340
Anal. Chem. 1985, 57,340-346
by thermionic emission, the standard error of the difference is less than 0.097 across the range of 96 samples. That is, with 95% confidence, the absolute difference will be less than 0.19%. In summary, it has been established that the precision and accuracy of the CI method for isotopic boron determination is equivalent to the thermionic emission technique while extensive sample preparation and purification, isotopic fractioning, and source memory effects are eliminated.
(4) Cortzee, P. P.; Pretorius, R.; Peisach, M. J . Radioanal. Chem. 1975, 25. 283-292. (5) Braman, R. S. Talanta 1963, 70, 991. (6) Bussel, H. United States Atomic Energy Commission Report NBL-216, New Brunswick Lab, New Burnswick, NJ, 1969, p 27. (7) Fujiwana, S.; Yano, Y.; Nagashima, K. Chem. Insfrum. 1989, 2 , 103-109. 18) Bentiey, P. G. J . Sci. Instrum. 1960, 37, 323-328. 1 Horton, J. C. Anal. Chem. 1966, 38, 198-199. 1 Meiton, C. E.; Giipatrick, L. 0.;Baidock, R.; Heaiy, R. M. Anal. Chem. 1956. 28. 1049-1051. Cameron,'A. E., Stevens, C. M., Roddin, C. J., Ed. "Mass Spectrometry in Analysis of Essential Nuclear Reactor Materials"; U.S. Government Printing Office: Washington, DC, 1964; p 987. Cantanzaro, E. J.; Champion, C. E.; Garner, E. L.; Marinnenko, C.; Sappenfieid, K. M.; Shields, W. R. "Boric Acid; Isotopic and Assay Standard Reference Materials", National Bureau of Standards: Washington, DC 1970; Special Publication 260-17. Lerner, M. W. "The Analysis of Elemental Boron", United States Atomic Energy Commission, Division of Technical Information TID-25, Washington, DC, 1980, p 190.
ACKNOWLEDGMENT We express appreciation to John Grice for his assistance in the statistical analysis and Maxine Pennington for her technical assistance in preparing this paper. Registry No. 1°B, 14798-12-0;B, 7440-42-8. LITERATURE CITED (1) Oliver, C.; Pierce, T. B. Radiochem. Radioanal. Lett. 1974, 77, 335. (2) Oliver, C.; Pelsach, M. J . Radioanal. Chem. 1972, 72, 313. (3) Oliver, C.; McMiiian, J. W.; Pierce, T. B. J . Radioanal. Chem. 1976, 37,512-523.
RECEIVEDfor review June 28, 1984. Accepted October 16, 1984. This work was supported by the U.S. Department of Energy under Contract DE-AC04-76-DP00613.
Liquid Chromatographic Characterization of Picolyl Kel-F as a Reverse-Phase, Weak Anion-Exchange Column Packing Maribeth Kruempelman and Neil D. Danielson*
Department of Chemistry, Miami University, Oxford, Ohio 45056
Picolyl Kel-F (PIKF) Is prepared by reacting Kel-F 6061 ( 100 % poly(chiorotrlfluoroethyiene)) and 4-plcolyillthium In THF under helium. Elemental analysis and I R data show that the primary reaction occurring Is substitution of the chlorine In the polymer for plcoiine. The anlon exchange capaclty was determined to be 22 pequlv/g. The effects of ionlc strength and pH on solute retention are studied to Indicate the presence of Ion exchange processes. Selectlvlty data also Indicate that the retention mechanism is reverse phase In nature. Appilcatlons of picolyi Kel-F for the separation of simple drug mixtures, nltrophenols, and carboxylic acids are possible.
A variety of materials have been developed for use in anion exchange chromatography. These materials are based on either silica or polystyrene-divinylbenzene. Bonded ion exchange stationary phases have been prepared with silica as the core by derivatization with organochlorosilanes (1,2) or by polymerization of the silica pores followed by chloromethylation and amination (3). The macroreticular crosslinked polystyrene beads, XAD-1, have been converted to anion exchange resins by aminating the surface of the beads (4-7). Chromatographic packing with multifunctional bonded phases have also been prepared. Octadecylsilane was bonded to commercially available cation and anion exchange packings (8). Reverse phase/anion exchange materials were prepared by derivatizing silica with monochloroalkylsilanes and silylalkyl halides followed by amination of the terminal chloro groups with benzyldimethylamine. The ratio of hydrophobic to ionic character was controlled during the synthesis and used 0003-2700/65/0357-0340$01.50/0
as a means to vary the stationary phase to effect the desired separation (9). Almost all investigations involving the separation of anions employ strong anion exchange columns. Tertiary resin bonded amino groups acting as weak anion exchange sites, such as pyridine groups, have been only slightly studied. Silica has been derivatized with /3-(trichlorosilyl)-2-ethylpyridine and the capacity factors of azobenzene dyes and steroids have been investigated (10). Alumina has been used as a support with a bonded phase formed by using 2-(4-pyridyl)-l-(trichlorosily1)ethane. This material has been used for the separation of aromatic hydrocarbons, barbiturates, and nucleoside components (11). A detailed study of the ion exchange properties, especially pH effects, was not made for either the pyridyl bonded alumina or