606
Anal. Chem. 1985, 57,606-610
Characterization of a Chiral Tripeptide Stationary Phase for the liquid Chromatographic Separation of Chiral Dipeptides William A. Howard,' Tau-Being HSU,and L. B. Rogers* Department of Chemistry, University of Georgia, Athens, Georgia 30602 David A. Nelson Department of Chemistry, University of Wyoming, Laramie, Wyoming 82071
A trlpeptlde-bonded stationary phase was synthesized by bondlng the trlpeptlde, L-Val-L-Ala-L-Pro, to a reactlve sllane bonded to slllca gel. The retentlon behavior of several lsomerlc dlpeptldes was examlned as a function of pH and buffer composltlon. Although retentlon behavlors were slmllar to those reported for the nomlnally same trlpeptlde phase prepared In another way, the data lndlcated that the phases were slgnlflcantly dlfferent. I n addltlon, for the new phase at pH 7.4, a buffer effect was found In which retentlons decreased on golng from a phosphate-cltrate-chlorlde (McIlvalne) to phosphate buffer alone, to cltrate alone, and to pyrophosphate. However, wlth respect to selectlvlty, the McIIvalne and pyrophosphate buffers were superlor to cltrate and, In turn, phosphate.
The present study follows up the very interesting work by Grushka and co-workers (1-3) in which a chiral tripeptide was synthesized on the surface of silica and then used to fractionate chiral dipeptides. There is no doubt that chirality was contributing to the observed retentions. For the most part, diastereomers were not examined, but crossovers in retentions of diastereoisomers with pH were noted in their work. Fong and Grushka prepared their tripeptides on the surface by a stepwise manner analogous to a Merrifield synthesis in which a solid phase is used as the support. Even if individual steps produced nearly a 95% yield, a substantial fraction of the surface should be derivatized with mono- and dipeptide groups. Hence, although their work demonstrated the feasibility of using a bonded tripeptide phase, the mixed nature of the derivatized surface introduced some ambiguity into the interpretations of results obtained using their packing. The present study was designed to circumvent the problems of multiple derivatives by synthesizing the intact tripeptide before attaching it to the surface. Admittedly, the surface would not be homogeneous; silanols, especially, would be present in abundance. However, although silanols would contribute to the retention of a dipeptide solute, they would make no chiral contribution. Assuming that significant racemization can be avoided, conclusions about the contribution of chirality, including any templates and conformational effects, should be sound. The first goal, therefore, was to prepare a phase having the same tripeptide but prepared by attaching the intact tripeptide. Then, the phase was to be tested by using some of the same dipeptides (pairs of sequence isomers) used by Fong and Grushka so as to compare on a relative basis their retentions in terms of the capacity ratio, k , and especially their changes with pH. It was also of interest to observe the effects on the ratios of the k values ( a values), both before and after Present address: Burroughs Wellcome Co., P.O. Box 1887, Greenville, NC 27834.
the derivatized packing had partially decomposed (with use). While it is possible that decomposition might have increased, rather than decreased, the k and a values as dilution has been shown to do ( 4 ) ,it seemed unlikely to be important for the relatively small extent of decomposition that was anticipated. Finally, it was of interest to explore in a preliminary way the effects of different anions in several buffers, each a t the same pH. Because an effect has been reported for phosphate (51, it was used alone in one buffer as was citrate. In addition, a McIlvaine buffer, which contains both citrate and phosphate (plus chloride to attain the desired ionic strength), was also examined. Finally, a pyrophosphate buffer was included so as to see if it behaved much the same as phosphate.
