2190
Anal. Chem. 1985, 57,2190-2196
(4) Armstrong, D. W.; Nome, F. Anal. Chem. 1981, 5 3 , 1662-1666. (5) Armstrong, D. W.; Stine, G. Y. J . Am. Chem. SOC. 1983, 105, 6220-6223. (6) Armstrong, D. W.; Stine, G. Y. J . Am. Chem. SOC. 1983, 105, 2962-2964. (7) Armstrong, D. W., personal communication. (8) TerweljGroen, C. P.; Heemstra, S.;Kraak, J. C. J . Chromatogr. 1978, 161, 69-82. (9) Arunyanart, M.; Cline Love, L. J. Anal. Chem. 1984, 56, 1557-1561. (10) Dorsey, J. G.; DeEchegaray, M. T.; Landy, J. S. Anal. Chem. 1983, 5 5 , 924-928. (11) Yarmchuk, P.; Welnberger, R.; Hirsch, R. F.; Cline Love, L. J. J . Chromatogr. 1984, 283, 47-60. (12) Mullins, F. G. P.; Kirkbright, G. F. Analyst (London) 1984, 109, 1217-1221. (13) Armstrong, D. W.; Stine, G. Y. Anal. Chem. 1983, 5 5 , 2317-2320. (14) Pramauro, E.; Pellzzettl, E. Anal. Chlm. Acta 1983, 154, 153-158. (15) Yarmchuk, P.; Welnberger, R.; Hlrsch, R. F.; Cline Love, L. J. Anal. Chem. 1982, 5 4 , 2233-2238. (16) Kirkbright, G. F.; Mullins, F. G. P. Analyst (London) 1984, 109, 493-496. (17) Landy, J. S.;Dorsey, J. G. J . Chromatogr. Sci. 1984, 2 2 , 68-70. (18) Armstrong, D. W.; Hlnze, W. L.; Bul, K. H.; Singh, H. N. Anal. Lett. 1981, 14, 1659-1667. (19) Weinberner, R.; Yarmchuk, P.; Cline Love, L. J. Anal. Chem. 1982. 5 4 , 155%-1558. (20) Dorsey, J. G.; Khaledi, M. G.; Landy, J. S.;Lin, J. L. J . Chromatogr. 1984, 316, 163-167. (21) Armstrong, D. W. Sep. Purif. Methods, in press. (22) Wall, R. A. J . Chromatogr. 1980, 194, 353-363. (23) Ghaemi, Y.; Wall, R. A. J . Chromatogr. 1980, 198, 397-405. (24) Ghaeml, Y.; Wall, R. A. J . Chromatogr. 1981, 212, 271-281. (25) Amano Pharmaceutical Co., Ltd. Jpn. Kokai Tokkyo Koho JP 58,215,554, Dec 1983, 3 pp; Chem. Abstr. 1984, 101, 3 5 5 6 5 ~ . (26) Carson, S. D.; Konigsberg, W. H. Anal. Biochem. 1981, 116, 398-401. (27) Regnier, F. E. “Receptor Protein Purification”; Alan R. Liss, Inc.: New York, 1984. Chang, J. P. “Abstracts of Papers”; 6th International Symposium on Column Liquid Chromatography, New York, NY, 1984; Abstr. No. 2a-
83. Barford, R. A.; Sllwinski, E. J. Anal. Chem. 1984, 5 6 , 1554-1556. Subcommittee E-19.06 Task Group on Llquid Chromatography of the American Society for Testing and Materials. J . Chromatogr. Sci. 1981, 19, 338-348. Krstulovic, A. M.; Brown, P. R. “Reversed-Phase Hlgh PerformanceLiquid Chromatography”; Wiley: New York, 1982; pp 16-24. Hamilton, R. J.; Sewell, P. A. “Introduction to High Performance Liquid
Chromatography”; Chapman & Hall: London, 1977; pp 16-27. (33) Karger, B. L.; Snyder, L. R.; Horvath, C. “An Introduction to Separation Sclence”; Wiley: New York, 1973; pp 135-146. (34) Synder, L. R.; Kirkland, J. J. “Introduction to Modern Liquid Chromatography”; Wiley: New York, 1979; Chapter 5. (35) Tyler, W. E. “Skew and Column Efflciency Program”; Exxon Research and Engineering Co.: Linden, NJ. (36) Mukerjee, P.; Mysels, K. J. “Critical Micelle Concentrations of Aqueous Surfactant Systems”; Natl. Stand. Ref. Data Ser. ( U S . Natl. Bur. Stand.) 