silica. Poly(viny1pyridine)resins have been used for the selective sorption of metals such as cobalt and copper (12). In addition, poly(viny1pyridinium dichromate) has been prepared and used as an inexpensive recyclable oxidizing agent for organic synthesis (13). The synthesis of various divinylbenzene-vinylpyridine polymers and their subsequent characterization as reversed-phase HPLC packings has just been reported (14). Porous vinylpyridine polymers apparently have not been used to any great degree for anion exchange HPLC. Previously our research program has demonstrated that poly(chlorotrifluoroethy1ene)or Kel-F can be derivatized with organomagnesium and organolithium reagents (15). Butyl and phenyl modified Kel-F have been used as reversed-phase packing materials for HPLC (16,17). Phenyl-modified Kel-F (PHKF) has been further derivatized to form both strong cation and anion exchange packing materials (18). Because 0 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985
the capacity of a weak ion exchange resin is dependent on the mobile phase pH, we felt a weak ion exchange group synthesized on a hydrophobic support could permit solute retention by either an ion exchange or a reverse-phase mechanism as desired. This report describes the derivatization of Kel-F with 4-picoline using organolithium chemistry and the subsequent HPLC characterization of this new polymer as a column packing. The synthesis of picolyl Kel-F (PIKF) has not been previously reported so optimization of reaction conditions was explored. Because of the hydrophobic nature of Kel-F, the reverse-phase property of PIKF should be present. In addition, a plot of the fraction of protonated 4-picoline vs. pH indicated that the polymer should potentially be an anion exchange resin in the pH range of 1.0-6.0. Therefore the ion exchange retention of certain solutes as a function of ionic strength and pH was expected and found. The dependence of solute retention with the composition of mixed solvents was also investigated. The application of PIKF for the separation of various charged and neutral organic species was demonstrated. Control of the retention of anions by changing the capacity of the packing using pH was possible. EXPERIMENTAL SECTION Preparation of Column Packings. Kel-F 6061 (100% poly(chlorotrifluoroethy1ene))was ground and sized as previously described ( I 7). 4-Picoline (Aldrich Chemical Co., Milwaukee, WI) was purified by vacuum distillation at 125 "C. The exchange reaction of Walter (19) was used for the preparation of picolyllithium. A desired volume (100mL for 101stoichiometry) of 1.55 M n-butyllithium (Aldrich Chemical Co., Milwaukee, WI) mixed with 220 mL of gold label THF (Aldrich Chemical Co., Milwaukee, WI) was cooled to -70 "C and purged with helium for 30 min. A known volume of 4-picoline (16.7 mL for 101 stoichiometry) was added dropwise over a 15-min period. The reaction mixture was stirred at -10 "C for 20 min. The appearance of a goldenorange color indicated the presence of picolyllithium. The picolyllithium was then injected, via a syringe through a septum, into 100 mL of dry He-purged THF at 0 "C which contained 2 g of Kel-F. After complete addition of the picolyllithium within 30 min, the cooling bath was removed. The reaction mixture was allowed to warm to 25 "C and stirring was continued for another hour. For termination of the reaction, 3-5 mL of water was added dropwise. The product was filtered and washed with hexane, acetone, methanol, and water. Kel-F is white, while the derivatized polymer, as expected from other organolithium reactions, was brown. The exchange capacity of the PIKF polymer was determined by titration of a methano1:water (1090 % v/v) slurry of the polymer with 0.003 N HCl. Elemental analyses (C, H, N, C1, and F) were performed by Galbraith Laboratories (Knoxville, TN). After KBr pellet preparation, further characterization of the polymer by infrared spectrometry was performed with a Perkin-Elmer Model 683 infrared spectrophotometer and a PerkinElmer Model 3500 infrared data station (Norwalk, CT). Surface area measurements were calculated from the BET adsorption isotherm of nitrogen. Particle size distribution was measured with a Spectrex laser particle counter (Redwood City, CA). Chromatography. Fines in the PIKF were removed by repetitive sedimentation in methanol. Stainless steel columns, 150 mm X 4.1 mm i.d. were packed at 7000 psi (48 MPa) with a methanol slurry of the polymer using a Model 10-600-300pneumatic amplifier pump (SC Hydraulic- Engineering Corp. Los Angeles, CA) and a high-pressure slurry packer (Alltech, Deerfield, IL). Chromatographic measurements were made with an IBM LC/9533 ternary gradient liquid chromatograph and operator station (Danbury, CT) and a Fisher Recordall Series 5000 recorder (Houston Instruments, Austin, TX). All solvents used as mobile phases were distilled-in-glass quality (Burdick and Jackson or Mallinckrodt). The organic and inorganic chemicals used as chromatographic solutes and buffers were generally reagent grade and were from a variety of sources. Column efficiency data were reported as plate height, HETP where HETP = L / N and N = 5 . 5 4 ( t ~ / u ~ /Retention ~)~. data were reported as capacity factors, k', where k'= ( t R - t o ) / t o The abbreviations L , wlj2, t R , and to