EXPERIMENTAL SECTION Chemicals and Reagents. LiChrospher Si 100, silica having a 10-pm particle diameter, was used for preparation of the bonded phase (Alltech Associates, Norcross, GA). The 1-(dimethylchlorosilyl)-2-(4-chloro-3-methylphenyl)ethane silane reagent used to bond the peptide to silica was obtained from Petrarch Systems (Bristol, PA). All chemicals were J. T. Baker (Phillipsburg, NJ) reagent grade, HPLC grade, or better unless otherwise stated. The following reagents were obtained for synthesizing the tripeptide L-valineL-alanine-L-proline (L-Val-L-Ala-L-Pro) using isobutylchloroformate. The amino acid derivatives used in this synthesis were tert-butoxycarbonyl-L-valine(BOC-L-V~) from US.Biochemicals Corp. (Cleveland, OH) and the methyl ester hydrochlorides of alanine and proline from Chemalog (South Plainfield, NJ). The reagents used for coupling the amino acids were isobutylchloroformate and N-methylmorpholine, both from Aldrich Chemical Co. (Milwaukee, WI) as were the triethylamine and tetrahydrofuran (THF). The THF was distilled over potassium prior to use. The workup of the coupling reactions employed house doubly distilled deionized water plus ethyl acetate, anhydrous magnesium sulfate, sodium chloride, and hydrochloric acid, along with sodium hydroxide from Fisher Scientific Co. (Norcross,GA). The methyl ester of the peptide was hydrolyzed and worked up with sodium hydroxide, methanol, diethyl ether, citric acid monohydrate, sodium chloride, and chloroform. Solvents used for bonding the silane reagent to silica and then the peptide to the silane-derivatized silica were toluene and acetonitrile. The toluene was either distilled or dried over calcium chloride before being stored over sodium. Proton Sponge, a catalyst from Aldrich (Milwaukee, WI), was used to accelerate the bonding reactions. Formic acid (88%) was used to remove the BOC protecting group. Chloroform, THF, and DME were used in the syntheses of the tripeptide. The DME was dried over sodium or molecular sieve 5A prior ta use. Isooctane, cyclohexanol, carbon tetrachloride, and 2-propanol were used to pack the column. Methylene chloride and methanol were used to wash the column after packing. Dipeptide standards were used to evaluate the peptide bonded phase columns. These included L-valine-L-phenylalanine(LVal-L-Phe),L-phenylalanine-L-valine (L-Phe-L-Val),L-alanine-Lphenylalanine (L- Ala-L-Phe), and L-phenylalanine-L-alanine(LPhe-L-Ala),all from Sigma Chemical. The peptides were prepared as 1 mg/mL solutions in distilled water and stored in the refrigerator overnight.
0 1985 American Chemical Society 0003-2700/85/0357-0608$01.50/0
ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985
Doubly distilled deionized water was used for preparing the buffers used as mobile phases. The buffer reagents included anhydrous monosodium phosphate and tetrasodium pyrophosphate decahydrate (both from Sigma), anhydrous dibasic sodium phosphate, dibasic sodium phosphate dodecahydrate, tribasic sodium citrate dihydrate, and citric acid monohydrate. The citrate, phosphate, and pyrophosphate buffers were prepared as 0.10 M solutions and adjusted to the appropriate pH using 0.1 M solutions of sodium hydroxide and hydrochloric acid. The McIlvaine buffer was prepared according to Elving et al. (6). The mobile phases were filtered through 0.50 pm cellulose acetate and nitrate (HAWP) Millipore filters (Bedford, MA) before being degassed with helium prior to use. Apparatus. The chromatographic system consisted of an isocratic Spectraphysics Model SP8770 pump (San Jose, CA), a Rheodyne Model 7125 injection valve (Cotati, CA), a Kratos Model 770 liquid chromatographic detector (Ramsey, NJ) set a 254 nm and 0.04 AUFS, unless specified otherwise, and a Linear Model 585 (Reno, NV) chart recorder. The columns were packed with a SC Hydraulics Model 10600-50 pump (Loa Angeles, CA) and a homemade slurry bomb. Procedures, Derivatization Reactions. The peptide L-ValL-Ala-L-Prowas synthesized by using the procedures of Vaughan and Osato (7) and Galobardes (8). A typical peptide synthesis proceeded as follows. N-Methylmorpholine (11mmol) dissolved in THF (10 mL) was added to a 0 "C solution of BOC-L-Val(10 mmol) in THF (20 mL). Isobutylchloroformate (11mmol) in THF (10 mL) was added and the reaction stirred for 30 min. A well-stirred suspension of L-Ala-OMeHCl (11 mmol) and N methylmorpholine (11mmol) in THF was added to the reaction mixture and stirred for 12 h. The reaction mixture was filtered and the THF removed in vacuo. The crude product was dissolved in ethyl acetatewater (80 mL to 80 mL) and waehed successively with 80 mL each of water, 1%aqueous citric acid, water, 10% aqueous sodium bicarbonate, and water. The solvent was removed in vacuo to yield the product BOC-L-Val-L-Ala-OMe. The protected dipeptide, BOC-L-Val-L-Ala-OMe,was hydrolyzed to yield a free carboxylic acid for coupling. A solution of the dipeptide methyl ester (10 mmol) in methanol (50 mL) containing sodium hydroxide (28 mmol) was stirred for 3 h at room