Anal. Chem. 1998, 70, 3629-3635
Electrically Polarized Ion-Exchange Beds in Ion Chromatography: Eluent Generation and Recycling Hamish Small*
HSR, 4176 Oxford Drive, Leland Michigan 49654 Yan Liu and Nebojsa Avdalovic
Dionex Corporation, 445 Lakeside Drive, Sunnyvale, California 94088
This paper describes how electrically polarized ionexchange beds pumped with water can produce electrolyte of steady and controllable concentration. Such devices make it possible to use water as the pumped phase in ion chromatography (IC), thus avoiding off-line eluent preparation. Control of the electrical current flowing through the devices allows precise control of the concentration of eluent that they deliver. This provides a new way of performing gradient and isocratic elutions. Using water as the carrier and two small beds of resin, one as a generator the other as a suppressor, and periodically reversing their roles through automatically switched valves, we have developed a form of continuous IC that involves little intervention by the user. The paper presents the principles of the new method and examples of its use in anion analysis. All forms of liquid chromatography use eluents that the user must prepare and replenish. Because the preparation of eluents in the conventional way is time-consuming and often tedious, and precludes using chromatography in unattended or remote settings, more automated methods of eluent preparation are desirable. The eluents of some widely practiced forms of ion chromatography (IC) are aqueous solutions of common acids and bases. These simple electrolytes facilitate the automation of eluent preparation for IC, and a number of successful systems have been described. Dasgupta and co-workers have pioneered the application of electrolytic methods to purify eluents for IC and to electrically control their concentration.1,2 More automatic, less user-involved techniques have also changed suppression in IC so that this once bothersome procedure is now virtually invisible to the user.3-5 And recently, using a new concept called ion reflux, eluent generation and suppression have been integrated into an automated and, in principle, perpetual form of IC.6 This paper (1) Strong, D. L.; Dasgupta, P. K. J. Membr. Sci. 1991, 57, 321-336. (2) Strong, D. L.; Dasgupta, P. K.; Friedman, K.; Stillian, J. R. Anal. Chem. 1991, 63, 480-486. (3) Strong, D. L.; Dasgupta, P. K. Anal. Chem. 1989, 61, 939-945. (4) Rabin, S. A.; Stillian, J. R.; Barreto, V.; Friedman, K.; Toofan, M. J. Chromatogr. 1993, 640, 97-109. (5) Stillian, J. R.; Barreto, V. M.; Friedman, K. A.; Rabin, S. A.; Toofan, M. U.S. Patent 5, 248, 426, issued September 28, 1993. (6) Small, H.; Riviello, J. Anal. Chem. 1998, 70, 2205-2212. S0003-2700(98)00268-6 CCC: $15.00 Published on Web 07/11/1998
© 1998 American Chemical Society
describes another approach to automated IC that permits continuous, nearly unattended operation. Like ion reflux, it uses electrically polarized ion-exchange beds and water as the carrier. The electrolyte required for the ion-exchange separation is generated in situ, so off-line preparation of eluents is avoided. As in ion reflux, control of the current and water flow rate controls eluent concentration, so gradient elutions can be carried out with a simple isocratic pump. By using two small ion-exchange beds and flowswitching valves, this new method also integrates eluent preparation and suppression. We will describe the principles of these new techniques as they apply to anion analysis by IC; the principles for cation analysis follow by analogy. PRINCIPLES OF ELUENT GENERATION AND RECYCLING Eluent Generation. Ion-exchange resins, because of their mobile counterions, are good conductors of electrical current.7,8 A typical strong cation-exchange resin (Dowex 50 × 8) in the sodium form is as conducting as 0.23 M NaCl; in the hydrogen form it is as conducting as 0.55 M HCl.7 A packed bed of ionexchange beads is thus a complex network of electrical pathways involving the resin beads and their regions of contact. Consider a bed of typical, strong acid-type cation-exchange resin beads retained between porous metal electrodes and pumped with water (Figure 1). If a dc potential is applied to the bed, the electrical field will induce the positive counterions to move toward the cathode and a current will be sustained if there are means for replenishing the counterions leaving the anode and for disposing of those arriving at the cathode. The reactions of water at the electrodes provide the means. The anodic oxidation of water produces hydronium ions and oxygen gas
H2O f 2H+ + 1/2O2 (gas) + 2e
The hydronium ions electromigrate into the resin, displacing other hydronium ions and, in turn, potassium ions toward the cathode.9 (7) Small, H. Some Electrochemical Properties of an Ion Exchanger. M.Sc. Thesis, Queen’s University of Belfast, Belfast, 1953. (8) Sauer, M. C.; Southwick, P. F.; Spiegler, K. S.; Wyllie, M. R. J. Ind. Eng. Chem. 1955, 47, 2187-2193.
