Comparison of pulsed coulometric detection and potential-sweep

Periodically interrupted amperometry. A way of improving analytical performance of membrane coated electrodes. Livia Nagy , Nikoletta Kálmán , Géza...
0 downloads 0 Views 599KB Size
Anal. Chem. 1989,6 1 , 555-559

555

Comparison of Pulsed Coulometric Detection and Potential-Sweep Pulsed Coulometric Detection for Underivatized Amino Acids in Liquid Chromatography Lawrence E. Welch,' William R. Lacourse, David A. Mead, Jr.,2a n d Dennis C. Johnson*

Department of Chemistry, Iowa State University, Ames, Iowa 50011 Terry Hu

Cypress Systems, Lawrence, Kansas 66046

Pulsed coulometric detectlon (PCD) and potential-sweep pulsed coulometric detection (PS-PCD) are applied to the direct detection of amino acids in protein hydrolyrates. The detection mechanisms are based upon surface-catalyzed oxidation of the amine functionailties activated by the transient formation of surface oxide on Au electrodes. PCD uses a triple-step potential waveform In whlch the integration of electrode current at a constant potential Is followed by anodic and cathodic polarizations to clean and reactivate the eiectrode surface. PS-PCD incorporates a cyclic potential sweep into the triple-step waveform which proceeds through the formation and subsequent removal of the surface oxide with simultaneous current integration. Reactions catalyzed by the formation of the noble metal oxide can be monitored in PSPCD with the automatic reJectlonof the surface oxide background. A significant decrease is obtained also in the fluctuation and drift in the base line resulting from gradient elution and variations of the electrode surface. PCD and PS-PCD following gradient elution chromatography are demonstrated to allow for the direct detectlon of 20 amino acids including secondary amino acids. Detection limits for lyslne are ca. 220 ppb (1 1 ng, 75 pmoi) by PCD and 60 ppb (3 ng, 19 pmoi) by PS-PCD applied at gold electrodes.

INTRODUCTION With the burgeoning interest in genetic engineering and biotechnology, the determination of amino acids for protein sequencing and analysis has become increasingly important. Recent progress in amino acid determinations can be attributed to technological advances in liquid chromatography and chromatographic detectors ( I ) . Separations of amino acids and their derivatives in liquid chromatography (LC) are readily achieved by using reversed-phase stationary phases (2-5) and ion exchangers (6-8). For the separation of complex mixtures, gradient elution chromatography is essential. The common amino acids do not fluoresce, and only tyrosine, phenylalanine, and tryptophan have significant UV-vis absorbance. Sensitive photometric detection requires some form of pre- or postcolumn derivatization. o-Phthalaldehyde (OPA), a fluorescent derivatizing agent, offers excellent sensitivity and subpicomole detection limits (4, 9). But the derivatives of OPA are not stable, and OPA does not form adducts with secondary amino acids (e.g. proline and hydroxyproline). Ninhydrin (10) and phenyl isothiocyanate (11-13) Permanent address: Department of Chemistry, Knox College, Galesburg, IL 61401. Permanent address: Commonwealth Edison, Maywood, IL 60153.

are used to enhance UV-vis detection. However, sensitivity for these reagents is less than that for OPA. Detection methodology that does not require derivatization is preferred, when available, for convenience and simplicity. Laser-based optical activity (14) and refractive index have been applied to the detection of underivatized amino acids, but limits of detection and reproducibility are usually poor. In recent years, electrochemical detection with liquid chromatography has gained prominence as a sensitive and selective detection technique for electroactive groups. Amino acids generally have not been considered to be electroactive (15-1 7). Direct anodic detection at constant applied potential (dc) can occur by catalytic mechanisms on certain transition-metal oxides, e.g. nickel and copper ( 1 4 1 9 ) . However, catalytic dc detections on noble-metal electrodes is accompanied by loss of electrode activity with rapid decay of analytical sensitivity. Pulsed amperometric detection (PAD) and pulsed coulometric detection (PCD) following liquid chromatography have proven to be selective and sensitive techniques for the determination of alcohols, polyalcohols, carbohydrates (20-23), amino acids ( 8 ) ,aminoalkanols (24),and many inorganic and organic sulfur-containing compounds (24). PAD and PCD use a triple-step potential waveform to combine amperometric (PAD) and coulometric (PCD) detection followed by alternating anodic and cathodic polarizations to clean and reactivate the electrode surface. In the detection of amino acids, the waveform exploits the surface-catalyzed oxidation of the amine functionality activated by the formation of noble metal surface oxides. Pulsed coulometric detection does not require elevated temperatures, complex nor timed injection systems, nor pre- or postcolumn derivatizations. Sensitivity in PCD is optimum at ca. pH >11, and postcolumn addition of base may be desired. However, the catalytic nature of PCD for amino acids limits the use of gradient elution chromatography because the base-line signal corresponds to the oxide formation process which is very sensitive to small changes in the mobile phase composition, especially the pH. Potential-sweep pulsed coulometric detection (PS-PCD) incorporates a cyclic potential sweep within the detection period of the triple-step waveform which proceeds through the formation and removal of the surface oxide (25). Accordingly, reactions catalyzed by the simultaneous formation of the noble metal oxide can be coulometrically monitored with the automatic rejection of the surface oxide background signal. PS-PCD can substantially decrease base-line drift associated with variations in pH and composition of the mobile phase, and the slowly changing surface area of the noble-metal electrode. Thus, PS-PCD improves considerably the base-line characteristics in gradient elution chromatography. This paper compares the application of PCD and PS-PCD following gradient elution chromatography to the direct de-