temperature. After dilution with water (40 mL) and extraction twice with 40 mL of ethyl acetate, the organic layer was discarded. The aqueous layer was cooled, acidified to pH 2 with citric acid, and saturated with ammonium chloride before being extracted 3 times with 50 mL of chloroform. The chloroform extracts were combined, washed with water, and concentrated in vacuo to yield the free dipeptide acid, BOC-L-Val-L-Ala-COOH. The dipeptide acid, BOC-L-Val-L-Ala-COOH,was coupled with L-Pro-OMeHC1using the coupling procedure described previously. The product, BOC-L-Val-L-Ala-L-Pro-OMeHC1,was hydrolyzed ta the free acid, BOC-L-Val-Ala-L-Pro-COOH. The average overall yield of the tripeptide (five runs) was 78%. The mixed-anhydride methods used for this synthesis were reported to give no racemization (7). One indication of the stereochemical homogeneity of the product was a single peak on HPLC analysis using two different C-18 columns and a variety of mobile phases. The silane-derivatized silica for bonding the tripeptide to the silica surface was prepared by suspending silica gel in boiling toluene (100 mL) and adding the silane reagent 1-(dimethyl(17 mmol). The chlorosily1)-2-(4-chloro-3-methylpheny1)ethane reaction was allowed to proceed for 24 h before the product was collected and washed successively with 100 mL of each of the following solvents: toluene, methanol, methanokwater (l:l),water, methanol, and ether. The derivatized silica was dried at 110 "C in vacuo. Elemental analysis of this silica yielded 10.2% carbon, 1.55% hydrogen, and 2.35% chlorine (Atlantic Microlab, Inc., Atlanta, GA). This is equivalent to a ligand density of 3.1 pmol/m2, assuming a surface area of 300 m2/g for LiChrosphere Si-100. Calculations of surface coverage were made by using the formulas of Berendsen and de Galan (9, IO). The tripeptide was bonded to the derivatized silica by suspending the derivatized silica (2.0 g) in acetonitrile (100 mL) and adding the free tripeptide BOC-L-Val-L-Ala-L-Pro(13 mmol) and Proton Sponge (3.5 g) as a catalyst. After 2 days, the tripeptide bonded phase was collected and washed successively with 100 mL each of acetonitrile, methanol, and water, before being dried at
807
110 "C in vacuo for 12 h. The BOC group was removed with formic acid. Elemental analysis of the peptide bonded phase yielded 11.5% carbon, 1.80% hydrogen, and 0.82% nitrogen (Atlantic Microlab, Inc., Atlanta, GA). This corresponds to a 30% conversion of the original chlorine to tripeptide, based on nitrogen analysis. This percent conversion is similar to that achieved by Grushka (1, 3) for other tripeptide phases. All of the reactions were monitored by reversed-phase liquid chromatography where appropriate. A Spectraphysics octadecyl column (C18),eluted by 30:70 acetonitri1e:water adjusted to pH 2 with phosphoric acid, was employed to fractionate the reaction mixtures. An increase in the concentration of a product or decrease in reactant was used to follow the reaction's progress. Also, proton and carbon-13 nuclear magnetic resonance (NMR) spectrometry were used to confirm the structure of the peptide intermediates. Chromatographic Procedures. The tripeptide bonded phase was packed into 316 stainless steel (SS) tubing 25.0 cm X 0.32 cm i.d. x 0.635 cm 0.d. (Alltech Associates). The columns were terminated with 0.635 cm (1/4 in.) to 0.159 cm (1/16 in.) zero dead volume Swagelok reducing unions (Georgia Valve and Fitting, Atlanta, GA) having 2-wm stainless frits (Alltech). All tubing and reducing unions were washed successively with THF, methanol, water, 3 N nitric acid, water, methanol, and acetone before being dried with nitrogen. The columns were packed using the viscosity technique (11). A 10% slurry (w:v) of the bonded phase in cyclohexanol:2-propanol (3:1, v:v) was placed in the slurry packer and displaced into the column with isooctane at a pressure of 650 bar. Carbon tetrachloride, placed in the column prior to packing, prevented the slurry from settling into the column. Methylene chloride and methanol (100 mL each) were used to wash the column prior to evaluation. The mobile phase flow rate was 0.33 mL/min for all studies. The column was equilbrated with the appropriate mobile phase for 3 h prior to a chromatographicrun. A precolumn, 5 cm X 0.46 cm i.d. X 0.635 cm 0.d. packed with Corasil (Water Associates, Milford, MA) was placed between the pump and the injection valve in order to saturate the mobile phase with silicic acid. This helped to minimize degradation of the tripeptide bonded phase column at high pH. In addition, columns were usually flushed with methanol before being stored overnight, but for shorter times, the column was flushed only with distilled-deionized water. The time (volume) for a nonretained substance, used in calculating the capacity factor, k , was determined by injecting water into the eluent. Because peaks for the solutes were quite symmetrical, the retention times (volumes) were taken at the peak maximum. Values for k from replicate runs usually agreed within 1%.