341
Table I. Addition of Picolyl Groups t o Kel-F Based on Elemental Analysis reaction stoichiometry Kel-F 1:l 3:1 101
C
H
21.30 22.44 24.15 23.00
0.00 0.22 0.57 0.54
element, % N c1 29.40 28.46 28.00 27.08
0.00 0.21 0.40 0.39
F 49.30 48.64 46.30 46.21
Scheme I CH.Li
\ F
IY
a/,
L; c-c
+ LiCl
correspond to column length, peak width at half height, and retention time of the retained and unretained solute,respectively. The separation factor, a, = k i / k < . RESULTS AND DISCUSSION Characterization of Picolyl Kel-F. The expected derivatization of Kel-F was qualitatively identified by infrared spectrometry. Absorption bands present in underivatized Kel-F occur at 1000 to 1400 cm-l, due primarily to the C-F stretch and at 940 cm-l due to the C-Cl stretch. Upon reaction of Kel-F with picolyllithium, the peak appearances of the out-of-plane bend at 850 cm-l and the C=C and C=N frequencies at 1500 to 1600 cm-' indicated addition of the picolyl moiety to the polymer. The effect of stoichiometry was studied for the reaction between picolyllithium and Kel-F using elemental analysis data of C, H, N, and C1 (Table I). A 1:l stoichiometry of picolyllithium to Kel-F gave a slight increase in hydrogen, nitrogen, and carbon, and a decrease in chlorine. A stoichiometry of 3:l showed a doubling of percentage for hydrogen and nitrogen with a continued increase in carbon. For a stoichiometry of l O : l , the coverage of hydrogen, nitrogen, and carbon was essentially the same as the 3:l stoichiometry. On the basis of these data, extent of modification of Kel-F by picolyllithium was generally only about one-third as great as compared to butyl or phenyllithium (15). T o ensure a complete as possible reaction between picolyllithium and Kel-F, the 1O:l stoichiometry was used for further preparations of PIKF. The chemical reaction occurring between picolyllithium and Kel-F was believed to be that shown in Scheme I. Any possible loss of fluorine was small as evidenced by a total element percent of 97.22. The reaction between picolyllithium and Kel-F was believed to proceed by a oneelectron exchange mechanism (15). The anion exchange capacity of picolyl Kel-F was determined by titration of the polymer with 0.003 N HC1. Titration of the 3:l polymer gave an exchange capacity of 4.7 wequivlg. The exchange capacity of the 1O:l polymer was determined
342
ANALYTICAL CHEMISTRY, VOL. 57,
NO. 1, JANUARY 1985
I
k'
4 4 \
3.0
k'
1
t
I I
I
I
01005 01010 (Citrate),M
'0!001
I
I
0.'015
01020
Flgure 1. Effect of ionic strength on the retention of m-chlorobenzoic acid: mobile phase, methano1:sodium citrate buffer, pH 5.0 (60:40, v/v); flow rate, 0.5 mL/min.
l
I
,d
I
3.0
I
I
I
I
I
5.0
7.0
pH
(44
Flgure 3. Retention (k') vs. pH using phenol (8)and picoline as the solutes: mobile phase, methanol:0.0096 M sodium citrate (40:60, v/v) for €3 and polnts; mobile phase for [ points the same except 0.0096 M phosphate used; flow rate, 0.5 mUmin.
k'/
II
4
6,
\6
\
PH Flgure 2. Retention (k') vs. pH using nitrate (X) m-chlorobenzoic acid
(e),and 2,4dinitroaniline (4)as the solutes:
mobile phase, metha-
nol:0.0096 M sodium citrate (60:40,v/v); flow rate, 0.5 mL/min.
to be 9.5 pequivlg. The particle range was quite wide, from about 10 to 100 pm for this sample. Titration of the 1O:l polymer with a tighter particle size range of about 20-35 pm gave an exchange capacity of 21.6 pequivlg. The entire 1 O : l synthesis procedure was repeated by using another 2-g sample of the 20-30 pm Kel-F, and upon titration, the PIKF product was found to have a capacity of 22.5 pequivlg. This average value of 22 pequivlg falls within the exchange capacity range of 5-50 pequiv/g listed for pellicular packings (20). The low coverage of picolyl groups, as evidenced by elemental analysis, and the low capacity indicated that the derivatization was basically a surface phenomenon. Recent ESCA and ATR infrared spectroscopy results also confirmed the surface nature of the coverage (21) for modified Kel-F. Chromatographic Properties. The effect of solute retention with ionic strength was explored first. Figure 1 illustrates a plot of capacity factor (k') for m-chlorobenzoicacid vs. ionic strength expressed as the concentration of citrate. The optimum citrate concentration was determined to be 0.0096 M. Above an ionic strength of 0.01 M citrate, the retention of m-chlorobenzoic acid was essentially independent of ionic strength. Solute retention as a function of pH was investigated with nitrate, m-chlorobenzoic acid, and 2,4-dinitroaniline as the