0 1989 American Chemical Society 0003-2700/89/0361-0555$01.50/0

556

ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989

tection of amino acids in protein hydrolyzates. Previous work concerning application of PAD to amino acids utilized a platinum electrode (8, 26). More recent research indicates that gold may be a more universal noble electrode material for oxide-catalyzed detections activated by the waveforms of PCD and PS-PCD. EXPERIMENTAL SECTION Apparatus and Procedures. Voltammetric data were obtained at a Au rotated disk electrode (RDE) (Pine Instrument Co., Grove City, PA). Current decay was studied with a Model 2230 digital storage oscilloscope (Tektronix Corp., Beaverton, OR). All voltammetric experiments used rotation speeds of 900 revolutions m i d and scan rates of 6.0 V min-' unless noted otherwise. Liquid chromatographic work employed either an isocratic or gradient chromatography system (Dionex Corp., Sunnyvale,CA). For gradient chromatography, a gradient mixer was placed after the pump outlet to enhance mobile phase mixing. The injection volume was 50 pL. Separations were performed with an AS-8 anion-exchange analytical column with an AG-6 guard column (Dionex). All mobile phases were filtered before use with 0.45-fim Nylon-66 filters (Rainin Corp., Woburn, MA) and a solvent filtration kit (Millipore Corp., Milford, MA). All flow rates were 1.0 mL min-'. PCD was performed with the Model PAD-2 electrochemical detector (Dionex), and PS-PCD was performed by softwaregenerated waveforms from a modified computer aided electroanalysis system (Cypress Systems, Lawrence, KS). The Cypress Systems potentiostat was interfaced via a 12-bit analog to digital (A/D) converter to a System 1800 IBM-AT compatible personal computer (Everex Systems, Inc., Fremont, CA) with a 20-MB hard disk drive, an EGA color monitor, and a Model 7440A ColorPro plotter (Hewlett-Packard, San Diego, CA). Two electrochemical cell configurations were used interchangeably. A wire electrode configuration consisted of a Au working electrode, Pt counter electrode, and a saturated calomel reference electrode (SCE). A thin-layer electrochemical cell (Dionex) was modified to incorporate a Universal Glass pH Electrode (Fisher Scientific Co., Springfield, NJ) or a SCE as the reference, and the working electrode and counter electrode materials were Au and glassy carbon, respectively. Reagents. All solutions were prepared from reagent grade chemicals. Amino acids standards were from Aldrich Co. (Milwaukee, WI), Fisher Scientific Co. (Springfield, NJ), and Pierce (Rockford, IL). Protein hydrolyzates were from Pierce, and inM (25 nmol), dividual amino acid concentrations were 5 X except cystine, which was 2.5 X lo4 M (12.5 nmole). Water was purified in either a Millipore MILLI-Q system or a Barnstead NANOpure I1 system, followed by filtration (0.2 pm). R E S U L T S A N D DISCUSSION Voltammetry. Recent experimental results indicate that Au rather than Pt is the electrode material of choice for pulsed coulometric detection of carbohydrates (21-23, 27) and the detection of many inorganic and organic sulfur-containing compounds (24). In addition, overall electrochemical characteristics of Au electrodes are well-suited to pulsed coulometric detection techniques, vide infra. Thus, it was desired to study the use of Au working electrodes for the detection of amino acids by the pulsed coulometric detection techniques. The current-potential (I-E) response is shown in Figure 1 for lysine at a Au RDE in 0.05 M NaOH. An anodic wave is observed for the positive potential scan in the region ca. 0.1-0.7 V vs SCE corresponding to the anodic formation of the electrocatalytic surface oxide on the Au surface. On the reverse scan, the growth of surface oxide ceases and the oxide is cathodically dissolved in the region +0.2 to -0.2 V. Additions of lysine enhance the anodic current for the positive potential scan, and the sensitivity for lysine at Au electrodes is highest in alkaline media (pH ca. >11). It is highly significant that lysine is detected only during the positive potential scan when surface oxide is being generated. No signal is observed in Figure 1 for the negative scan when oxide growth

