Anal. Chem. 1983, 55,2109-2112
2109
Counterion Effects in Ion-Exchange Partition Chromatography Debra S. Dieter and Harold F. Walton’ Chemistry Department and Cooperative Institute for Research in Environmental Sciences, Campus Box 449, University of Colorado,, Boulder, Colorado 80309
Nonlonlc organlc compounds, including “parabens” and analgesic drugs, are separated by liquld chromatography on cation-exchange resins with aqueous eluents and different counterlons, Li, Na, K, Mg, and Ca. Retention volumes and band widths are affected by the counterion, with LI generally givlng the best chromatography. Water uptakes of the different resin forms were measured by centrifugation, and a theory based on ionic hydration Is proposed to account for the counterlon effects.
and noted certain trends (1,6). From these studies it seemed that retention increased in the order K+ < Na+ < Li+ and that doubly charged counterions gave greater retention than singly charged ions. A tentative explanation involving ion-dipole interactions was offered for the binding of trigonelline and other dipolar ions. This paper reports a study of chromatographic retention of a number of polar aromatic compounds on cation-exchange resins of two cross-linkings, carrying five cations, Li+, Nu+, K+, Mg2+,and Ca2+. Because ionic hydration seemed to be important, we measured the water uptake of the different resin forms, using a centrifugal technique.
Ion-exchange resins absorb neutral organic compounds from solutions and may be used for analyzing these compounds by liquid chromatography. The absorption is specially marked with aromatic solutes and resins of the common polystyrene type. Such solutes may, of course, be separated on columns of macroporous, nonionic polystyrene, but ion-exchange resins have certain advantages. They may be obtained in small particle sizes, 10 pm and below, the resins are wet by water and columns arie easy t o pack, they may be used in purely aqueous solutions or solutions with small proportions of organic solvent modifier, their capacity is high, their physical structure i s uniform and they give symmetrical elution bands, and their chromatographic behavior toward organic solutes can be “fine-tuned” by properly choosing the inorganic counterion. The nature of the calunterion is obviously important in “ligand-exchange chromatography”, where columns of cation-exchange resins are used that carry transition-metal ions, and these ions form complexes with the ligands to be separated. Copper(l1) is the counterion most often chosen, and spectaculm separations of amino acids and peptides by ligand exchange have been relported, including the separation of optical isomers of amino acids. Complex formation between the metallic counterion and the organic solutes is the basis of separation of organic acids on a lanthanum-loaded cation-exchange resin ( I ) , and undoubtedly affects the chromatography of carbohydrates on calcium-loaded (2) and silverloaded (3) cation-exchange resins. It is not generally recognized, however, that the inorganic counterion affects retention in cases where complex formation is out of the question or a t least extremely unlikely. The name “ioin-mode~atedpartition chromatography” has been given to the chromatography of organic compounds on ion-exchange resins. A review by Jupille e t al. (4) shows chromatograms of various compound classes-carbohydrates, organic acids, phenols, and chlorinated phenols-on cationexchange resin columns carrying three counterions, hydrogen, calcium, and lead. No systematic study of the function of the counterion is presented. The often-cited paper of Goulding (5)gives retention data for carbohydrates and sugar alcohols on a cation-exchange resin of the sulfonated polystyrene type with 13 counterions, but, again, no attempt was made to correlate the behavior of the different counterions. In our laboratory we have observed the effect of the counterion on the retention of various compounds, including xanthine5 and aromatic hydrocarbons, on gel-type and macroporous resins,
EXPERIMENTAL SECTION Instrumentation. The chromatographic system consisted of a Waters Model 6000A pump, Rheodyne Model 7120 injector with 20-fiL sample loop, Kratos-SchoeffelSF-770 variable-wavelength detector, Linear Instruments recorder, and Bioanalytical Systems Model LC-23A column heater with LC-22 temperature controller. The columns were of stainless steel, 4.6 mm i.d. and 15 cm long (for 8% cross-linked resin) or 11 cm long (for 4% cross-linked resin); they were packed upflow with a Micromeritics Model 705 stirred-slurry column packer. In packing the resin columns and in using them, careful attention must be paid to flow rate; if flow is too fast, the resin beads will deform and plug the tube; this problem is more acute the lower the cross-linking. The 8% cross-linked resin column was run at 0.5 mL/min and the 4 % cross-linked column at 0.3 mL/min. The temperature was normally 50 “C, though some runs were made at higher temperatures. The void volume was measured by injecting solutions of potassium bromide or potassium nitrate and noting the first sharp rise in absorbance. It varied somewhat with the