test solutes (Figure 2). As the pH decreased, the k'of nitrate increased. Since nitrate is always an anion, the k' should increase as the pH decreases, assuming the protonation of the picolyl group increases. As the pH decreased from 7.0 to 4.0, the k'of m-chlorobenzoic acid increased due to ion exchange between solute and stationary phase. At a pH below 4.0, the solute (pK, = 3.82) becomes protonated causing repulsion between the solute and the stationary phase, decreasing the k'. Retention of 2,4-dinitroaniline increased from a pH of 7.0 to 4.5 due to an increased probability of hydrogen bonding between the solute and the picolyl moiety. At lower pH values, 2,4-dinitroaniline (pK, of 4.57) becomes protonated and therefore the k'decreased. A comparison of chromatograms at pH 3.5 and 4.5 showed there was a reversal in the retention order of nitrate and m-chlorobenzoic acid. The k ' of mchlorobenzoic acid decreased from 3.0 to 2.0 and for nitrate the k ' increased from 1.0 to 3.0. The retention of phenol and picoline as a function of pH was also investigated (Figure 3). Because phenol has a pK, of 9.89, its k'was fairly independent of pH due to protonation at the pH range of interest. The k'of picoline decreased at the lower pH as expected, but surprisingly increased from pH 7.0 to 6.0. It was thought the k'would remain constant at the higher pH values. The buffering range of citrate is from a pH of 2.1 to 6.4, therefore the k'of picoline was investigated with a phosphate buffer at a pH of 6.0-7.5. A similar trend was also observed with the phosphate buffer. The pK, of picoline is 6.0 and from the titration data we determined that the pK, of the picolyl moiety on the polymer was about 7.0. As the pH decreased from 7.5 to 6.0, the polymer became protonated and the probability of hydrogen bonding increased between the solute and the stationary phase. With a further decrease in pH, the solute became protonated, setting up a repulsion between picoline and the stationary phase, thus lowering the k'value. The data in both Figures 1and 2 show the retention mechanism of PIKF is clearly a function of the ionized state of both the packing and solute. To obtain a better understanding of the reverse-phase interaction between PIKF and substituted aromatics, the retention of various organic acids and bases was investigated as a function of the percent organic modifier in water. The acids tested were benzoic acid and o-chlorobenzoic acid. These acids, as well as most other aromatic carboxylic acids, were unretained by PIKF even with mobile phases of only 20%
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985
343
Table 11. Retention Characteristics for Various Functionalized Aromatics” compound benzamide benzyl alcohol
phenylethanol pyridine benzaldehyde
phenol
0.11
benzylamine 0 10
20
30
40
I
I
I
50
60
70
aniline benzonitrile benzene picoline anisole nitrobenzene
SURFACE TENSION, dyne/cm Flgure 4. log a for the CH, group vs. surface tension for PHKF (A), C-18 (0),and PIKF (0): mobile phase, methanokwater (1OO:O to 7 5 9 2 . 5 , v/v).
methanol. Possibly repulsion between anions and the lone pair of electrons on the nitrogen of the picolyl moiety is a dominant factor. The bases that were investigated were p chloroaniline (pK, of 4.0) and o-nitroaniline (pK, of 10.0). The capacity factors ranged from 0.3 to 9.7for a 80-40% methanol content in the mobile phase. The k’ of o-nitroaniline was consistently higher by two units. Since p-chloroaniline would not be protonated in the mobile phases investigated, hydrogen bonding would only exist between it and the picolyl moiety while protonated o-nitroaniline could interact more strongly with the lone pair of electrons on the picolyl nitrogen. Therefore, when using the polymer in the reverse-phase mode, a preference seems to be shown for cations and neutral species. The dependence of solute retention on the volume fraction of methanol was also investigated for benzyl alcohol and phenylethanol. Plots of log k’for benzyl alcohol and phenylethanol vs. organic volume fraction for methanol in water were linear, with k’values ranging from about 10.6 for 10% methanol to 1.2 for 40% methanol. At 100% water, the k’ values were interpolated to be 9.0 and 24.0 for benzyl alcohol and phenylethanol, respectively. A similar study has alsd been reported with PHKF (17). The k’values were consistently about 10 times larger on PHKF. For this packing, essentially irreversible adsorption of the solutes was noted, with k’values interpolated to be approximately 600. This difference is probably due to the greater extent of modification of PHKF. For a truly liquid-liquid chromatographic system, it has been shown that the relationship between a, the separation factor between two solutes differing by one group substituent, and surface tension, y,is given by log a =
A(AGvnw)
+ rN(AAj - AAt’) 2.3RT