Analytical Chemistry, Vol. 70, No. 17, September 1, 1998 3629
Figure 1. An electrically polarized cation-exchange resin bed as a generator of potassium hydroxide. (A) is a conventional way of illustrating a packed bed of cation-exchange resin particles: upper layer in the potassium form and lower layer in the hydronium form. We have found the more abstract representation of (B) to be a better means of illustrating the various ion fluxes and fluid flows. The split rectangle in the left side of the column depicts the resin phase in its two-layer form, while the rectangle in the right side of the column represents the contiguous mobile phase. The porous metal electrodes are represented by the crosshatched regions (in A and B) and by the horizontal broken lines in (C). (C) depicts the directions of cation migration and water flow through the bed and the formation of KOH at the cathode.
The cathode reaction produces hydrogen gas and an equivalent number of hydroxide ions as companions for the potassium ions arriving at the cathode -
depends on the size of the bed and how “hard” the generator is driven, that is, the concentration of base that it is called upon to produce. It also depends on how efficiently the resin counterion is utilized. Figure 1 could imply that the hydronium-potassium boundary is sharp and remains so throughout the depletion process, which, in turn, would imply a completely efficient use of the potassium to provide eluent of requisite, steady concentration. But that is not the case; the boundary does not remain sharp and hydronium can break through to the cathode before all of the potassium is displaced. When this occurs, part of the current is spent in producing water at the cathode at the expense of potassium hydroxide and the critical link between current and eluent concentration is broken. At this point, the generator must be replaced or regenerated. If one uses generator beds of typical IC dimensions and common elution conditions, this point can be reached in a matter of just hours. Because generators with such short lifetimes are an unattractive alternative to the conventional means of preparing eluents, we have used two different strategies to prolong their lifetimes: (1) much larger generator devices with lifetimes of weeks or months and (2) methods wherein the eluent is captured and recycled. The large generator approach will be the subject of other papers. Eluent recycling has been accomplished in two ways. The first method, ion reflux, uses water as the carrier in conjunction with an electrically polarized ion-exchange bed and provides virtually perpetual eluent generation and suppression.6 The second method exploits the dual character of a partially spent generator bed (Figure 1). Not only does this bed have residual generator capacity but it has acquired capacity to suppress potassium hydroxide thus,
KOH + H+resin f K+resin + H2O
H2O + 2e f 2OH + H2 (gas)
If water is pumped through the bed in the anode-to-cathode direction while potential is applied, a solution of potassium hydroxide is carried from the bed for as long as some potassium ions remain in the resin. The directions of ion migration, flow of water, and electrode products are shown in Figure 1C. Now if the flow rate of water and the current through the bed are held steady, the device should deliver a steady concentration of potassium hydroxide suitable as an eluent for the isocratic elution of anions in an IC application. Furthermore, changing the current should lead to eluent concentration changes that are predictable by the laws of electrolysis. It also follows that programming the current, while maintaining a constant flow of water, affords a means of providing steps or gradients in the eluent concentration. In other words, by using electrical current to control eluent concentration, a simple isocratic chromatographic pump can take the place of a more complex gradient pump. Clearly there is some limit to the period that the ion-exchange bed can deliver base, for it will eventually become depleted in the counterion that forms the base. How quickly this happens (9) Because resins are conducting current, the best choice of resin is usually the resin with the best conductivity, providing no other factor overrides that choice. Potassium form resin is approximately twice as good a conductor as the sodium form.