RESULTS AND DISCUSSION Retentions of Amino Acids. The amino acid components of the dipeptide solutes were examined using a 0.1 M phosphate buffer a t p H 7.4 so as to estimate the relative contribution of each L-amino acid to the overall retentions of the dipeptides. The k values for alanine, valine, and phenylalanine were 0.086,0.117, and 0.290. These show that the amino acids eluted in the order of increasing hydrophobicity, an order consistent with that reported by Fong and Grushka (3),for the phenylthiohydantoin (PTH) derivatives. The quantitative relationships developed by Rekker (12) and cited by Horvath (13)for the hydrophobic contributions of the amino acid side chains also agree very well for the elution order and are useful for making predictions. Therefore, a reversed-phase mechanism contributes in a major way to the overall retention of the dipeptides on a tripeptide bonded stationary phase. Effect of pH. A series of dipeptide solutes were examined in order to characterize their retention behaviors on the tripeptide bonded stationary phase. Again, the peptides eluted in the order of increasing hydrophobicity. Figure 1shows that the k values increased as the pH of the eluent increased with the greatest retention near pH 7-and were still rising. The k value for L-Phe-L-Val increased faster with pH than those of the other dipeptides.
608
ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985
c
1.3
,
1
.? ,
2.0,I
,' I '
,'
0.9-
1
I
2
4
I'
6
8
PH Flgure 1. Effect of pH on k values of dipeptides in McIlvaine buffers of ionic strength, I = 0.5 M: A, L-Ala-L-Phe; 0, L-Phe-L-Ala, 0, L-Val-L-Phe, 0, L-Phe-L-Val.
The elution order and the increase of k values 89 a function of pH agree very well with the data reported by Fong and Grushka (3). Both the magnitudes and trends of the k values are consistent for the two studies as seen in Figure 2. The main difference occurred at about pH 7.4 where the dipeptide solutes and the tripeptide bonded stationary phase were most sensitive to small changes in proton concentration. The $electivity, a,changed steadily as a function of pH as seen in Figure 3. The selectivity values have been defined as follows: a1 =
kL-Val-L-Phe ItL-Phe-L-Va1
a2 =
kL-Ala-L-Phe
kL-Phe-L-Ala
a3 =
kL-Phe-L-Ala kL-Phe-L-Val
aq =
IZL-Ala-L-Phe
0.0' 2
'
,
1
I
L
6
8
PH Flgure 2. Effect of pH on k values of dipeptides in McIlvaine buffer, I = 0.5 M, comparison of Fong and Grushka vs. the present study: (ref 3) 0, L-Val+-Phe; 0, L-Phe-L-Val; (present study) 0,L-Val-L-Phe; W, L-Phe-L-Val.