1971, NSRDS-NBS 3 6 , 6-16, (37) Mukerjee, P. J . Phys. Chem. 1962, 66, 1733-1735. (38) Kalyanasundaran, K.; Thomas, J. K. I n “Micellization, Solubilization, and Microemulsions”; Mittal, K. L., Ed.; Plenum Press: New York, 1977; Vol. 2, pp 569-588. (39) Becker, P. J . Colloid Sci. 1985, 2 0 , 728-729. (40) Mathai, K. G.;Ottewlll, R. H. Trans. Faraday Soc. 1966, 62, 750-758. (41) Kuno, H.; Abe, R. Kolloid-2. 1961, 177, 40-44. (42) Corkill, J. M.; Goodman, J. F.; Tate, J. R. Trans. Faraday SOC. 1966, 6 2 , 979-966. (43) Abe, R.; Kuno, H. Kolloid-Z. 1962, 181, 70. (44) Rosen, M. J. “Surfactants and Interfacial Phenomena”; Wiley: New York, 1976; Chapter 2. (45) Attwood, D.; Florence, A. T. “Surfactant Systems”; Chapman & Hall: New York, 1983; Chapters 1 and 9. (46) Sorel, R. H. A.; Hulshoff, A.; Wiersema, S.J . Li9. Chromatogr. 1981, 4 , 1961-1985. (47) Hung, C. T.; Taylor, R. B. J . Chromatogr. 1981, 209, 175-190. (48) Knox, J. H.; Hartwick, R. A. J . Chromatogr. 1081, 204, 3-21. (49) Hung, C. T.; Taylor, R. B. J . Chromatogr. 1980, 202, 333-345. (50) Clunie, J. S.; Ingram, B. T. I n “Adsorption from Solution at the Solid/ Liquid Interface”; Parfitt, G. D., Rochester, C. H., Eds.; Academic Press: New York, 1983; Chapter 3. (51) Hinze, W. L.; Singh, H. N.; Baba, Y.; Harvey, N. G. Trends Anal. Chem. 1984, 3 , 193-199.
RECEIVED for review January 29,1985. Accepted May 24,1985. The authors thank the R. J. Reynolds Tobacco Co., Winston-Salem, NC, for their generous support of this research. This work was presented at the 17th Middle Atlantic Regional of the American Chemical Society, Middle Atlantic Regional Meeting of the Amercian Chemical Society, White Haven, PA, April 8, 1983, Abstract No. 89, and at the 4th Annual Liquid Chromatography Symposium, Research Triangle Park, NC, October 5, 1983, Presentation No. 12.
Hydro-Organic and Micellar Gradient Elution Liquid Chromatography with Electrochemical Detection Morteza G. Khaledi’ and John G. Dorsey* Department of Chemistry, University of Florida, Gainesuille, Florida 32611 The parameters affecting base-iine shifts in gradient eiutlon wlth electrochemlcal detection are dlscussed. These parameters are applied potentlal, moblle phase conductance, pH, flow rate and vlscoslty, resistance of the electrochemlcal cell, moblle phase electroactlve lmpurltles, electrode sensltlvlty, temperature, and solvent type. The magnltude of residual current change Is reported for water-methanol gradients at different potentlals. As expected, the magnitude of the base-line shin is far greater at high potentlals where masslve oxldatlon of water occurs. The extent of base-line shlfts caused by mlcellar concentratlon gradients at dlfferent potentials, pH, and lonlc strengths ls also studied. The size of mlcellar gradlent induced base-llne shifts can be greatly reduced, especially at high potentials, by balancing the pH and the conductance of the two micellar solutlons. The Improved compatlblllty of micellar gradlents with electrochemlcal detectors Is a great advantage over that of hydro-organic moblle phases. Present address: D e p a r t m e n t of Chemistry, U n i v e r s i t y of N e w Orleans, N e w Orleans, LA 70148.