3630 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
from the hydronium ion that it has accumulated. Early versions of suppressed IC used such packed-bed suppressors. They had some drawbacks that could be attributed to their size,10 but these problems can be minimized by using very small suppressor beds that are frequently regeneratedsas often as once every chromatographic run.11 This frequent regenerations by chemical means in this casesis conveniently handled by automatically actuated valves. Furthermore, by switching between two identical small beds, the suppressing function is, from a practical viewpoint, continuous. The recycling method of this paper has a similar strategy, except that electrochemical regeneration takes the place of chemical regeneration; it differs from the chemical regeneration method in requiring no chemical other than water. Figure 2 is a simplified schematic of eluent recycling. Two key elements in the system are the small cation-exchange resin beds, one preceding the anion separator column, the other following it. They are of identical size, and each is equipped with a pair of porous metal electrodes. At start-up, each bed is roughly half in the hydronium form and half in the potassium form. The upper bed is electrically polarized and takes on the role of eluent (10) Small, H. Ion Chromatography; Plenum Press: New York, 1989; p 170. (11) Small, H.; Riviello, J.; Pohl, C. A. U.S. Patent, 5, 597, 734, issued January 28, 1997.
regeneration of these two beds. Injecting analytes as their acids avoids this. Analogous arguments apply to cation analysis.
Figure 2. The basis of eluent recycling. Note: The chromatographic components are not drawn at the same scale.
generator, while the lower bed is maintained in an electrically passive state and acts as a suppressor for the KOH flowing from the separator column. During a chromatographic run, the upper bed (the generator) becomes partially depleted in potassium while at the same time it acquires an equivalent amount of hydronium ion. Similarly, the lower bed becomes equally depleted in suppressor capacity but picks up an equivalent amount of potassium. At the end of a single chromatographic runsor several runs if conditions allowsthe roles and positions of the two beds are reversed. What was previously the generator is placed after the separator and becomes the suppressor, and the suppressor is moved ahead of the separator and becomes the generator. In addition, the orientation of each bed with respect to the direction of water flow is reversed, so that resin of the appropriate ionic form is presented to the water flowing from the pump and to the eluent flowing from the separator. This is accomplished by automatic valves which we will illustrate later. If the chromatography uses an appropriately low concentration of eluent, the switching of roles can wait until several samples have been analyzed. The beds have an ion-exchange capacity such that the degree of depletion between switches leaves their ionic composition comfortably short of that point where the proper functioning of either bed is compromised. An earlier paper6 showed how sample preconditioning is critical to the optimal operation of ion reflux-type devices; the same applies to the eluent recycling technique. In anion analysis, salts in the samples should be converted to their acids before injection, otherwise the cations from the samples will eventually consume the entire suppressor capacity of columns A and B. At that point, chromatography must be interrupted to carry out a corrective
EXPERIMENTAL SECTION Ion-Exchange Resins. To generate and suppress bases for anion analysis we used either commercially available Dowex 50WX8 200-400 mesh or proprietary cation-exchange resins of similar chemical composition and particle size (Dionex Corp., Sunnyvale CA). Resins were converted to the appropriate ionic forms by common ion-exchange procedures using reagent grade chemicals as regenerants. The separator columns were standard commercial varieties (Dionex Corp.). CAUTION: Anion-exchange resins are used to generate and suppress acids for cation analysis. We know that electrically polarized ion-exchange beds can become hot, particularly when current is high and water flow rate is low. We are also aware that anion-exchange resins in the nitrate form have been known to decompose explosively when heated. Therefore, we urge those who experiment with electrically polarized beds to use extreme caution when working with anion-exchange resins with oxidizing counterions such as nitrate or perchlorate. Metal Electrodes. The metal electrodes were customfabricated porous platinum disks (Mott Metallurgical, Farmington, CT) with diameters varying from 4.1 to 4.2 mm. The disks were approximately 1.45 mm thick and had a 2-µm nominal porosity. Power Supply. An electrophoresis power supply (EPS 600, Pharmacia Biotech, Uppsala, Sweden) was used in most of the experiments. This power supply has the ability to deliver constant current under conditions of varying electrical resistance, an important requirement in this work. Managing the Electrolysis Gases. During electrolysis, the gases produced in the eluent generator can be a problem for the chromatographic analysis; we solved this by adding a flow restrictor to the outlet of the chromatographic system. The restrictor raises the pressure in the system and compresses the gases to an insignificant volume. Two meters of plastic capillary tubing, 0.005-in. i.d., placed at the end of the system, raises the operating pressure in the conductivity cell from 200 to 500 psi at normal flow rates. This is usually adequate for eliminating any problems arising from gas bubbles in the IC system, particularly in the conductivity cell. Miscellaneous Hardware. Chromatographic pumps, valves, conductivity cells and meters, and chromatographic columns were available from Dionex Corp. RESULTS AND DISCUSSION Concentration of the Eluent and the Current in a Generator. To determine the relationship between current applied to a generator and the strength of eluent it delivers, a 3-mm-i.d. × 175-mm-long column was filled with Dowex 50W X 8 in the sodium form and each of its ends was equipped with a porous Pt disk. Deionized water was passed through the column at 2 mL/min while the bed was electrically polarized to give various levels of steady current. A conductivity cell placed at the outlet of the generator was used to determine the concentration of eluent being produced at the various current levels. The results, shown in Figure 3, illustrate the very good linearity between current and the specific conductance of the sodium hydroxide produced. Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
3631
Table 1. Capacity Factors for Several Anions versus Current Applied to the Potassium Hydroxide Generator ion
current (mA)
k′
ion
current (mA)
k′
fluoride chloride nitrate sulfate
10 10 10 10
0.055 0.41 1.47 4.22
fluoride chloride nitrate sulfate
5 5 5 5
0.11 0.85 2.91 16.4
Figure 3. Specific conductance of the effluent of a sodium hydroxide generator as a function of the current that it carries.
