Langmuir 1988,4, 341-345 gradually shifts to lower wavenumbers, losing intensity, while the 13CN-band gradually shifts to higher wavenumbers, developing its intensity. Figure 9 shows the C-N stretching frequency of the two isotopic CN- ions adsorbed on the gold surface as a function of the percent concentration of the 13CN-ions in the adsorbed layer. This result shows clearly that the lateral interaction, most likely dipole coupling, shifts the C-N stretching frequency with coverage by ca. 4-5 cm-l. This is almost equal to the total shift observed by changing the surface coverage in the pure 12CN-solutions. These are important results because they clearly point to insignificant lateral interactions between adsorbed CN- ions, compared to the potential-driven shift of 30 cm-'/V. The shift seen here should be compared to the much larger shifts observed from the CO/Pt system" or the
341
CO/Cu system18 in ultrahigh vacuum. The small frequency shift as a function of coverage for the gold/cyanide system may be explained by assuming that the maximum coverage of CN- on gold a t -1.0 V is smaller than a full monolayer, although the IR absorption intensity is a maximum at this potential. This is frequently true for the anions specifically adsorbed on negatively charged electrode surfaces, which is the case here; -1.0 V is much more negative than the PZC of polycrystalline gold, 0.005 V.19
Acknowledgment. This work was supported in part by the Office of Naval Research. Registry No. Au, 7440-57-5;CN-, 57-12-5;K2S04,7778-80-5. (19) Clavilier, J.; Van Huong, C. N. J. Electroanal. Chem. 1977, 80, 101.
Studies of Adsorption Processes in the Absence of Added Electrolyte: Phase Changes in Coumarin Adsorbed at Conventional and Micro Mercury Electrodes A. M. Bond*l and F. G. T h o m a ~ * ~ ? ~ Division of Chemical and Physical Sciences, Deakin University, Waurn Ponds, 321 7,Victoria, Australia Received May 1, 1987. In Final Form: August 13,1987 Electrochemical studies of adsorption processes generally require the presence of large concentrations of electrolyte in contrast to other physical methods of measurements where the electrolyte is not regarded as mandatory. In this work alternating and direct current polarographic and voltammetric techniques for studying adsorption processes in the absence of added electrolyte have been investigated at conventional sized electrodes and at microelectrodes. The model system chosen was an examination of the effect of decreasing the added sodium fluoride concentration from 0.9 M to zero on the adsorption of coumarin onto mercury electrodes from its saturated solution at 15 "C. At all electrolyte concentrationsthe adsorbed monolayer of the neutral coumarin molecule formed three surface phases in the relevant potential ranges, and data are consistent with those predicted by theory. However, quantitative dc and ac data were obtained only for solutions containing greater than M concentrations of soditim fluoride at conventional sized electrodes. At M or lower added electrolyte concentrations, no useful ac response was observed because of the high level of uncompensated resistance present even when a three-electrode potentiostat was used with positive feedback circuitry. Under these conditions the dc response gave qualitatively, but not quantitatively, useful data. The cyclic voltammetric response obtained at low or zero added electrolyte concentrationswith a hemispherical mercury microelectrode of 5-pm radius was significantly less distorted by Ohmic IR drop than it was with conventional sized electrodes. Stochastic and other features of the phase transitions were readily observed even without added electrolyte. Although the alternating current voltammetric response with the microelectrode was vastly superior to that from a conventional sized electrode in the absence of added electrolyte, the task of accurately measuring picofarad capacitances in the presence of megaohm resistances by means of alternating current techniques presents considerable difficulty.
Introduction Electrochemical techniques of the voltammetric kind are widely used to study electrode surface p r o c e ~ s e s . Al~~ most by definition, electrolyte is present for the purpose (1)Deekin University. (2) James Cook University, Townsville, Queensland, 4811, Australia.
(3) On leave at Deakin University. (4) Bard,A. J.; Faulkner, L. R. In Electrochemical Methods, Fundamentals and Applications; Wiley: New York, 1980. (5) Bond, A. M. In Modern Polarographic Methods in Analytical Chemistry; Marcel Dekker: New York, 1980. (6) Greeb, R.; Peat, R.; Peter, L. M.; Pletcher, D.; Robinson, J. In Instrumental Methods in Electrochemistry; Ellis Horwood: Chichester, 1985.
