector (See Experimental Section and Table I1 for further experimental details. (I
h
@
16
2
6
4
8
10
PH Figure 5. pH dependance for the enantiomeric enrichment of N-fBOC-phenylalanine. See Experimental Section for all conditions. The buffer was 0.01% triethylammonium acetate.
controls the ionization state of the analyte and in some cases of the chiral selector. Consequently, enantioselective interactions are strongly pH dependent. It is not unusual in HPLC and capillary electrophoresis for an enantiomeric separation to occur at one pH and not at another.23,24,27*28 Clearly, optimization of the pH can be as important in foam flotation as in most other waterbased enantioseparation techniques.
-
W
r/l
M 8 W
0
12 -
CONCLUSIONS
u
. I
7
6
12
18
Foaming Time (hrs) Figure 4. Effect of reflux equilibration time and concentration of the chiral collector on the enantiomeric enrichment of 0-N-t-BOCphenylalanine. Each curve represents a different concentration of the permethyl-p-cyclodextrin chiral collector: (a) 1%, (b) 0.1YO,(c) 0.01%, and (d) 0.05%. Experiments were done at room temperature using a 25 cm long packed foam column. The chiral collector was permethyl-P-cyclodextrin. See the ExperimentalSection and Tables 1 and 2 for further details.
that allows a prediction of the best absolute and/or relative concentrations. Another factor that affects the enantiomeric excess is the pH of the sample solution. Figure 5 indicates that L-N-t-BOC-phenylalanineis more highly enriched at pH's near 3 than at more basic pHs. This is not surprising in view of past enantioseparation results in reversed phase HPLC.26,27The pH (26) Pawlowska, M.; Chen, S.; Armstrong, D. W. J. Chromatogr. 1993,641,2257.
(27) Armstrong, D. W.; Li, W. Chromatogr. Forum 1987,2,43. (28) Han, S. M.; Annstrong, D. W. J. Chromatogr. 1987, 389, 256.
4282 Analytical Chemistry, Vol. 66,
No. 23, December 1, 1994
A variety of different chiral collectors were used in this work. This indicates that the foam flotation of enantiomers is not narrowly applicable. When associating with an analyte, the chiral collector can make use of a variety of interactions including the following: inclusion complex formation and hydrogen bonding (as in the case of the derivatized cyclodextrins in Table 2); ligand exchange interactions Cable 3) ; steric interactions (which are important for all of the chiral collectors tested); and combinations thereof. The relatively low cost, the ease with which the separations are performed, and the ability to enrich large samples on a continuous basis are all important factors for large-scale enantiomeric separations. It should be noted that a complete economic analysis of this technique (which was not the focus of this work) must take in to account the ease with which the chiral collector can be separated from the product and recycled. Also, it is possible to design chiral collectors for specific separations. This would not only enhance the enantioselectivity but would also allow optimization of the surface-active properties. It is hoped that the data presented here will stimulate an increase in interest in this area of separations. ACKNOWLEDGMENT
Support of this work by the Department of Energy, Offices of Basic Science (DE FG02 88ER13819), and the Biotechnology Research and Development Corp. are gratefully acknowledged. Received for review July 25, 1994. Accepted September 17, 1994. @Abstractpublished in Advance ACS Abstracts, October 15, 1994