where N , R , and Tare Avogadro’s number, the gas constant, and the absolute temperature, respectively, AGWw is the van der Waals component of the free energy of interaction of the solute with the solvent, and ( A A j - AAi) is a group constant describing the effect of a substituent in altering the surface contact area of the solute with the stationary support (22). Figure 4 illustrates the plot of log a for the methylene group vs. y of methanol-water for (2-18, PHKF (23), and PIKF packings. For all three columns, the plots are quite linear from about 30 to 40 dyn/cm or 70 to 30% methanol content; however curvature in the plot from 30 to 10% methanol (34-55 dyn/cm) definitely exists. Linearity between log a and y has been shown previously for various functional groups for surface tension values ranging from 20 to 40 dyn/cm (24). However, upon plotting log a data derived from k’values reported by Horvath et al. (25) for cinnamic and benzoic acids (CH=CH group) vs. surface tension, a gradual curvature at 40 dyn/cm and above is observed consistent with our results. The cause
k’ 0.72 0.83 1.6 1.7 2.6 2.8 2.8 2.9 3.9 6.1 7.2 7.6 7.9
01‘
0.11
0.13 0.26 0.28 0.43 0.46 0.46
0.47 0.64 1.0 1.1 1.2
1.3
Mobile phase 15% acetonitri1e:water. as = k’ of functionalized benzenelk‘ of benzene. a
of nonlinearity is unclear at this time. One possible explanation is it is assumed the bulk surface tension and the microscopic surface tension of the mobile phase are the same. The factor between the two, k”, has been shown to be fairly constant from 30 to 100% methanol but decreases linearily with lower organic percentages (22). Interestingly, the log a data for the three packings were consistently ordered PHKF > C-18 > PIKF. This order is likely controlled by the degree of accessible hydrophilic groups. PHKF should be very hydrophobic while the residual silanols of C-18 and the heterocyclic nitrogens of PIKF could lower the nonpolar nature of the surface. Picolyl Kel-F was further characterized by investigating the capacity factor, k’, of various functionalized aromatic compounds. Capacity factors for 13 compounds are listed in Table 11. Similar data have also been reported for PHKF (17). As expected, the capacity factors were generally smaller on PIKF by about a factor of 2 due to the lighter loading of the picolyl moiety. Many of the compounds followed the same retention order on both polymers. Notable exceptions include phenol, aniline, anisole, and nitrobenzene. The higher k’ of phenol on PIKF is probably due to the following type of interaction. Recently Kawabata and Ohira have used 4-vinylpyridinedivinylbenzene for the removal of phenol from aqueous solutions (26). The mechanism of interaction they proposed was that the nitrogen of the polymer abstracts a hydrogen from phenol allowing a complex to form between the protonated polymer and the phenoxy anion. The slight rise in k‘ with pH for phenol shown in Figure 2 also fits this mechanism. The greater retention of aniline on PIKF could be due to hydrogen bonding between the solute and the picolyl moiety. The greater relative retention of nitrobenzene and anisole on PIKF is more complex and cannot be explained based on a charged transfer interaction between the solute aromatic ring and the picoline moiety. The axvalues for the aromatics with polar functional groups such as OH or NH2 were 3 to 4 times higher than those found for PHKF. The combination of an enhanced retention through hydrogen bonding for polar aromatics and a lower hydrophobic retention for benzene on PIKF is responsible. However it is clear based on these data that unprotonated PIKF is not an identical reverse-phase packing to either PHKF or C-18 silica (27),which both retain this set of solutes in the same order. Two other compounds of interest studied only on PIKF are pyridine and 4-picoline. A similar moderate retention was expected for both solutes. The hydrophobic effect of the methyl group was greater than expected for the retention of 4-picoline. The plate height, HETP, was calculated for several of the compounds listed in Table 11. For phenol, benzonitrile, and nitrobenzene the plate heights were 0.183, 0.343, and 0.716
344
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985 T 0.01 a.u.
2
1 T
’ 2
0.01 a.u.
1 . 3
1
3
I
I
0
19
I
Time, min
Flgure 5. Chromatogram of (1) toluene, (2) p-nitroanlllne, and (3)
T
I
I
phenanthrene (all 50 pg/mL): mobile phase, methanoi:water (80:20, v/v); flow rate, 0.5 mL/min.
13 T h e , min
0
Figure 7. Chromatogram of (1) pyridoxamine (R = CH,NH,, 500 pg/mL), (2) pyridoxal (R = -CHO, 50 pg/mL), and (3) pyridoxine (R = -CH,OH, 200 pg/mL): mobile phase, acetonltri1e:water (20:80, v/v); flow rate, 0.5 mL/min.
0.01 a.u.
T 2
0.02 a.u.
3
0
10 Time, min
20
Figure 8. Chromatogram of (1) salicyllc acid (300 pglmL), (2) salicylanilide (50 pg/mL), and (3) salicylhydroxamlc acid (450 pg/mL): mobile phase, methano1:water (70:30,vlv); flow rate, 1.O mL/min. I
I
0
112
Time, min
Figure 6. Chromatogram of (1)aniline, (2) p-nitroaniline and (3) 2,4dinboanlline (all 60 pg/mL): mobile phase, methano1:water (80:20,v/v); flow rate, 0.5 mL/min.