counterion, but was close to 0.90 mL for the 14 cm, 8% cross-linked column. To measure water uptake, glass tubes 12 mm i.d., carrying coarse frits, were taken and cut 5 mm below the frit and 85 mm above. Resin was placed in the tubes, and the tubes were spun in 15-mL polycarbonate centrifuge tubes in which were inserted small rings cut from Tygon tubing to serve as seats for the fritted tubes. A Dynac centrifuge was used with horizontal head, giving about lOOOg centrifugal force; spinning time was 15 min. Tubes were spun in pairs to balance the centrifuge, and they and their contents were weighed to 0.1 mg after placing them in wide-mouth screw-cap bottles. Each series of measurements was started with air-dried Na-form resin, 100-200 mesh, in portions of about 0.7 g (dry basis). The resin was converted in situ into the desired cationic form. Knowing the initial moisture content and the exchange capacity of the resin, the weight of each resin form on a dry basis was easily calculated, and the weight of the swollen resin was found after wetting thoroughly with water and centrifuging. No matter what the spinning speed, some water is always hcld between the resin beads at the zones of contact (7), and we can expect some water to be held in the glass frits. To estimate the amount of water held in this way, we wetted portions of 8% cross-linked resin in the tubes with 0.02 M potassium iodide solution, spun the tubes, removed the solution from the bottoms of the polypropylene tubes, and then wetted the resin with water and spun the tubes again. We measured the amount of iodide ions removed from the resin in this second centrifugation by combining the water from two tubes, transferring it to a separating funnel, adding 0.01 M potassium dichromate and 0.5 M hydrochloric acid, and then extracting the liberated iodine with 3 mL
0003-2700/83/0355-2109$01.60/0 0 1983 American Chemical Society
2110
ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983
- 60
as
8% C R O S S L I N K I N G
- 50
E
I
0
s10.0
O . I y ACETATE, p H 5 . 5 , 0.5 M L / M I N . , 50"
- 40
-I
=
7.5
- 30
5.0
- 20
z
z W
+ 2.5 W
as
Li
No
K
Mg
Co
ACM +--+-+--+-' L i No K
-10 Mg
Ca
Flgure 1. Retention volumes vs. counterions. See text for abbreviations. Retention volumes are given in milliliters for a 15-cm column and are corrected for the void volume.
of methylene chloride and measuring intensity of absorption at 505 nm. From experiments of this kind we found that 1.54 g of Na-resin (dry basis), containing6.81 mequiv of Na+ and imbibing 1.32 g of water, retained, in all, 0.0011 mequiv of I- (average of two closely agreeing data sets). The theoretical Donnan retention would be 0.0001 mequiv of I-. Thus 0.0010 mequiv was held between the beads, correspondingto 0.05 mL of 0.02 M KI. This is only 4% of the observed water uptake by the Na-form resin. The correction is small, but it is beyond experimental error, and the water uptakes tabulated below have been corrected for the water held between the beads. Resins. The columns were packed with Benson 8% crosslinked cation-exchange resin, 7-10 pm, and Hamilton 4% cross-linkedresin, 10-15 pm, both obtained in the sodium form. The packed column with 8% cross-linkedresin, 15 cm long, had an ionic capacity of 4.42 mequiv, while that with 4% cross-linked resin, 11cm long, had a capacity of 1.75 mequiv. For the centrifuge tests a coarser resin was necessary, and Bio-Rad A 100-200 mesh, were used. The exchange capadties of were 4.41 and 4.75 mequivfg of dry Na resin for 8% and 4%
Compounds Studied. The following compounds, obtained in best commercial grades, were used as solutes and their chromatographic retentions measured. Abbreviations are used as shown to facilitate reading Figure 1: TRIG, trigonelline, N methylpyridinium-3-carboxylate;CAF, caffeine, 1,3,7-trimethylxanthine;THB, theobromine,3,7-dimethylxanthine;THP, theophylline, 1,3-dimethylxanthine; ACET, acetanilide, N acetylphenylamine; ACM, acetaminophen (paracetamol), 4'hydroxyacetanilide; PHA, phenacetin (acetophenetidin, 4'-ethoxyacetanilide; MEB methyl benzoate; ETB, ethyl benzoate; MEP, methylparaben, methyl p-hydroxybenzoate; ETP, ethylparaben, ethyl p-hydroxybenzoate. RESULTS AND DISCUSSION Retention Volumes with 8% Cross-Linked Resins. Most experiments were made with resin of 8% cross-linking, because it tolerated a higher flow rate in the column than did 4% resin and gave smaller volume changes when one cation was substituted for another. The results of a large number of tests at 50 "C are summarized in Figure 1. Figures 2 and 3 show typical chromatograms; in both cases the lithium counterion gives better performance than Na+ or K+. One is reminded that lithium buffers have been advocated for amino acid chromatography for many years (8). Certain trends are seen in Figure 1;six compounds show definite decreases in retention as one proceeds from Li to Na to K, and these confirm our earlier results (6);phenol and the parabens show decided increases in retention. A possible reason for this effect will be offered later. Noteworthy is the strong retention of
I
J Figure 2. Chromatography of a mixture of theophylline and caffeine with two counterions.