+

Iz ("..AI

Figure 1. Voltammetric response (I-€) for lysine at a Au rotating disk electrode: lysine concentration (mM) (a) 0, (b) 0.065, (c)0.12, (d) 0.26.

has ceased. Hence, it is concluded that dc detection for a constant potential in the range 0.1-0.7 V is not expected to produce useful analytical signals. This conclusion is consistent with the general notion that amino acids are not electroactive. The net anodic current for lysine increases markedly with increases in the potential scan rate, yet little change results from variations in electrode rotation speed. This behavior is indicative of processes in which the reaction rate is under the control of electrode surface processes (8). These results are further evidence for an anodic detection mechanism that is electrocatalytic, requiring simultaneous formation of surface oxide. It is the amino acid which has been adsorbed at E < 0.1 V which is detected during the positive scan a t E > 0.1 V. Therefore, maximum anodic response is expected to be limited by the adsorption isotherm for the particular amino acid being detected. Because the fully developed oxide-covered surface exhibits insignificant catalytic reactivity, detection of lysine at Au electrodes is only possible via a transient electrocatalytic mechanism achieved within a pulsed potential waveform. All other essential amino acids display nearly voltammetric behavior identical with that of lysine. All voltammetry experiments were performed with degassed solutions to emphasize amino acid detection. Under nondegassed conditions, the dissolved oxygen reduction wave is well-resolved from the cathodic stripping peak of gold oxide. This aspect of Au electrodes is in direct contrast to Pt electrodes where the dissolved oxygen reduction wave overlaps the cathodic stripping peak of Pt oxide. Thus, Au electrodes allow one to perform pulsed coulometric detection techniques without precise control of dissolved oxygen concentration or degassing the entire LC system, both of which are difficult if not impossible to achieve. For these reasons, analyte response at Au electrodes under nondegassed conditions show better reproducibility and repeatability; even though the limit of detection for lysine at Pt electrodes was determined to be 3 times lower under degassed conditions. Liquid Chromatography-Pulsed Coulometric Detection. Table I describes the optimized triple-step waveform for PCD of lysine. Figure 2 illustrates the transient amperometric signal (I-t) obtained upon stepping to the detection potential (E,) for the Au RDE. For application of PCD for liquid chromatography, the base-line response corresponds to the value of the time integral of the electrode current given in curve a for an integration period of 16.7 ms beginning a t time t d in period t,. For lysine, the maximum signal-to-noise ratio (S/N) was obtained for t d = ca. 540 ms. Since the electrode activity is diminished due to inhibition by the fully developing surface oxide and, perhaps, by fouling of the surface by adsorbed oxidation products, electrode activity is renewed by subsequent anodic (E2)and cathodic (E3)polar-

ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989

Table I. Pulsed Coulometric Detection Waveform Characteristics t,

557

Table 11. Solvent System for Amino Acid Separation with the Dionex AS-8 Column solution 1 (regenerant) solution 2 solution 3

solution 4 flow rate

0.56 M NaOH/0.64 M boric acid 0.23 M NaOH/0.005 M Na2B407 0.08 M NaOHI0.018 M NazB407/2%MeOH 0.4 M NaOAc/0.001 M NaOH/2% MeOH 1.0 mL min-'

Gradient Program potential (mV vs SCE)

time, ms

E l , 500 Ez, 1050 Es, -550

t,, 560 t z , 180 t,, 240

t

sol 1, %

time, min

I\ i,

function detection oxidative cleaning reductive reactivation and adsorption of analyte

0.0 4.0

.-

Time-

-

Flgure 2. Amperometric response ( 1 4 ) for lysine following the poE , at a Au rotating disk electrode: lysine concentential step E , tration (mM) (a) 0, (b) 0.044,(c) 0.28, (d) 2.0.