mm, respectively. Although these plate heights are quite a bit higher than what is theoretically predicted by the Knox equation for “good” pellicular columns, they compare favorably to similar data published by Sugii et al. (14) for various 10-15 pm divinylbenzene-vinylpyridine HPLC packings. For example, their HETP values for phenol ranged from 0.7 to 2.27 mm depending on the polymer type and degree of cross linking. The major causes of the disappointing efficiency of the PIKF column were undoubtedly the relatively wide size distribution and the low surface area of the packing. Using a Spectrex laser particle counter, approximately 75% of the particles were found to be between 8 and 45 pm. The surface area was measured to be only 1.6 m2/g. Considering these data, the chromatograms to follow are quite competitive to those reported for vinylpyridine HPLC packings of narrower particle distribution and a higher surface area (14). Several applications employing the PIKF column were carried out. The separation of toluene, p-nitroaniline, and phenanthrene is shown in Figure 5. Phenanthrene is retained the longest on PIKF due to hydrophobic interactions with the Kel-F backbone. p-Nitroaniline is retained longer than toluene
due to hydrogen bonding between the amine and the nitrogen of the bonded phase. Separation of the same mixture on PHKF gave an elution order of p-nitroaniline, toluene, and phenanthrene (17). Figure 6 shows the separation of aniline, p-nitroaniline, and 2,4-dinitroaniline. In addition, the separation of o-nitrophenol,p-nitrophenol, and 4,6-dinitro-o-cresol has been carried out on the PIKF column using a 30:70 acetonitri1e:water mobile phase. Previously we have shown that PIKF has an affinity for polar solutes (such as the k’of nitrobenzene, Table 11). Because the positive charge of the ring is enhanced by nitro groups, a charge transfer interaction with the high electron density of the nitrogen heterocycle is promoted. Although the column efficiency of picolyl Kel-F is presently not high enough for the separation of multicomponent samples, this PIKE’ column was useful for the separation of simple mixtures of pharmaceutical interest. Figure 7 shows the separation of compounds that consitute vitamin Bg: pyridoxamine, pyridoxal, and pyridoxine. Generally these compounds are separated by cation exchange chromatography with a retention order of pyridoxal, pyridoxine, and pyridoxamine (28). Protonation of both nitrogens of pyridoxamine would explain its longer retention by cation exchange. Hydralazine, an antihypertensive drug, and hydrochlorothiazide,a diuretic, are combined in drug mixtures for the treatment of hypertension. The separation of hydralazine and hydrochlorothiazide was accomplished with base line resolution in 13 min
ANALYTICAL CHEMISTRY, VOL. 57, NO. 1, JANUARY 1985
A.
I
I
I
1
T O m By
I
2
1.4
Figure 9. Chromatogram of (1) acetylsalicylic acid (270 pg/mL), (2) caffeine (50pg/mL), (3) acetanilide (29 pg/mL), (4) salicylamide, (200 pg/mL), and (5)Phenacetin, (50 pg/mL); (A) mobile phase, methanokwater, pH 7.1 (17:83, v/v); (B) mobile phase, methanol:2.5 X M sodium acetate, pH 5.5 (17:83, Wv); (C) mobile phase, methanol:2.5 X M sodium acetate, pH 4.1 (17:83, vlv). Flow rate was 1.0
mL/min for A, B, and C. using a 25:75 acetonitri1e:water mobile phase. Again the more hydrophilic solute hydrochlorothiazide has the longer retention; the two peaks are reversed when PHKF is used. Salicylanilide and salicylhydroxamic acid (pK, of 5.19) are topical antifungal agents. The separation of salicylic acid and the topical antifungal agents is illustrated in Figure 8. This separation was investigated with methanollwater, primarily in the reverse-phase mode. Although possessing two aromatic groups, salicylanilide was retained less than salicylhydroxamic acid. The pK, of the latter compound is only 5.2 inferring the presence of two OH groups that could both hydrogen bond to the picolyl moiety by a mechanism similar to that proposed by Kawabata (26). Various analgesic substances, including acetylsalicylic acid, acetanilide, salicylamide, phenacetin, and a stimulant, caffeine, have been separated in Figure 9. Caffeine, acetanilide, and phenacetin are present in Bromoseltzer tablets and acetylsalicylic acid (pK, = 3.5) and salicylamide are present in aspirin tablets. A comparison was made between the reverse-phase and the ion exchange modes. For this separation, the reverse-phase mode did not give resolution between caffeine, acetanilide, andd salicylamide. At a pH of 5.5 the separation of all five solutes WBS possible. Decreasing the pH
345
to 4.1 resulted in the K'of acetylsalicylic acid increasing to that of salicylamide. Therefore only four solutes were resolved at this pH. Clearly, the increase in the ion exchange capacity of the column was responsible. In general, the ion exchange capability of PIKF was particularily effective at or below pH 4. This property of PIKF was demonstrated by the resolution of relatively small similar anions such as acrylate and methacrylate in about 10 min using a 10:90 methano1:sodium acetate, pH 4 mobile phase. In addition, the cis and trans isomers maleate and fumarate could also be distinguished under similar conditions. Column performance during the course of this study was reasonable. The column stability of PIKF was quite good with pressures of only 500 psi with a variety of mixed organicwater or buffered mobile phases at 1mL/min. Columns were stable for up to 4 months with steady use. After approximately 8 months of use, the column was unpacked and the polymer titrated with 0.003 N HCl. The exchange capacity of the polymer remained constant at 20 pequiv/g. However, peak tailing occurred for most components having a retention time greater thdn 15 min. The peak asymmetry factors for phenol, benzonitrile, and nitrobenzene were 0.75, 1.5, and 2.0, respectively. The increase in HETP values with k' for these solutes also reflects this problem. The peak tailing may be partially caused by the mixed retention mode due to the nature of the packing and primarily by a poorly packed column due to the heterogeniety of particle size (20). Future work will be directed toward the improvement of the surface area and the particle distribution of the PIKF packing. Kel-F 6300, a copolymer of 97% chlorotrifluoroethylene and 3% vinylidene fluoride, has been found to react more extensively with organometallics than Kel-F 6300 (16). Additional elutriation steps using a small capacity sedimentation pipet appear to be required both before and after reaction. These procedures should lead to modified Kel-F packings of better efficiency and less peak tailing. Hopefully, use of PIKF because of its low and controllable ion exchange capacity for ion chromatography would be a feasible extension of this work.