Knowing the characteristics of the conductivity cell, one can calculate the concentration of sodium hydroxide generated. And knowing the current and the flow rate, and applying the simple laws of electrolysis, one can calculate the concentration of sodium hydroxide to be expected. In one experiment where the current applied was 11.9 mA, the specific conductance of the sodium hydroxide was 859 µS, corresponding to a NaOH concentration of 0.003 45 M. From the current and flow rate, we calculated an expected concentration of 0.0037 M.12 The agreement is good. In another similar experiment using a 4-mm-i.d. × 150-mmlong bed of potassium form resin,9 the concentration of base was measured more directly by collecting the effluent and titrating it with standardized acid. At a water flow rate of 1 mL/min and a current of 10 mA, the effluent was 0.0051 M potassium hydroxide. The current through the bed was then reduced to 5 mA, and the effluent was found to be 0.0025 M. This experiment confirmed the linear dependence of effluent concentration on current and illustrates the ability to precisely control the concentration of eluent produced by controlling the current through the resin bed. A Generator Used in a Chromatographic System. The KOH generator was used in a chromatographic system that included two Dionex Ion Pac AG 11 columns connected in series as the separator; a small, chemically regenerated packed-bed suppressor (Dowex 50W X 8, 200-400 mesh, H+ form, 4 mm i.d. × 50 mm long); a conductivity cell; and associated electronics. We determined the elution behavior of several common anions at various current levels in the generator. The results for currents of 5 and 10 mA are summarized in Table 1. Figure 4 shows chromatographic separations carried out at two different generator currents and illustrates how conveniently chromatographic behavior can be controlled simply by manipulating the current. Figure 5 shows the elution of three “low-affinity” anions: fluoride, acetate, and formate. This particular experiment is remarkable in that (1) the voltage applied to the generator was only 6V, (2) the current was only 0.2 mA, and (3) the concentration of KOH produced by the generator was only 0.000 12 M (calcu(12) A current of 11.9 mA is equivalent to 0.714 C/min, that is, (0.714 × 1000)/ 96500 (Faraday constant) ) 0.0074 mequiv of NaOH generated per minute. At a flow rate of 2 mL/min of water, the generator will therefore deliver NaOH at a concentration of 0.0037 M.
3632 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
Figure 4. Separation using a potassium hydroxide generator: separator, two 4-mm-i.d. × 50-mm-long Dionex AG-11 columns, connected in series; sample size, 10 µL; analytes (in order of elution) fluoride, chloride, and nitrate, 0.0001 M in each analyte; flow rate of water, 1 mL/min; current, 5 (A) and 10 mA (B).