(1)
(2) \
\
(3)
\
'\
(4)
kL-Val-L-Phe
A selectivity ratio near unity represents the most difficult separation whereas any increase or decrease in the absolute difference from unity indicates an easier separation. Note that a1and a2involve dipeptide pairs of sequence isomers while a3 and a4 measure the separations between the two sets of dipeptides in which valine and alanine (combined with phenylalanine) can be compared. The separation as defined by a1 and a2would be the most difficult between pH 6 and 7 where a is close to unity. For pH >4 the a1was greater than a2while a3was greater than aq.Below pH 4,the selectivities were reversed. The a values for the isomeric pair L-Val-L-Phe and LPhe-L-Val are compared in Figure 4 with those from Fong and Grushka's study under identical conditions. The selectivities found in the present study change in ways very similar to theirs. Effect of Buffer Composition. The effect of buffer compositions at pH 7.4 was examined to evaluate the contribution of the buffer to the retention mechanism. For the dipeptide solutes the lines between the data points in Figure 5 have no significance and were added only to ease visual interpretation. All buffers were at 0.1 M except for the McIlvaine at Z = 0.5 M. The retention was greatest for the McIlvaine buffer and may be an ionic strength effect because
PH Flgure 3. Effect of pH on the a values in a McIlvaine buffer, I = 0.5 M: 0 , kL.val.L.phelkL-phe-L-val; 0,kL-Ala-L-PhelkL-Phe-L-Ala; L-Phe-L-Ala1 A g
k L-phe-L-Val; A,
L-Ala-L.PheJk
L-Val-L-Phe.
the retentions were greater in it (a mixture of phosphate and citrate) than in either of its components alone. The pyrophosphate buffer gave the smallest retentions. Hence, one can change significantly the absolute retentions of dipeptides on a tripeptide stationary phase by changing the buffer composition. These differences in retention were probably due to differing abilities of the buffer species to interact with the
ANALYTICAL CHEMISTRY, VOL. 57, NO. 3, MARCH 1985
609
1.3-
o( 1
I
1.1I
I
A
B
D
C
Buffer
2
c
6
a
PH Flgure 4. Effect of pH on the a values for the dipeptide pair L-phe-L-Val and L-Vak-Phe in a McIlvaine buffer, I = 0.5 M: 0, ref 3; 0 , present
Figure 6. Effect of buffer composition at pH 7.4 on the lla values for sequence isomeric pairs with phenylalanine: valine (0)and alanine (0)pairs; Mobile Phases were (A) McIlvaine, (B) phosphate, (C) citrate, ~L-V~K-~~LW-V~I; and (D) pyrophosphate. 0 , ~ L - v ~ K + ~ ~ L - - v E I ; 0,~ L - A ~ . L - ~ ~ ~ ~ L - ~ ~ ~ - A ~ . Table I. Retentions of Amino Acids in 0.1 M Phosphate Buffer, pH 7.4
study.
k
1.3-
amino acid
1st expt
2nd expt
Ala Val Phe
0.086 0.117 0.290
0.036 0.093 0.296
Table 11. Retentions of Dipeptides in 0.1 Buffer, pH 7.4
0.9-
hl Phosphate
1st expt
k' 0.5-
1
dipeptide
k
L-AIa-L-Phe
0.52
L-Phe-L-Ala L-Val-L-Phe
0.61 0.94
L-Phe-L-Val
1.00
2nd expt a
0.86
k 0.35
0.80
0.44 0.64 0.94
0.1
A
B
C
a
0.90
0.70
D
Buffer Figure 5. Effect of buffer composition at pH 7.4 on k values: A, L-Ala-L-Phe; 0 , L-Phe-L-Ala; 0,L-Val-L-Phe; 0, L-Phe-L-Val. Mobile phases were (A) McIlvaine, (B) phosphate, (C) citrate, and (D) pyrophosphate. dipeptide solute and/or with the tripeptide stationary phaee. The effect of the buffer composition on 1/a values can be seen in Figure 6. The McIlvaine and pyrophosphate buffers gave the best selectivities while phosphate gave the worst. It is interesting that the pyrophosphate buffer exhibited the least retention but gave good selectivity. This served to emphasize the independence of retention and selectivity. Long-Term Stability. The long-term stability of the tripeptide bonded phase columns as a function of time in contact with aqueous solutions was a prime concern since a pH >7.5 can solubilize silica gel. Therefore, some experiments were repeated under identical conditions after about 2 months of continuous use to see if there were any changes in the k or a values. Results of the amino acids retentions are shown in Table I. The retentions decreased by more than 50% for alanine and about 20% for valine, while phenylalanine remained unchanged. This suggests that the silanol interaction was smallest for alanine. The k values for the dipeptides changed in a manner consistent with the behavior of the amino acids they contained. The alanine-containing dipeptides changed
more than the valine-containing ones. The a values also changed, but correspondingly less due to the presence of phenylalanine in each dipeptide.