ACKNOWLEDGMENT The authors thank Alan ULlman of the Procter and Gamble Company and D. Lautermilk of the Armco Corp. for assistance with the particle size and the surface area measurements, respectively. Registry No. Toluene, 108-88-3; p-nitroaniline, 100-01-6; phenanthrene, 85-01-8;aniline, 62-53-3;2,4-dinitroaniline,97-02-9; o-nitrophenol,88-75-5;p-nitrophenol, 100-02-7;4,6-dinitro-o-cresol, 534-52-1;pyridoxamine, 85-87-0; pyridoxal, 66-72-8; pyridoxine, 65-23-6; hydralazine, 86-54-4;hydrochlorothiazide, 58-93-5; salicylanilide, 87-17-2; salicylhydroxamic acid, 89-73-6;salicylic acid, 69-72-7; acetylsalicylicacid, 50-78-2;acetanilide, 103-84-4;salicylamide, 65-45-2; phenacetin, 62-44-2; caffeine, 58-08-2. LITERATURE CITED Cox, G. 6.;Loscombe, C. R.; Slucutt, M. J.; Sugden, K.; Upfield, J. A. J . Chromafogr. 1976, 117, 269. Asmus. P. A.; Low, C.-E.; Novotny, M.J . Chromatogr. 1976, 119, 25. Caude, M.;Rosset, R. J . Chromatogr. Sci. 1977, 15, 405. Gierde, D. T.; Frltz, J. S. J . chromafogr. W79, 176, 199. Gjerde, D. T.; Schmuckler, G.; Fritz, J. S. J . Chromafogr. 1979, 186, 509. Gjerde, D.T.; Schmuckler, G.; Fritz, J. S. J . Chromatogr. 1980, 187, 35. Barron, R. E.; Fritz, J. S. React. f o l y m . 1983. 1 , 215. Liao, J. C.; Vogt, C. R. J . Chromatogr. Sci. 1979, 17, 237. Crowther, J. B.; Hartwick, R. A. Chromafographia 1982, 16, 349. Majors, R. E.;Hopper, M. J. J . Chromafogr'.Sci. 1974, 12, 767. Knox, J. H.; Pryde, A. J . Chromatogr. 1975, 112, 171. Sugli, A.; Ogawa, N.; Iinuma, Y.;Yamamura, H. Talanfa 1981, 28, 551. Frechet, J. M. J.; Darling, P.; Farrall, M. J. J . Org. Chsm. 1981, 4 6 , 1728-1730. Sugii, A.; Ogawa, N.; Harada, K.; Sato, I . J . Chromatogr. 1984, 294, 185.
346
Anal. Chem. 1985, 57,346-348 Danielson, N. D.; Taylor, R. T.; Huth, J. A.; Slergiej, R. W.; Galloway, J. G.; Paperman, J. B. Ind. Eng. Chem. Prod. Res. Develop. 1983, 22, 203.
Huth, J. A.; Danielson, N. D. Anal. Chem. 1982, 54, 930. Siergiej, R. W.; Danielson, N. D. Anal. Chem. 1983, 5 5 , 17. Siergiej, R . W.; Danielson, N. D. J. Chromatogr. Sci. 1983, 21, 362. Smlth, L. I . "Organic Synthesis"; Wiley: New York, 1943; Vol. 23, p 83.
Snyder, L. R.; Kirkland, J. J. "Introduction to Modern Liquid Chromatography"; Wlley-Interscience: New York, 1979; p 418. Mattern, D.; Danielson, N. D.; Hercules, D. M., unpubllshed results. Horvath, C. "HPLC, Advances and Perspectives"; Academic Press: New York. 1980: Vol. 2. OD 215-217. (23) Siergiej, R. W. Ph.D. Dissertation, Miami University, 1982, p 76. (24) Rlley, C. M.; Tomlinson, F.; Jefferies, T. M. J. Chromatogr. 1979, 185, 197. (25) Horvath, C.; Melander, W.; Molnar, 1. J . Chromatogr. 1976, 125, 129.
(26) Kawabata, N.; Ohlra, K. fnvlron. Sci. Techno/. 1979, 13, 1401. (27) Tanaka, N.; Goodelf, H.; Karger, B. L. J . Chromatogr. 1978, 158, 233. (28) Willlams, R. C.:Baker, D. R.; Schmlt, J. A. J. Chromatogr. Scl. 1973, 1 1 , 618.
RECEIVED for review July 9,1984. Accepted October 29, 1984. This work was supported in part by grants from the Research Corporation and the donors of the Petroleum Research Fund, administered by the American Chemical Society. Donation of the Model 9533 liauid chromatoaraDh bv IBM to the Chemistry Departmentis greatly appreciated. This work was presented at the 35th Pittsburgh Conference on Analytical Chemistry and Spectroscopy, Atlantic city, NJ, March 8, 1984.