30 MIN.
Figure 3. Chromatography of a four-component mixture with two counterions. Chrornatographlc conditions are the same as those given in Figure 2.
Table I. Resin Cross-Linking and Retention retention, mL/mequiv of resin 4% 8% cross-linked cross-linked Li K L 4 i K caffeine acetanilide phenacetin methylparaben
2.2 2.5
1.7 2.2
5.0 6.7
4.0 8.0
1.3 1.65 3.5 5.6
1 .o
1.45 2.7 8.6
the benzoate esters and their p-hydroxy derivatives, the parabens. Effect of Cross-Linking. Table I compares retention volumes with 4% and 8% cross-linked resins, both at 50 "C. When the retention volumes are referred to the quantities of
ANALYTICAL CHEMISTRY, VOL. 55, NO. 13, NOVEMBER 1983
Table 11. Theoretical-Pl.ate Numbers in 15-cm Column of 8% Cross-Linked Resin at 50 “C and 0.5 mL/min counterion
arcetanilide
Li K Mg Ca
660 520 400 -.560
phenac- methyl methyl etin benzoate paraben 680 400 280 320
950 850 418 350
87 0 900 350 41 5
resin in the two columns, one sees that t b e lower cross-linking gives more retention. The implication is that the organic solute molecules must approach closely to the polymer chains in order to be absorbecl; being somewhere in the internal solution of‘the resin is not enough. We presume that the chief mechanism of absorption is a-electron overlap between the aromatic solute molecules and the benzene rings of the polystyrene chains. Plate Numbers. Table I1 shows plate numbers in the 15-cm column of 8% cross-linked resin for four counterions and four solutes, all at 50 “Cand 0.5 mL/min flow rate. Plate numbers can be increased by using slower flow; a t 0.3 mLJmin and Li+ counterion, they were 900 for phenacetin, 1000 for acetanilide. Plate heights were nearly the same with 4(% cross-linking as with 8%; the larger particle size of the 4(% cross-linked resin may offset the faster diffusion. Lithium ions give the highest plate numbers, ana the doubly charged ions cause poorer plate numbers. Temperature. Raising the column temperature from 50 “Cto 60 “C lowers the retention volumes with 8% cross-linked resin by 20-3070, corresponding to enthalpies of absorption of some 5000 cal/mol. With 4% cross-linked resin the effect is smaller. Many factors can affect the enthalpy, and very accurate measurements are necessary if ope is to identify entropy and enthalpy effects; therefore, the matter was not pursued. Some runs were made with the column at 70 “C and 80 “C,to see if resolutioin could be improved, but new peaks appeared that indicated. hydrolysis of esters and amides. Water Uptake. The results of the centrifuge tests described in the Experimental Section are summarized in Table 111. We want to q a k e a distinction between the water of hydration that is bound to the cations and the rest of the imbibed water, which we shall call “free” water. We used the quantities calculated by the Debye-Huckel theory for the distance of closest approach of the cations and anions in chloride solutions, using the activity coefficients of the salts in dilute solution; these numbers are tabulated by Harned and Owen (9). Harned and Oiwen considered i,h@ only the cations carry water of hydration, not the chloride ions; we in turn have ignored the hydration of the fixed sulfonate ions of the resins.
While the total water in the resin is greatest with Li+ (among the alkali-metal forms), it seems that the “free” water increaqes as one goes from Lit to Nat to K+.