lated). In the conventional way of preparing eluents by diluting a concentrated solution, it is a formidable challenge to keep basic solutions entirely free of contamination by carbon dioxide. When the concentration of base is very low, as in this experiment, the presence of even small amounts of carbonate contaminant can lead to widely varying elution behavior. In contrast, by electrochemically preparing eluent using a base generator and pure water as the carrier, carbon dioxide contamination is prevented and chromatographic behavior is very stable. This is another of the many advantages of preparing eluents in this way. Separation Columns for Hydroxide Eluents. Historically, most anion separation columns used carbonate-based eluents. Even at the low-millimolar concentration range, carbonate-based eluents can efficiently elute polyvalent ions such as sulfate and phosphate. This is not the case for hydroxide eluents; therefore, to realize the full potential of hydroxide eluents, hydroxideselective separation columns will be required. Fortunately, such columns are being developed because ion chromatographers have
Figure 5. Separation of “low-affinity” anions using a potassium hydroxide generator: separator, two 4 × 50-mm Dionex AG-11 columns, connected in series; sample size, 10 µL; analytes (in order of elution) fluoride, acetate, and formate, 0.2 ppm fluoride, 1 ppm in each of acetate and formate; flow rate of water, 1 mL/min; current, 0.2 mA; concentration of KOH, 0.000 12 M (calculated).
recognized the advantages of hydroxide-based eluents:13 low background conductivity after suppression, minimal water dip, and improved resolving power for weakly held species. Exhaustion of the Generator. When called upon to deliver more concentrated eluents at flow rates typical for IC, small generators can exhaust within impractically short periods. Furthermore, the convenient link between current and concentration can break down before all of the ion in the generator has been dispensed. We carried out a number of experiments to define this behavior for a typical base generator. A 4-mm i.d. × 77-mm-long column was packed with an 18-µm sulfonated styrene-divinylbenzene resin in the potassium form and pumped with water at a flow rate of 1 mL/min while carrying a current of 15 mA. The concentration of KOH in the generator effluent was monitored by measuring the conductance of the effluent with a conductivity meter. The generator produced an essentially constant level of KOH (0.0093 M) for the first 72 min after which time the concentration dropped steadily; at 130 min, the concentration had dropped to about 0.005 M. Clearly, the generator had ceased to give acceptable behavior at 72 min, where we calculated that only 36% of the potassium had been used. This underutilization of the potassium ion is caused by ion boundary instabilities and is aggravated when the displacing ion is more mobile than the ion being displaced. Consider a hypothetical, perfectly flat boundary between the hydronium and potassium resin layers. Now imagine that some unstabilizing influence causes an embryonic “plume” of hydronium ions to advance into the potassium layer. This slight penetration by hydronium represents a less resistive pathway for the current since the hydronium form resin is more conducting than the potassium form. This, in turn, will cause a proportionally larger (13) Jackson, P. E.; Pohl, C. A. TrAC, Trends Anal. Chem. 1997, 16, 393-401.
Figure 6. Valving arrangement for eluent recycling. Valve system for switching resin beds A and B between the roles of generator and suppressor. The power supply is not shown, just the polarizing potential on each bed in each mode.
current to flow through this “bulge” in the boundary, and the more the current flows the less resistive the pathway becomes and the greater the penetration by the hydronium ions. This selfpromoting plume of hydronium will, therefore, reach the cathode before the bulk of the boundary does. When this happens, part of the current produces water rather than KOH and the linear relationship between current and eluent concentration ceases. When the displacing ion is slower than the displaced species, similar arguments will show that the boundary is self-sharpening. To counteract this premature exhaustion of generators we have employed a dual-bed generator. A dual-bed device was constructed by filling a 4-mm-i.d. × 77-mm-long column with a 63-mm upstream section of 18-µm sulfonated styrene-divinylbenzene resin in the potassium form. The 14-mm downstream section was filled with a 10-µm macroporous styrene-divinylbenzene resin with surfacegrafted R-chloroacrylic acid functional groups. This weak acidtype resin was also in the potassium form. The anode was at the inlet of the strongly acid resin and the cathode at the outlet of the weakly acid resin. When this dual-bed device was operated at 15 mA and 1 mL/min flow rate, it produced a steady Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
3633
Figure 7. Constant-current separation using eluent recycling. This shows the chromatograms obtained in each mode after the 67th injection: separator, 4 × 250-mm, Dionex AS-11 column; analytes (1) fluoride (2 mg/L), (2) chloride (3 mg/L), (3) sulfate (15 mg/L), (4) nitrate (10 mg/L), and (5) phosphate (15 mg/L); flow rate of water; 0.5 mL/min; current, 18 mA (voltage 30 V); concentration of KOH, 0.022 M (calculated). The small peaks at the beginning and the end of the chromatograms in Figures 7 and 8 were artifacts associated with the valve switching at the end of each run.