Simultaneous Gas-Liquid Chromatographic Determination of Aldonic Acids and Aldoses Jacob Lehrfeld Northern Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture, Peoria, Illinois 61604
A method has been developed for the slmultaneous quantltatlon of several aldonic aclds and aldoses. The aldoses are converted to aldltol acetates, and the aldonic aclds are converted to N-propylaldonamlde acetates. A standard mixture contalnlng the derlvatlzed rlbonlc, mannonlc, gluconlc, and galactonlc aclds and arabinose, xylose, mannose, glucose, and galactose was separated by GLC In 18 mln. Phenyl P-o-glucopyranoslde was used as an Internal standard.
Analysis of complex mixtures, such as those found in fermentation broths and pulping residues, often requires multiple analytical metbodologies. For example, a kinetic evaluation of the production of gluconic acid by Pseudomonas ovalis (1) required separate analyses of glucose and gluconic acid. A simple method which would simultaneously quantitate both components would be useful. A differential method (2) for quantitating mixtures such as this is available. However, it is tedious and time-consuming because it requires three borohydride reductions and a duplicate analysis. A new procedure, described here in Figure 1,is capable of simultaneously analyzing a wide variety of aldose and aldonic acid mixtures. Aldoses are reduced to their corresponding alditols, and the aldonic acids are converted into the corresponding Npropylaldonamides. The new derivatives are acetylated by pyridine-acetic anhydride; the peracetates are readily separated by GLC in 18 min on a SP 2340 column. EXPERIMENTAL SECTION Materials. Propylamine, L-arabinose, D-Xylose, D-mannose, D-glucose, and D-galactose were obtained from Aldrich Chemical Dco. Phenyl 0-D-glucopyranoside, ~-g~ucono-1,5-lactone, galactono-1,4-lactone,L-mannono-l,4-lactone,sodium D-glUCOnate, and myo-inositol were obtained from Pfanstiehl Laboratory, Inc. Potassium ribonate and ammonium xylonate were a gift from M. E. Slodki. GLC-coated support (3% Sp 2340 on 100/120 mesh Supelcoport) was obtained from Supelco, Inc. Cation-exchange resin (AG 5OW-XB 200-400 mesh H') was obtained from Bio-Rad Laboratory. GLC Analysis. Analysis by GLC was performed on a Packard Instrument Model 428 gas chromatograph equipped with dualflame ionization detectors and dual electrometers. Peracetylated derivatives were separated on a glass column (1m x 2 mm i.d.)
packed with 3% SP 2340. The temperature was programmed from 190 to 260 "C at a rate of 5 "C/min and held there until the last peak was eluted. Helium flow rate was 20 mL/min. Conversion of Aldoses and Aldonic Acids to Peracetylated Alditols and N-Propylaldonamides. A 0.1 M sodium carbonate solution (0.6 mL) was added to a mixture containing approximately 2 mg each of L-arabinose, D-Xylose, D-mannOSe, D-glucose, D-galactose, myo-inositol, D-ribonolactone, ~-mannono-1,4-lactone, D-galactono-1,4-lactone, and phenyl p-DD-glucono-~,~-~actone, glucopyranoside in a 16 X 125 mm culture tube. The solution was maintained at 30 "C for 1 h and then treated with sodium borohydride (0.5 mL of a 4% solution) for 1 l/z h at 22 "C. Excess sodium borohydride was decomposed by the dropwise addition of acetic acid (25%) until bubbling stopped (6-8 drops). To remove sodium ions, the solution was poured onto a column of cation-exchange resin (2 mL) and eluted with 7 mL of water. The eluate was evaporated to dryness in a Buchler vortex evaporator (45 "C in vacuo). Borate was removed by twice evaporating methanol (3 mL) from the residue. Heating at 85 "C in vacuo for 2 h converted the aldonic acids into aldonolactones. This residue was dissolved in pyridine (1mL), 1-propylamine (1 mL) was added, and the tube was capped and heated at 55 "C for 30 min. The solution was cooled (under 45 "C), the cap was removed, and nitrogen was bubbled through the reheated solution (55 "C) until dry. The residue was dissolved in pyridine (0.5 mL) and acetic anhydride (0.5 mL) and heated at 95 "C for 1h. Sodium gluconate, when treated as above, gave similar results. The sample is suitable for GLC as is; usual injection size is 0.8 pL. If sample size is small (less than 1-2 mg), tailing from pyridine and acetic anhydride can interfere with detection and/or quantitation. If so, remove them by bubbling nitrogen through the solution until dry and reconstitute with 100 pL of acetone, chloroform, or methylene chloride. Optimization of Lactone Hydrolysis. A mixture of Lmannono-l,4-lactone (39 mg), ~-glucono-1,5-lactone(41 mg), ~-galactono-l,4-lactone(39 mg), and phenyl P-D-ghcopyranoside (36 mg) was dissolved in 0.1 M sodium carbonate (8 mL) and kept at 30 "C. Aliquots (0.5 mL) were removed at 15, 30,45,60, 90, and 120 min and treated with a 4% sodium borohydride solution (0.5 mL). The aliquots were then treated as above. Similar experiments were performed at temperatures down to 22 "C and with more dilute base. RESULTS AND DISCUSSION The standard methods for the analysis of aldoses as alditol acetates (3) or aldononitrile acetates ( 4 , 5 )and aldonic acids
This article not subject to US. Copyrlght. Published 1984 by the American Chemical Society