One of the major drawbacks of electrochemical (EC) detectors is their limited compatibility with gradient elution ( I , 2). It is not uncommon to read in LCEC detection literature: “EC detectors are often incompatible with gradient elution” (1). Some even believe that gradient elution cannot be used with EC detectors (2,3). Their logic is based on the fact that background current is dependent on solvent and electrolyte composition, so any change in mobile phase composition would shift the base line. Recently, however, some authors have reported: “Contrary to previous reports, the compatibility of LCEC with gradient elution is excellent” ( 4 ) . This is based on the successful experience of some workers in using gradient elution with EC detectors (4-6). Different experimental conditions have caused this contradiction. Undoubtedly in order to answer the problem, it is necessary to study the parameters involved, the importance of each in contributing to the shift, and the possibility of decreasing the gradientinduced base-line shift to an “acceptable” value by adjusting the contributing factors. Despite the large number of publications on LCEC, to our knowledge, no one has addressed the issue in detail. Some workers have briefly mentioned the parameters involved; however, their efforts are far from a
0003-2700/85/0357-2190$01.50/0@ 1985 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
2191
Table I. Water-Methanol Gradient-Induced Base-Line Shiftsn gradient A.
0.10 M NaC10,
in H20 B.
C.
E, V
i,,, nA
i,,, nA
nA
Ai,"'",
5 min
0.10 M NaC10, in CHBOH
0.80 1.00 1.20
11 39 125
7 16
48
10 60 235
0.80 1.00 1.20
16 26 70
6 15 43
10 24 50
0.80 1.00 1.20
24 52 120
15 46 107
150 275
--+
buffer pH 2.50 0.09 M NaClO,
--+
0.10 M NaC10,
5 min
in CH30H
buffer pH 6.25 0.09 M NaC10,
-*
0.10 M NaClO,
5 min
in CHBOH
% change
63 92 71
40
air,,initial residual current corresponding to 100% aqueous solution; irf, final residual current corresponding to 100% methanol; Airma Lmm- i,"": % change = (Ai.""/i,) X 100.
comprehensive study of the subject (6, 7). The work reported by Allison et al. ( 4 ) is probably a good example of the potential use of gradient elution with EC detectors. The initial mobile phase composition of 20% methanol in a pH 5.0 buffer was changed to CH30H:THF: buffer (765:19) in 10 min, Le., a gradient range of 60% organic modifier. The OPA derivatives of 21 amino acids (167 pmol each) were separated and detected a t a working potential of 0.70 V vs. Ag/AgCl and a t a sensitivity of 500 nA full scale. Potentials such as 0.70 or 0.80 V, where the overall residual current is small cause a small base-line shift even with a wide organic modifier concentration gradient and the subsequent large changes in mobile phase properties such as pH, conductance, etc. In addition, the gradient-induced base-line shifts of a few nanoamperes look amost nonexistent when operating a t a sensitivity of 500 nA full scale. In this case, the compatibility of gradient elution with electrochemical detection is "excellent" (4). At high potentials, however, (e.g., 1.20 V), where massive oxidation of water occurs, the magnitude of the shift (typically 100 or 200 nA) is totally unacceptable for any typical sensitivity range. One can then say that gradient elution cannot be used with EC detection (2, 3). However, gradient elution can be performed at even high potentials by narrowing the gradient range and using proper experimental measures. A study of parameters causing the shift and affecting its size should then give a better insight into these problems. In amperometric detection, the application of a potential to the working electrode generates a large transient or charging current, which decays rapidly. The steady-state background current is composed of two components. Firstly, the residual current results from electrochemical processes associated with the electrode surface. For glassy carbon electrodes these processes extend from the oxidation of functional groups on the electrode surface and formation of oxide layer(s) whenever oxygen evolution occurs to the dissolution of adsorbed species and oxide layer, etc., depending on the history of the solid electrode (8, 9). The second component is a result of electrolysis (oxidation in this work) of traces of electroactive impurities in the solution and/or oxidation of the solvent itself, Thus, the magnitude of the residual current is dependent on the surface condition of the solid electrode and the rate of impurity and solvent oxidation. The extent to which the impurities or mobile phase undergo oxidation is a function of bulk mobile phase properties such as pH, viscosity, etc., along with the magnitude of the working potential. Variation of the mobile phase composition during gradient elution can then alter the extent of electrolysis of the mobile phase and impurities, thus resulting in a residual current change. Also any change in the electrode surface condition due to changes in mobile phase composition contributes to the base-line shift. Furthermore, changes in the double layer caused by solvent
=
composition variation induces charging current flow (at least transiently) (10). In general, changes in the bulk mobile phase properties such as pH, viscosity, electroactive impurities, ionic strength, and solvent type along with other parameters like temperature, flow rate, cell resistance, potential, and electrode sensitivity influence the base-line shift.