concentration of KOH at the same concentration as the single resin column, 0.0093 M, but continued to do so for a period of 96 min. We attribute this better utilization of potassium to the arresting effect of the carboxylic acid resin. If a hydronium plume should reach the carboxylic acid resin layer, its progress will be retarded since the conductivity of the hydronium form of a carboxylic acid resin is much lower than the conductivity of its potassium form. This slowing down of hydronium plumes will allow the rest of the boundary to catch up and delay hydronium reaching the electrode. Although this strategy gives a marked improvement in generator lifetime, it is not enough to turn a small generator into a practical one; the exhaustion times are still too short to make this a viable approach.14 Building much larger generators is one solution to this problem; eluent recycling is another. Eluent Recycling. Figure 6 is a schematic of an experimental system that was assembled to implement the eluent recycle concept illustrated in Figure 2. Bed A was a 4-mm-i.d. × 50-mm(14) However, used with capillary separators, generators of these capacities will have much longer lifetimes because the eluent flow rates are much lower. So generators of these dimensions may be very appropriate for capillary systems.
3634 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
Figure 8. Separation using eluent recycling and a current gradient. Separator, sample, and flow rate same as in Figure 7. Current was changed from 0.5 to 34 mA at a rate of 1 mA/min. The chromatograms were obtained by running the separation in each mode. Analytes in same order of elution as in Figure 7. The peak between the sulfate peak and the nitrate peak came from dissolved carbon dioxide (or carbonate) in the sample.
long electrode-equipped column filled with an 18-µm sulfonated styrene-divinylbenzene resin initially in the hydronium form, while bed B was an identical column filled with the same resin, initially in the potassium form. The separator was a Dionex AS11 anion separator (4 mm i.d. × 250 mm long). The various columns were connected to the pump and detection system through three Dionex BF-2 double-stack, four-way, two-position valves as shown in Figure 6. A computerized controller automatically switched the four-way valves between their two positions. Computer relays connected a power supply to whichever of the two columns was the generator. In the mode I valve position, the column in position A was supplied with current to produce KOH while the column in position B was used as the suppressor column. The system was operated in this position until the bed in position A was about 50% converted to the hydronium form by hydronium ions generated at the anode. At this point, the bed in position B had been converted to about 50% in the potassium form by the KOH released from the bed in position A. Both beds had now acquired the ability to generate and suppress KOH, so by continuously cycling between the two modes, KOH was continuously generated and suppressed. It took about eight cycles for the system to
stabilize to the point where consecutive chromatograms were essentially identical. In one experiment, the KOH generator was supplied with a constant current of 18 mA (the applied voltage was 30 V) and the water flow rate maintained at 0.5 mL/min. Under these conditions, the KOH concentration was 0.022 M. Figure 7 shows an example of the separation of several anions obtained after the system had been switched automatically between the two valve positions for more than 60 sample injections. These results demonstrated the successful application of eluent recycle to the isocratic separation of anions. In another experiment, the current supplied to the KOH generator was changed from 0.5 to 34 mA at a rate of 1 mA/min to generate a gradient of KOH from 0.0062 to 0.042 M at a flow rate of water of 0.5 mL/min. Figure 8 shows an example of the separation of several anions obtained with the system cycled between the two modes. This is a successful demonstration of gradient elution and cosuppression by the recycling technique. CONCLUSIONS We have demonstrated that electrically polarized ion-exchange beds pumped with water can be used to generate eluents for ion
chromatography. The eluent concentration can be precisely controlled simply by controlling the current and flow rate through the bed, a great convenience for both isocratic and gradient elutions. We have successfully demonstrated how switching the roles of two small ion-exchange beds solves the problem of premature exhaustion of a generator, and though it is not as simple as the ion reflux method, it provides yet another version of perpetual ion chromatography using simply water as carrier. We anticipate that there will be ample opportunities to apply these recycling techniques to other forms of chromatography as well as IC. ACKNOWLEDGMENT H.S. thanks the Dionex Corp. for their support and hospitality while he worked at the Dionex research laboratories in Sunnyvale. Note. The methods and apparatus described in this publication are the subject of pending patents.
Received for review March 10, 1998. Accepted June 11, 1998. AC980268X
Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
3635