CONCLUSIONS This study demonstrated the feasibility of bonding an intact tripeptide, L-Val-L-Ala-L-Pro,to silica gel for separating dipeptide isomers. When the retentions of the dipeptides (Table 11) and the selectivities of the separations using this phase were examined as a function of pH, the trends were very similar to those reported by Fong and Grushka (3), and the dipeptides that contain valine have the greatest k values at higher pH. However, retentions of dipeptides reported by Fong and Grushka (3) at high pH were significantly greater than those in the present study in spite of having approximately the same carbon percentages (12.3% (3)and 11.5%) in the two studies. Greater retention suggests that there was more valine on their surface because its hydrophobicity is greater than that of alanine or proline. Therefore, their smface is quite probably not the pure tripeptide, but a mixture of valine, valine-alanine, and valine-alanine-proline. The a values for dipeptides as a function of pH reported in Figure 3 show that, at low pH, protonation of the N-terminal aliphatic amino acid appears to increase the selectivity for the isomeric pair more than protonation of the N-terminal phenylalanine. Going to higher pH values decreased the a values. In fact, below an a value of 1.00, ionization resulted in greater se-
610
Anal. Chem. 1985, 57. 610-615
lectivity for L-Phe-L-Val and L-Val-L-Phe; at least its effect was greater for aliphatic than for aromatic amino acids. A greater effect was also found for the ionization of valine relative to alanine. The retentions of the dipeptides and the selectivities of the separations were also examined as a function of buffer composition. Pietrzyk and co-workers (14-16) discussed as a function of pH the different mechanisms for retention by more conventional (nonchiral) packings. In the present study, there appears to be an additional factor influencing the separations as evidenced by the following. The McIlvaine buffer, which contains both phosphate and citrate, gave greater retention and selectivity than a buffer of either of its componenb. This difference may be the result of a difference in ionic strength. The greater ionic strength may have resulted in greater dehydration of ions which may, in turn, have increased the selectivity of the separation, as cited by Niederwieser (17). However, Krummen and Frei (18) have also shown the opposite trend to be possible, depending upon the range of the ionic strength involved. The pyrophosphate buffer gave the smallest retention. However, it gave almost as good a selectivity as the McIlvaine buffer, even though its retention values were smaller. Hence, one wonders if ion pairing or a conformational change was the source of the differences. As a result, the use of pyrophosphate as an eluent is being explored. In the future, it will be of interest to immobilize different tripeptides so as to examine in more detail the effects of chirality on the k and a values for selected isomeric dipeptides. It will also be interesting to examine the effects of having Zn (19,20) or cupric ion (21) in acidic buffers, such as those of ammonium acetate, so as to take advantage of ligand exchange behavior.
ACKNOWLEDGMENT We thank Mercedes Galobardes for her preliminary work. We also thank Spectrophysics for the loan of a SP8770 pump.