EXPERIMENTAL SECTION The system incorporated an Altex (Berkeley, CA) Model 322 gradient liquid chromatograph with two Model lOOA dual reciprocating pumps, an Altex 210 injection valve with 20-pL loop, and an Altex Ultrasphere octyl column (4.6 X 250 mm) with 5-bm particle diameter. A hand-packed saturator precolumn was prepared by using irregularly shaped silica gel of 25-40 pm diameter (Macherey-Nagel Co., West Germany) and was located before the injector. The mobile-phase temperature was controlled by using water jackets for both the precolumn and the analytical column. A Neslab Model 850 (Neslab, Portsmouth, NH) bath circulator was used to thermostat the columns. The electrochemical detector was an LC-4 amperometric controller connected to a TL-5 thin-layer detection cell (Bioanalytical Systems, West Lafayette, IN). The glassy carbon working electrode was polished to a mirror finish with 0.1 pm alumina (Fisher Scientific, Fair Lawn, NJ) and polishing paper (Buehler Co., Evanston, IL) and was then electrochemically pretreated by cyclic polarization at +1.50 V and -1.50 V each for 2 min before each experiment. The reference electrode was Ag/AgCl. A Wescan (Wescan Instruments, Santa Clara, CA) 213 conductivity detector was employed in series before the EC detector. Deionized, doubly distilled water was irradiated by UV light to remove trace organics. The surfactant was sodium dodecyl sulfate (SDS, Puriss. grade, Fluka Chemical, Hauppauge, NY) and was used as received. Surfactant solutions were made in water:l-propanol (97:3) and were filtered through a 0.45-pm Nylon-66 membrane filter (Rainin Instruments, Woburn, MA). The solvent (H201-propanol)was stored in a 6.5-gallon container to ensure homogeneity of impurities for all experiments. Solutes. The phenolic compounds were purchased from Mallinckrodt (St. Louis, MO), Eastman Kodak (Rochester, NY), and Chem Service (Westchester, PA) and were dissolved in a solution of 0.10 M SDS. RESULTS AND DISCUSSION Table I shows the maximum background current change caused by water-methanol gradients at three different potentials. The percentage of the change as compared to the initial residual current is shown for gradient B. As a measure of gradient compatibility, Airm" is the important factor and not the percent change. For example, while the Airma" value at 1.20 V is 5 times that of the 0.80 V, the corresponding percent changes are almost equal. For practical purposes, a residual current change of 10 nA is much more "acceptable" than 50 nA, as it is the operating sensitivity (nanoamperes full scale) which determines whether a Airmax value is acceptable. Table I also illustrates a limiting problem of hydro-organic gradients; the residual current goes through
2192
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
a maximum (or minimum) during the gradient. Note that Ai,"" is greater than Airf- Airl for all cases. This problem will be discussed later. PH. One of the major constituents in reversed-phase eluents is water. The oxidation potential of water is a function of pH, as H+ is one of the products of water oxidation: 2Hz0 --+ Oz + 4H+ + 4e-. Thus, EHlo= 1.23 - 0,05gpH, considering the pressure of O2as 1atm and NHE as reference electrode. Water can then be oxidized more easily as the pH increases, producing larger background currents. In addition, H+ is involved in glassy carbon surface redox reactions (11). The pH variation during the course of gradient elution can then greatly influence the base-line shift, especially at potentials high enough to Cause solvent decomposition. At high tentials, buffering the mobile phase composition during the gradient is essential in order to minimize the p~ contribution to the base-line shift. However, this is difficult a t best for a wide change of concentrations in water-methanol mixtures. Comparison of gradients A, B, and c, s h o w in Table I, reveals the pH effect to some extent. The final mobile phase cornposition for all three gradients wm 0.10 M NaC104 in CH,OH, while the initial pH of the aqueous mobile phases was different. The supporting electrolyte and the buffer concentrations were chosen so that the conductance range for all three gradients was equivalent. In gradient B where the initial aqueous mobile phase has a pH of 2.50, the initial residual current and the maximum change are smaller a t high potenti& compared to those of gradients A and C. This is probably due to the overall smaller extent of water decomposition at pH 2.5. At low potentials, such as 0.80 V, however, where water oxidation will not occur, the values are almost equal for gradients A and B. In gradient C, the use of an aqueous buffer of pH 6.2 should yield comparable (if not smaller) results to those of gradient A where the initial solution is unbuffered and the