B 0.511A

5 min

TimeFigure 3. Comparison of dc and pulsed detection: (A) PCD, see Table I; (6)dc detection at 0.50 V vs SCE.

izations. During the application of E$,adsorption of analyte can occur on the oxide-free surface prior to the next detection step. The optimized PCD waveform was applied initially to a 0.1 mM lysine solution in 50 mM NaOH mobile phase a t a Au electrode in a flow-through electrochemical cell under conditions of flow injection, i.e. no chromatographic column in line. One of the advantages of PCD is aptly demonstrated in Figure 3 where the anodic response is compared to that for a constant potential (dc). The dc response to lysine is 100 times smaller than PCD response and decays rapidly to a level of uselessness during successive sample injections. The long-term stability of the PCD response was confirmed by repeated injections of lysine solution over a 6-h period. The measured peak response showed only a 1.3% relative standard deviation over this time span. The PCD response for the direct detection of amino acids is optimum at pH ca. >11. The pK, of the amine group of glycine is 9.57 (28) and other amino acids have similar values. A t high pH values, the amine functionality of an amino acid

100 100 90

10

100

26.0

100 100 100

36.0

100

56.0

sol 4, %

100 100

14.0

25.8

sol 3, %

100

4.1 13.9

40.0 46.0 46.1

L

sol 2, 70

inject

100

is deprotonated and the amine group(s) can be oxidized. In addition, at high pH values, amino acids can be separated via anion-exchange chromatography. Solutions of NaOH as eluents provide the desired high pH for efficient chromatography and with sufficient ionic strength necessary for satisfactory electrochemical detection. Statistical analysis of calibration data was based on a modified regression analysis which assumed that variance in the signal is proportional to concentration (29). PCD response for lysine was linear (intercept, 1.77 pC; slope, 6.79 X lo4 pC M-l; and S , = 1.6 X pC) over nearly 2 decades with significant deviation from linearity for concentrations greater than 15 ppm (750 ng, 5.1 nmol). The limit of detection for lysine by PCD was 220 ppb (11ng, 75 pmol). Response for other primary amino acids was within f 3 times that for lysine. Detection limits for the secondary amino acids, hydroxyproline and proline, were 420 and 490 ppb, respectively. A major limitation toward a single injection determination of all the underivatized essential amino acids has been the inability to efficiently couple gradient chromatography with pulsed electrochemical detection. The AS-6 column used previously for carbohydrate separations lacks the ability to separate similar amino acids during isocratic elution. Application of a four-solvent gradient system (three for the gradient and one for regeneration, Table 11) with an AS-8 analytical column has been described for separation of underivatized amino acids without severe base-line shift using UV-vis detection following postcolumn addition of ninhydrin (30). This four-solvent gradient system is sufficiently alkaline for electrochemical oxidation of the amino acids without postcolumn pH adjustment. However, the solvent gradient is severe, with changes in pH, ionic strength, and an organic modifier. Comparison of the cyclic voltammograms for each of the three mobile phase solvents revealed that there is no potential value that can be selected for catalytic anodic detection which will give a constant base line. An attempt to use LC/PCD with the separation of a 17-component protein hydrolyzate resulted in such a severe base-line shift, as predicted from the cyclic voltammograms, so as to make the chromatograms virtually useless. The major causes for the shifting of the surface oxide background by the three-gradient mobile phase in LC/PCD are the change in pH and the addition of the first few percent of methanol to the aqueous system. The voltammetric waves for surface oxide formation and dissolution shift with pH (-59

558

ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989

I

1

Table 111. Waveform for Potential-Sweep Pulsed Coulometric Detection (PS-PCD)

!