Registry No. Ala, 56-41-7; Val, 72-18-4; Phe, 63-91-2;L-AlaL-Phe, 3061-90-3;L-Phe-L-Ala,3918-87-4;L-Val-L-Phe,3918-92-1; L-Phe-L-Val, 3918-90-9. LITERATURE CITED Kikta, E. J., Jr.; Grushka, E. J. Chromatogr. 1977, 735, 367. Fong, G. W.-K.; Grushka, E. J . Chromafogr. 1977, 142, 299. Fong, G. W.-K.; Grushka, E. Anal. Chem. 1978, 50, 1154. Felbush, B.; Cohen, M. J.; Karger, B. L. J. Chromatogr. 1983, 282, 3. Hancock, W. S.;Bishop, C. A,; Prestidge, R. L.; Harding, D. R. K.; Hearn, M. R. W. J. Chromafogr. 1978, 153, 399. (6) Elving, P. J.; Markowitz, J. M.; Rosenthal, I.Anal. Chem. 1956, 2 8 , 1179. (7) Vaughan, J. R., Jr.; Osato, R. L. J. Am. Chem. SOC. 1952, 74, 676. (6) Galobardes, M., The University of Georgia, unpubllshed results, 1982. (9) Berendsen, G. E.; deGalan, L. J. Llq. Chromafogr. 1978, 1 , 561. (10) Berendsen, G. E.; Pikaart, K. A.; deGalan, L. J. Liq. Chromatogr. 1980, 3 , 1437. (11) Slemon, C. C. J . Liq. Chromafogr. 1983, 6 , 765 (12) Rekker, R. F. “The Hydrophobic Fragmental Constant”; Elsevier: New York, 1977; p 301. (13) Molnar, I.; Horvath, C. J. Chromatogr. 1977, 142, 623. (14) Kroeff, E. P; Pietrzyk, D. J. Anal. Chem. 1978, 50, 1353. (15) Kroeff, E. P.; Pietrzyk, D. J. Anal. Chem. 1978, 5 0 , 502. (16) Pietrzyk, D. J.; Smith, R. L.; Cahill, W. R., Jr. J. Liq. Chromafogr. 1983, 6, 1645. (17) Niederwieser, A. I n “Chromatography”; Heftmann, E., Ed.; Van Nostrand-Reinhoid: New York, 1975; p 417. (18) Krummen, K.; Frei, R. W. J. Chromatogr. 1977, 132, 27. (19) Cooke, N. H. C.; Viavattene, R. L9; Eksteen, R.: Wong, W. S.; Davies, G.; Karger, B. L. J. Chromafogr. 1978, 749, 391. (20) Karger, B. L.; Wong, W. S.; Viavattene, R. L.; Lepage, J. N.; Davies, G. J. Chromafogr. 1978, 167, 253. (21) Roumeliotis, P.; Unger, K. K.; Kurganov, A. A,; Davankow, V. A. J. Chromafogr. 1983, 255, 51. (1) (2) (3) (4) (5)
RECEIVED for review February 6, 1984. Resubmitted September 27, 1984. Accepted November 16, 1984. We acknowledge partial support of this work by the National Science Foundation, Grant No. CHE78-13269. W.A.H. also thanks the Eastman Kodak Co. for a fellowship and the University of Georgia, Department of Chemistry, for an H. L. Richmond Scholarship.
Reversed-Phase Liquid Chromatography with Fourier Transform Infrared Spectrometric Detection Using a Flow Cell Interface Charles C. Johnson,l John W. Hellgeth, and Larry T. Taylor* Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0699
The results from preliminary lnvestlgatlons Into the use of segmented, flowing extraction of aqueous effluents from a reversed-phase HPLC system for flow cell FTIR detection are presented. The use of a hydrophoblc membrane phase separator facllltated the operatlon of the RP-HPLC/FTIR instrument. By lncorporatlngthis technlque wlth chloroform as the extraction solvent, one observes slgnlflcantly improved Infrared transparency over the aqueous, elution solvent system. Further Improvements were observed with carbon tetrachloride, not only in the Infrared transparency of the solvent system but also In the wider range of compounds that were extracted from the methanol/water elutlon solvent system. Infrared spectra obtalned match standard reference spectra peak-for-peak.
‘Current address: The Procter and Gamble Co., 6210 Center Hill
Rd., Cincinnati, OH 45224.
Recently we have reported several developments involving the Fourier transform infrared (FTIR) spectrometer as a detector for normal-phase high-performance liquid chromatography (NP-HPLC) employing a flow cell interface (1-5). These developments described the use of semipreparative, analytical, and microbore scales of NP-HPLC applied to the identification of nonpolar and polar components in various coal-derived products. Earlier work had shown moderate sensitivity for this technique, but a greater problem at the time was the infrared absorbance by the mobile phase (6). The use of chlorinated and deuterated solvents in our work has afforded ample mobile phase transparency for flow cell HPLC-FTIR. Furthermore, various methodological and flow cell improvements have now yielded detection limits below 40 ng of injected material (5). Aqueous-based, reversed-phase systems are reportedly used in well over 60% of all HPLC separations. Unfortunately the intense infrared bands of water and methanol, and to a lesser
0003-2700/85/0357-0610$01.50/00 1985 Amerlcan Chemical Soclety