pH is within the same range. Surprisingly, the overall residual currents in gradient C are almost twice those of gradient A even a t 0.80 V. The anomalous residual current in gradient are probably due either to differing electrode surface conditions or to electroactive impurities present in the buffer Background current values a t solid electrodes are subject to some irreproducibility. However, the similar residual current values for the final ('.lo NaC104 in CH30H) for phase gradients A and a t o*80v and l-0 shows that pretreatment of 'he glassy carbon before each experiment minimized the I in problem. A comparative study of the resu1ts in terms of the PH effect is valid only to a certain degree. First, the aqueous buffer is not totally operative for hydro-organic systems. The phase pH can therefore be neither nor known Over a wide range Of water-methano1 compositions. Secondly, the residual Current Values carry an uncertainty component. At high potentials such as 1.20 v the values were irreproducible in terms of repetetive gradient runs. However, similar overall trends were observed as the gradients A, B, and C were compared S~veraltimes in a Period of a few days. The residual currents of gradient B Were consistently the smallest and those of gradient c were the largest. A more detailed discussion of the subject is given below. The Problems inherent in variation of pH with water-methanol gradients were resolved to a great extent with micellar mobile phases (vide infra). Viscosity. The effect of viscosity and flow rate on background current can be understood in terms of changes in the transport rate of electroactive impurities to the electrode surface. The resulting current from electrolysis of impurities is then a function of flow rate, viscosity, and diffusion coefficient. Therefore, higher residual current is observed at faster flow rates with less viscous media and large diffusion coef-
ficients. The viscosity variation in the gradient process then affects the base-line shift as it influences the diffusion coefficients of the electroactive species. Mobile Phase Purity. The role of mobile phase purity iS Clear. Variation in concentration of electroactive impurities during gradient elution obviously causes base-line shifts. Electrochemically pure reagents minimize the problem, and PreelectrolYsis of the mobile phase may occasionally be necessary, Cell Design and Ionic Strength. The detector cell design and ionic strength do not directly affect the residual current and its changes. However, both factors are important considerations as they determine the uncompensated resistance of the solution between the electrodes. The ohmic potential drop, which directly affects the applied potential, varies with changes in the position of the electrodes (cell design) and the conductivity Of the solution (ionic strength). Furthermore, changes in dielectric constant during a hydro-organic gradient Will affect the ionic strength (and conductance) even when the Pure solvents (water and methanol) are equimolar in Supporting electrolyte. The interfacial electrode potential, Ei, which iS the potential across the working electrode/solution interface, is related to the applied potential E , as follows:
Ei = E,
- Eref- IR,
where Erefis the reference electrode potential and R, is the uncompensated resistance of the solution. E, and Erefare constant; therefore the Ei value is directly affected by iR drop. The major problem with the cell design used in this work is a large uncompensated solution resistance due to the large distance between the working and auxiliary/reference electrodes positioned downstream from the working electrode. A typical cell resistance of 5 X lo6 D is reported using an aqueous buffer whose resistance in a macroscopic conductometric cell is only a few hundred ohms (13). In nonaqueous solvents, the problem of iR drop becomes more pronounced because of low conductivity. As a result, a large current flow between the auxiliary and working electrode creates a severe iR drop problem and loss of control Over working potential (14, 15). Background current and even the current generated due to peak elution can be the major contribution to iR drop (15, 16). In addition, due to the specific geometry in the thin-layer cells, the current between the working and the electrode has to pass along the channel electrode surface. This generates iR drop along the surface and thus nonuniform polarization of the working electrode (15). At higher currents where the iR drop is larger, the potential difference between the upstreamand downstream edges of the electrode is pronounced resulting in a nonhomogeneous rate of electron transfer along the face of the electrode (14, 15, 17). This translates into a nonlinear behavior of the detector when large