CONCLUSIONS Ion-exchange resins may be regarded as organic solvents, mixed solvents in which the solubility properties of polystyrene are modified by the presence of water. The more water that is incorporated in the resin phase, the less effective the resin will be in absorbing aromatic organic compounds from aqueous solution. With this argument, the solvent properties of the resin should increase as we go from Li+ to Na+ to K+ as counterions. Actually the reverse order is found for many of the solutes shown in Figure 1. We rationalize this order by saying that the free water, as distinct from that bound by hydration of the counterions, increases as we go from Li+ to Na+ to K+. The rise in retention going from singly to douhly charged ions may likewise be explained by saying that there is less free water with the doubly charged ions; these ions are much more strongly hydrated than the singly charged ions. From Figure 1it is seen that there is a group of compounds, phenol, the “parabens”, and possibly acetaminophen, whose behavior is the reverse of that we have just discussed. Their retention increases as we go from Li+ to Na+ to K+ and drops abruptly when strongly hydrated Mg2+ is substituted for weakly hydrated K+. These compounds all carry phenolic hydroxyl groups. We suggest that the phenolic OH forms hydrogen bonds with the water in the resin and that water molecules that are coordinated with cations, Li+, Mg2+,and so on, are not available to form hydrogen bonds with phenolic OH. Hydrogen bonding i s a secondary effect that reinforces the primary attractive force, which in our opinion is a bonding between the aromatic solutes and the polymer. Hydrogen bonding and a bonding are affected in different ways by the “free water”, so that where both forces operate, retention could either increase or decrease as the free water increased. Going from Mg2+ to ea2+,when the “free water” may decrease slightly, all retentions increase. The cooperative effect of hydrogen bonding would explain the unexpectedly strong attachment of the parabens (p-hydroxybenzoates). These compounds are stronger acids than phenol by a factor of 10 (10) and would therefore form stronger hydrogen bonds. We must be wary of treating the resin as a homogeneous phase, for we have seen that the weakly cross-linked resin, which contains more imbibed water than the more strongly cross-linked resin and therefore might be a poorer solvent €or organic compounds, is actually a better solvent if one considers the retention per unit quantity of polymer. Attachment of organic solute molecules to the polymer probably takes place
--
-
Table 111. Water Uptake by Ion-Exchange Resinsa count er
ion
-
crosslinking
water per rnequiv of resin mg mmol
“bound” water, mmol/meauiv of i a t i o i
“free” water, mmol/meauiv of resin
217 12.0 214 11.9 6.5 5.4 4% 357 19.8 6.5 13.3 Nd1 8% 188 10.4 3.5 6.9 4% 313 17.4 3.5 13.9 K 8% 161 8.9 1.9 7 .O 4% 278 15.4 1.9 13.5 ME! 8% 185 10.3 6.95 3.35 4% 288 15.9 6.95 8.95 Ca 8% 164 9.1 5.95 3.15 4% 253 14.0 5.95 8.05 a Note: A uniform correction of 7 mg, or 0.4 mmol, of water per mequiv of resin has been applied to compensate for the water not removed bly centrifugation. The observed values are all 7 mg, or 0.4 mmol, higher than those tabulated. The estimated precision is t 2 mg, or 0.1 mmol of water per mequiv of resin. H Li
_______
8% 8%
2111
2112
Anal. Chem. 1983, 55, 2112-2115
in the first instance through a-electron interaction, which is a short-range effect. The number and state of binding of the water molecules close to the polymer chains must be considered, rather than the water between the polymer chains that is several molecular diameters away. Whatever their explanation, the data we have presented should help the selection of counterions in partition chromatography. Selectivities can be changed by changing the counterion, but the narrowest bands are obtained by using lithium counterions. ACKNOWLEDGMENT Part of this work was supported by the National Science Foundation, Grant No. CHE79-21294. Registry No. Li+, 17341-24-1;Na+, 17341-25-2;K+, 24203-36-9; Mg2+,22537-22-0;Ca2+,14127-61-8;TRIG, 535-83-1; CAF', 58-08-2; THB, 83-67-0; THP, 58-55-9; ACET, 103-84-4;ACM, 103-90-2;
PHA,62-44-2; MEB,93-58-3; ETB,93-89-0; MEP,99-76-3;ETP, 120-47-8;water, 7732-18-5. LITERATURE C I T E D (1) Otto, J. L.; de Hernandez, C. M.; Walton, H. F. J . Chromatogr. 1982, 247, 91. (2) Wood, R.; Cummings, L.; Jupille, T. J . Chromatogr. Scl. 1980, 18, 551. (3) Scobell, H. D.; Brobst, K. M. J . Chromatogr. 1981, 212, 51. (4) Jupille, T.; Gray, M.; Black, B.; Gould, M. Am. Lab. (Falrfleld, Conn) 1981, 13 (part 8),80. (5) Goulding, R. W. J . Chromatogr. 1975, 103, 229. (6) Walton, H. F.; Eiceman, G. A.; Otto, J. L. J . Chromatogr. 1979, 180, 145. (7) Parrish, J. R. J . Appl. Chem. 1985, 15, 280. (8) Modino, A.; Bonglovannl, G.; Fumero, S. J . Chromarogr. 1972, 7 1 , 363. (9) Harned, H. S.;Owen, B. B. "Physical Chemistry of Electrolytic Solutions", 3rd ed.; Reinhold: New York, 1958; p 525. (10) Wu, A.; Blehl, E. R.; Reeves, P. C. J . Chem. Soc., Perkin Trans. 1 1972, 449.