-Id

\

I?

variable

waveform Aa

waveform Bb

E,, mV

0 600 0 800 0 350 50 5 1 344 50 10 100 90

0 600 0 800 -300 150 20 55 1 384 80 60 5 235

E2 E3

E4 E5 Ta,ms Tb

Tc Td

Te Tf Tg

Li\ i d Figure 4. Anion-exchange amino acid separation followed by PCD: (1) arginine, (2) lysine, (3) threonine, (4) alanine, (5) glycine, (6) serine, (7) valine, (8) proline, (9) isoleucine, (10) leucine, (1 1) methionine, (12) histidine, (13) phenylalanine, (14) glutamic acid, (15) aspartic acid, (16) cystine, (17) tyrosine.

mV/pH unit), but the reference potential obtained from an SCE reference electrode does not shift. Hence, the base-line response a t the detection potential (E,) will change with changes in pH. Mead (31) demonstrated that a glass pH electrode used instead of an SCE as the potential reference results in a shift of the reference potential in unison with the pH gradient. A comparison of the three-gradient solvents by cyclic voltammetry using a glass pH reference revealed that the anodic waves for oxide formation are nearly superimposed. Base-line shift attributable to the pH gradient is, therefore, substantially decreased by use of the pH reference. A chromatogram for LC/PCD using the pH reference is shown in Figure 4. The base-line shift in Figure 4 corresponds to the addition of the organic modifier in the gradient. Although the effects of a pH gradient can be compensated to a large extent by use of a glass pH reference electrode, the base line obtained by using the PCD waveform is highly sensitive to other factors that change the rate of surface oxide formation, i.e. organic modifier, ionic strength, temperature. This aspect of PCD virtually limits its use to isocratic chromatography. In addition, fluctuations (noise) in the large background current from the oxide formation process is the limiting factor in achieving lower detection limits for amino acids.

Liquid Chromatography with Potential-Sweep Pulsed Coulometric Detection. Potential sweep-pulsed coulometric detection (PS-PCD) (25) applies a waveform that is designed to reject the background by summing the charges due to oxide formation and oxide dissolution which are expected to be of equivalent magnitude but opposite in polarity. Table I11 depicts the PS-PCD waveform which superimposes on the detection period of the triple-pulse waveform a triangular scan which proceeds through oxide formation and dissolution to coulometrically remove oxide background signal. PS-PCD can virtually eliminate drift and changes associated with small variations in pH and composition of the mobile phase and changes in the total surface area of noble-metal electrode surface. Thus, PS-PCD techniques are expected to improve considerably the limits of detection and base-line characteristics in gradient elution chromatography. Variations in the length of the PS-PCD time periods and applied voltages can cause large changes in chromatographic

Th

Ti

Optimized for the least skew of chromatographic peak. Optimized for the best signal-to-noiseratio. a

T

16

I

Figure 5. Anion-exchange amino acid separation followed by PSPCD.

peak shape and sensitivity. Table I11 describes two waveforms. Waveform A produced the least skew of the chromatographic peak, whereas waveform B produced the best signal-to-noise ratio (S/N). The differences in response are attributable to differences in the oxide formation/dissolution kinetics and the inclusion of a secondary detection mechanism in the best signal-to-noisewaveform. The secondary detection mechanism involves the suppression of dissolved oxygen reduction by the adsorbed analyte. A thorough discussion of waveform optimization and detection mechanisms in PS-PCD will be covered in a later publication. Total waveform durations were limited to 1 s to ensure accurate representation of chromatographic peak shapes. Figure 5 illustrates the application of LC/PS-PCD with a saturated calomel reference electrode for the assay of the protein hydrolyzate containing 17 components. Although

ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989

2

T

500 nC I

,

10

niin

16

559

detected. A protein hydrolyzate containing 20 components was assayed by LC/PS-PCD using the glass pH reference electrode, and the result is shown in Figure 7. The underivatized primary and secondary amino acids were resolved and detected with excellent sensitivity. Excellent reproducibility was obtained for separations on repetitive injections, and no degradation of the performance of the glass pH electrode was apparent. Limits of detection for lysine by PS-PCD were determined to be 60 ppb (3 ng, 19 pmol) by waveform A (Table 111) and 40 ppb (2 ng, 13 pmol) for waveform B. Response for other primary amino acids was within &3 times that for lysine. These detection limites represent the present state-of-the-art for direct determinations of amino acids by pulsed electrochemical detection.

CONCLUSION

'I'irne

-

Figure 6. LC/PS-PCD for Pierce protein hydrolyzate using a glass reference electrode.