samples are injected (14,15) and also aggravates the gradient induced base-line shift. In gradient elution, the iR drop changes as a result of variations in conductivity and so the uncompensated resistance of the cell. Thus, the applied POtential along the electrode surface is continuously changing during the course of the gradient, and therefore the extent of electrolysis of impurities and the mobile phase is varied. It is possible to decrease the background current change even with a high impedance cell and high working potential providing the conductivity of the mobile phase is held constant. The result of the fixed uncompensated resistance of the solution is a constant iR drop. Although the iR drop still exists along the electrode surface and the electrode is yet polarized nonuniformly, that does not cause the base line to shift. In gradient B (Table I) where the overall residual current is kept small by operating at low pH, the contribution of iR variation is smaller than gradients A and C for equal changes in conductivity.
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
Potential. The extent of electrolysis of both mobile phase impurities and water is dependent on the potential. The condition of the electrode surface is also a function of working potential. As shown in Table I, the residual current corresponding to each mobile phase composition as well as the size of the change for each gradient increases with potential, despite the fact that the change in mobile phase properties for a given gradient is the same at all potentials. Although the effect of potential on residual current is quite clear, the dependence of base-line shift on potential cannot be explained unless the relationship between the potential and other parameters affecting the shift is well understood. For example, pH gradients have greater impact on the residual current change at higher potentials where water decomposition and/or other pH dependent electrolysis reactions occur. The conductance change, as well, has a greater influence in terms of the change of iR drop at 1.20 V than for 1.00 and 0.80 V simply because of the overall large, initial residual current. Also, the range of “electroactive” compounds becomes wider and more impurities are subject to oxidation at higher potentials. Thus, any change in their concentrations will be observed in terms of the sloping base line. Finally, the glassy carbon’s surface condition changes with potential and is probably more SUSceptible to the variations in the solution at its interface at extreme potentials. Electrode Sensitivity. Solid electrodes have poor reproducibility and generally produce high residual currents. This is mainly because of the surface inhomogeneity of the electrodes and their surface oxidation a t positive potentials (13). The main disadvantage of glassy carbon is the loss of surface activity with use, which results in a decrease in sensitivity and poor reproducibility. Pretreatment of the surface such as mechanical polishing followed by alternate polarization of the electrode is used to maintain the sensitivity and the reproducibility. A well-polished glassy carbon surface is reported to be virtually free of functional groups (18); however, carbonyl groups and even carboxyl groups can be formed as the electrode is immersed in a solution containing strong oxidizing agents and/or anodic potential is applied (18). Laser and Ariel (19) confirmed the existence of surface groups. They concluded that anodic polarization results in surface group formation such as carbonyl groups that subsequently can be reduced to hydroxyl groups at more negative electrode potentials. Also the formation of quinone/ hydroquinone cannot be ruled out (18,19). Dieker et al. (20)also supported these results and demonstrated that the large residual currents observed at glassy carbon electrodes are largely due to redox reactions of the electrode surface rather than electroactive impurities or charging the electrical double layer. Changes in electrode surface condition could then greatly affect the background current variation. This is likely to occur during a gradient run since the solvent composition at the interface of the surface is changing. This factor is more pronounced at higher potentials where one of the mobile phase constituents is being oxidized. For example, the oxygen evolved from oxidation of water is sorbed on the glassy carbon surface forming an oxide layer or layers. Replacing the water with an organic modifier during a gradient run alters the characteristics of the electrode surface at the solution interface; thus the size of the electrode’s contribution to residual current may vary. The effect of the surface condition of the electrode on the magnitude of residual current and base-line shift was studied by different pretreatment procedures. First the electrode was polished and treated by cyclic polarization of the electrode at +1.50 V and -1.20 V each for 5 min. The working potential of 1.20 V was chosen and the mobile phase composition was changed from acetate buffer pH 4.2 to 0.05 M NaC104 in