RECEIVED for review May 31, 1983. Accepted July 8, 1983.
Catalytic Activity in an I ndium-Sensitized Flame Photometric Chlorine Detector for Gas Chromatography Gregory Wells
Varian Instrument Group, Walnut Creek Division, Walnut Creek, California 94598
A sensltlve detector for halogenated organlc compounds for use in gas chromatography Is descrlbed. The detector Is based on an Indlum-sensltlred dual flame photometric detector. The column effluent Is decomposed In the lower flame, converted to the Indium hallde in the reglon between the flames, and excited In the upper flame to emit at a wavelength that Is characterlstlc of the lndlum hallde. The effect of varlous flame tip materlals on sensltlvlty, for chlorine detectlon, are dlscussed and evidence is shown to support a catalytlcally enhanced mechanlsm for the formation of InCI. Detection limlts In the low pg of Cl/s range have been obtalned by using nickel, tantalum, and actlvated stainless steel flame tlps. The sensltlvlty was found to decrease slowly over '" long periods of time for the catalytlc materials studled.
The determination of halogenated organic compounds is of particular interest in connection with environmental pollution. Many analyses that are done in this area by gas chromatography currently use packed columns and electrolytic conductivity detectors for the selective detection of halogens. The use of capillary columns for the analysis of complex environmental samples has become increasingly popular. Electrolytic conductivity detectors are not well suited for use with high-resolution capillary columns. Flame photometric detectors have the advantage of extremely fast response times and are ideal for use with high-resolution and high-speed capillary systems. The indium-sensitized flame photometric detector for halogen detection has been the subject of several investigations for use in both gas (1-7) and liquid (8) chromatography. I t has been shown to have good sensitivity, selectivity, and linear range. Previous studies have utilized variations of a two-flame
van der Smissen type burner with metallic indium suspended within a wire mesh between the flames. The sample is decomposed in the lower flame, converted in the region between the flames to the indium halide, and excited in the upper flame to emit at a wavelength that is characteristic of the indium halide. This work describes the use of various catalytic materials for the construction of the flame tips which improve the detection limits for chlorine by an order of magnitude over what has been previously reported. EXPERIMENTAL S E C T I O N The gas chromatograph used for this study was a Varian 6000 equipped with a 1075 capillary injector and a 1080 injector for use with packed columns. A Varian 401 chromatography data system was used to store and process the raw data. An ISA H-20 monochromator was used to scan the spectral regions of interest. The grating was blazed at 200 nm and had a 4 nm/mm dispersion; 50 wm front and back slits were used. The phototube was an RCA 1P28 and all lenses and windows were quartz. All measurements of sensitivity for chlorine were done with a 30 nm band-pass Hoya glass filter centered at 360 nm in place of the monochromator. An electrometer time constant of 1.2 was used for all packed column measurements; 200 ms was used in conjuction with capillary columns. The detector tower and burner are shown in Figure 1. The burner is a two-flame design with an inverted hydrogen-air lower flame similar to one described by Patterson (9). The indium reservoir consists of powered indium dispersed within a porous ceramic and cast into a cylindrical shape. The indium vapor is transported to the lower flame by diffusion from the heated reservoir. The temperature of the reservoir is controlled by the electrically heated detector oven of the gas choromatograph. The tower is designed to minimize the heat transport between the base of the tower and the top. This allows the base (and reservoir) to vary from 300 "C to 420 O C while the top varies from 120 "c to 145 "C. The temperature gradient is desirable so as t o allow high temperatures at the reservoir to produce large vapor pressures
0003-2700/83/0355-2112$01.50/0 0 1983 Amerlcan Chemical Soclety