I

l4

1

I

11c

4 1 31

I

15

'I'irnc

-b

Figure 7. LC/PS-PCD for 20 amino acids: (1) arginine, (2) lysine, (3) glutamine, (4) asparagine, (5) threonine, (6) alanine, (7) glycine, (8) serine, (9) valine, (10) proline, (11) isoleucine, (12) leucine, (13) methionine, (14) histidine, (15) phenylalanine, (16) glutamic acid, (17) aspartic acid, (18) cysteine, (19) cystine, (20) tyrosine. PS-PCD is less sensitive to changes in surface oxide background, the chromatographic base line still suffered from some drift due to the pH gradient, but not the addition of the organic modifier. The use of PS-PCD and a glass pH reference electrode with gradient chromatography is illustrated Figure 6 for the same hydrolyzate used in Figure 5. Only a slight base-line perturbation is observed. All 17 amino acids were resolved and

The determination of underivatized amino acids by using anion-exchange separation with pulsed coulometric detection techniques is direct, sensitive, and simple. Detection limits for LC/LS-PCD are superior to UV-vis detection of ninhydrin adducts. The electrochemical characteristics of Au electrodes are well-suited to pulsed coulometric detection techniques for nondegassed systems.

LITERATURE CITED (1) Lehninger, A. Biochemistry, 2nd ed.: Worth Publishers: New York, 1975. (2) Nakagawa, T.; Shbukawa, A,; Kaihara, A,; Itumochi, T.; Taneka, H. J. Chromatogr. 1986, 353,399. (3) Grego. B.; Hearn, M. J. Chromatogr. 1983, 255, 67. (4) Abecassis, J.; DavU-Eteve, C.; Soun, A. J. L i 9 . Chromatogr. 1985, 8 , 135. (5) Hayashi, T.; Tsuchiga, H.; Naruse, H. J. Chromatogr. 1983, 274,318. ( 6 ) Hughes, G.; Wilson, K. J. Chromatogr. 1982, 242,337. (7) Kawashiro, K.; Morimoto, S.; Yoshida, H. Bull. &em. SOC. Jpn. 1983, 56, 792. (8) Polta, J.; Johnson, D. C. J. L l 9 . Chromatogr. 1983, 6 , 1727. (9) Ogden, G.; FoMi, P. L C - G C 1987, 5 , 28. (10) Hamiiton, P. Anal. Chem. 1963, 35,2055. (11) Tarr, G. Anal. Blochem. 1981, 7 1 7 , 27. (12) Black, S . : Coon, M. Anal. Blochem. 1982, 121,281. (13) Granbarg, R. L C 1984, 2 , 776. (14) Reitsma, B. Ph.D. Dissertation, Iowa State University, Ames, IA, 1987. (15) Adams, R. N. Nectrochemistry at Solid Electrodes: Marcel Dekker: New York, 1969. (16) Malfoy, M.; Reynaud, J. A. J. Electroanal. Chem. 1980, 174, 213. (17) Joseph, H. M.; Davies, P. Current Separations 1982, 4, 62. (18) Fleischman, M.; Korinek, K.; Pletcher, D. J. Electroanal. Chem. 1971, 31,39. (19) Kafil, J.; Huber, C. Anal. Chim. Acta 1985, 175,275. (20) Hughes, S.;Johnson, D. C. Anal. Chlm. Acta 1981, 132,11. (21) Edwards, P.; Haak, K. K. Am. Lab. 1983, April, 78. (22) Rocklin, R. D.: Pohl, C. A. J . Liq. Chromatogr. 1983, 6 , 1577. (23) Johnson, D. C. Nature 1986, 327,451. (24) Johnson, D. C.; Polta, T. 2. Chromatogr. Forum 1986, I , 37. (25) Neuburger, G. G.; Johnson, D. C. Anal. Chem. 1988. 6 0 , 2288. (26) Polta, T. 2. Ph.D. Dissertation, Iowa State University: Ames, IA, 1986. (27) Neuburger, G. G.; Johnson, D. C. Anal. Chem. 1987, 59, 203. (28) Martell, A.; Smith, R. Critical Stabiliv Constants; Plenum Press: New York, 1977; Vol. 3. (29) Naturella, M. Experimental Statistics ; National Bureau of Standards Handbook 91; NBS: Washington, DC, 1963; pp 6-19. (30) Installation Instructions and Trouble-Shooting Guide for the HPICAS8 Column; Document No. 032626, Revision 2; Dionex Corp.: Sunnyvale, CA, 1987. (31) Mead, D.. Jr. M.S. Thesis, Iowa State University, Ames, IA, 1988.

RECEIVED for review October 11, 1988. Accepted December 5,1988. This research was supported by a grant from Dionex Corp., Sunnyvale, CA. The National Science Foundation contributed a portion of the funding for purchase of the Cypress Systems instrument through Contract CHE-8312032.