2193
CH30H. The second experiment was performed 24 h later under identical conditions except the electrode was not polished but was electrochemically pretreated at +1.50 V and -1.20 V each for 8 min. The magnitude of the residual current, and consequently the base-line shift, was larger for the second day by almost 100%. Therefore, the tedious job of glassy carbon pretreatment (both mechanical and electrochemical) was carried out before each experiment to improve the reproducibility of the surface conditions. Obviously, for the study of the effect of mobile phase compositions on the base-line shift, reproducing the electrode surface is of prime importance. The size and the trend of residual current change obtained for both forward and reverse gradients (e.g., methanol to water) as well as the residual currents corresponding to specific mobile phase compositions in the repetitive runs were used as the measure of reproducibility. The repetitive forward and reverse water-methanol gradients were extremely irreproducible at 1.20 V. For example, the residual current values for initial and final mobile phase composition as well as the size of the residual current change for gradient B at 1.20 V decreased by 20-30% in the second gradient run and 10-20% in the third run. At 0.80 V and 1.0 V the corresponding value was between 5% and 10%. However, the overall residual current and the size of the shift were consistently smaller for gradient B and larger for gradient C. Role of t h e Solvent. The change of solvent and the continuous potential change along the electrode surface during a gradient run change the structure and capacitance of the double layer at a solid electrode (10). As a result of the continuous change in solvent composition, charging current is generated continuously. Whether these variations contribute significantly to base-line shift is not known as the double layer structure at solid electrodes is not yet fully understood. Figure 1 shows the hydro-organic gradient-induced base-line shift for gradients A, B, and C and their corresponding reverse gradient (Le., methanol to water). Interestingly, the residual current passed through a maximum at 5-7 min after the start of the gradient for all cases. The maximum occurrences were also observed at 0.80 V and 1.0 V. The maxima were not due to preconcentration and subsequent elution of mobile phase impurities, which can happen during gradient elution, as no column was used. A t first, it was thought that the peak corresponded to a specific water:methanol composition where viscosity is maximum. If so, then the residual current trend for a reversed gradient should be the mirror image of that of the forward. On the contrary, for gradient A, the residual current rises exponentially to the new base line as the mobile phase composition is changed from 0.10 M NaC104 in CH30H to 0.10 M NaC104 in HzO. For gradient C, it passes through a minimum, then a maximum, as a result of a reversed gradient of CH30H to a pH 6.25 buffer. The magnitude of the residual current change for reversed gradients A and C was smaller than the corresponding forward gradient by a factor of 3 to 4 at 1.0 V and 1.20 V. The maximum occurrence was observed for reversed gradient B; however the peak position is again 5-7 min after the gradient start. Therefore it cannot correspond to the same mobile phase composition as that of the forward gradient. The shapes of the background current changes for the forward and reversed gradients were reproducible at all operating potentials. It is interesting to note that the first 20% change in methanol concentration in gradient B (Le., from buffer pH 2.4 to buffer:CH,OH 8020) results in a much larger (by a factor of 4 t o 5) residual current change than the subsequent 20% step gradients, especially at 1.20 V. Although the exact causes are not clear, it is likely that the change in the solvent type affects the electrode’s surface condition and so the base-line shift.
2194
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
Table 11. Reproducibility of Residual Current in Micellar Gradients" E, v
day 1
i, of M.P. I, nA day 2
0.80 1.00 1.20
18 25 60
23 33 67
i, of M.P. 11, nA day 3
day 1
day 2
day 3
10
13 32 76
15
36 78
16 29 78
17 70
i, of M.P. 111, nA day 1 day 2 38 58 98
39 50 97
"Mobile phase (M.P.). M.P. I, 0.01 M SDS,0.05 M NaClO,, pH 2.35; M.P. 11, 0.40 M SDS,0.05 M NaC104, pH 2.35; M.P. 111, 0.01 M SDS, 0.44M NaClO.,, p H 2.35. Forward Gradients
Reverse Gradients A