0743-7463/88/2404-0341$01.50/0
of minimizing the resistance, controlling the potential in the outer Helmholtz plane, and eliminating migration current. In contrast, nonvoltammetric techniques for studying surface processes do not have the mandatory requirement of the addition of an indifferent electrolyte. Unfortunately, with voltammetric studies, electrolytes are usually added a t very high concentrations, in the 0.1-1.0 M concentration range, which compounds the problem of comparing data from electrochemical measurements with those obtained from other techniques. Under these conditions of high electrolyte concentration, the potential in the outer Helmholtz plane (OHP) is approximately constant a t potentials removed from the potential of zero charge. This minimizes the influence of the diffuse part 0 1988 American Chemical Society
nond ana ~ " h o m a s
342 Langmuir, Vol. 4, No. 2, 1988 of the double layer on the adsorption behavior. In view of the limited knowledge of the effects of electrolyte on the adsorption of molecules at electrode surfaces, we have commenced a study of adsorption at electrode surfaces in the presence of small amounts or without deliberately added electrolyte. Recently, advances in instrumentation6 and and the development of microelectrode technique^^-'^ have opened the way for such studies to be undertaken with relatively minimal Ohmic IR drop problems. Conclusions from studies without added electrolyte have the distinct advantage of being directly comparable with those obtained from adsorption studies undertaken with nonelectrochemical techniques, e.g., colloid adsorption studies and fundamental studies of separation techniques based on adsorption. In this initial report, we describe the adsorption of coumarin on a variety of mercury electrodes in the presence of little or no added electrolyte. The adsorption of this compound a t dropping mercury electrodes has been well characterized in the presence of high electrolyte concentrations (0.5 M sodium sulfate or 0.9 M sodium fluoride),13-lsbut no information is available in dilute or zero added electrolyte. At low temperatures and near saturated concentrations, coumarin forms a quasi-solid adsorbed monolayer at the mercury electrode in addition to two structurally different "fluid" monolayers a t more positive and negative potentials, respectively. The condensed layer is characterized by a very low (3.5 pF cm-? potential-independent double-layer capacitance that is stable on slightly negatively charged electrodes. The condensed layer can be formed from either of the two fluid monolayers by a nucleation and growth mechanism that is induced by a potential step from the potential domain of stability of either of the two fluid monolayers to that of the condensed layer. In view of the fact that the transitions between the different structures in the adsorbed monolayers are characterized by large, sharp changes in the double-layer capacitance and charge-potential curves, coumarin was considered an ideal system to examine the influence, if any, of added electrolyte on the adsorption behavior of a neutral molecule a t an electrode surface. Studies in this work utilized ac polarography and cyclic voltammetry at conventional hanging and dropping mercury electrodes and a t mercury-coated microelectrodes. The techniques are all well established for studying different aspects of surface processes. The electrodes used cover a wide range of sizes so that the problems associated with uncompensated resistance and the measurement of extremely small currents could be examined.
Experimental Section Materials. Coumarin (Merck,zur Synthese)was recrystallized from aqueous methanol (mp 68.5-69 "C). Saturated solutions (7)Wightman, R.M.Anal. Chem. 1981,53,1125A. (8)(a) Bond, A. M.; Fleischmann, M.; Robinson, J. J. Electroanal. Chem. 1984,168, 299;(b) 1984,172,11. (9)Bond, A. M.;Lay, P. A. J. Electroanal. Chem. 1986, 199,285. (10)Caddy, J.;Khoo, S.B.; Pons, S.;Fleischmann, M. J.Phys. Chem. 1985,89,3933. (11)Chen, J.; Georges, J. J. Electroanal. Chem. 1986,210, 205. (12)Ciszkowska, M.; Stojek, Z . J. Electroanal. Chem. 1986,213,189. (13)Damaskin, B. B.; Dyamkina, S.L.; Petrochenko, S. I. Electrokhimiya 1969,5,935. (14)Partridge, L. K.; Tansley, A.C.;Porter, A. S. Electrochim. Acta 1966,11, 517. (15)Ghali, H.A.;Bekheet, A. M.;Mazhar, A. A. Electrochim. Acta 1982,27,595 and references cited therein. (16)Thomas, F. G.; Buess-Herman, Cl.; Gierst, L. J. Electroanal. Chem. 1986,214, 597.
II I
0 0
- 0.5
-1.0
- 1.5
E ( V vs SCE)
Figure 1. Double-layer capacitance curves as determined by ac polarography for saturated solutions of coumarin in the presence of sodium fluoride at 15 "C. ac voltage = 10 mV p p , quadrature current measured. dc scan rate = 2 mV s-l, frequency 113 Hz M (71.3 Hz) and zero added NaF (41.3 Hz): (-) except for 0.9 M NaF; (--) 0.1 M NaF; (-- --) 0.01 M NaF; (- - -) 0.001 M NaF; 0.0001 M NaF; 0 NaF. (e-)
(-e-)
were prepared by vigorous stirring of excess coumarin with electrolyte solution for at least 24 h at 15 "C. This solution was fiitered directly into the electrolytic cell. Sodium fluoride (BDH AnalR) was used as supplied. Distilled, deionized water and triple-distilled mercury were used, and all solutions were purged with high-purity nitrogen (C.I.G.). Equipment and Techniques. All measurements were made at 15 "C in a conventional water-jacketed cell (Metrohm),which enabled electrodes to be repositioned to within f l mm. Either a Radiometer 701 double-junction saturated calomel electrode (SCE) or a small platinum (1 mm X 0.5 mm diameter) quasireference electrode (PRE) was used as the reference electrode. Working electrodes were conventional dropping mercury electrodes (DME) with flow rates from 0.4815 to 0.5256 mg s-', a hanging mercury drop electrode (HMDE, Metrohm), or a mercury-coated platinum microdisk electrode. The platinum microelectrode onto which the mercury was plated via the procedure of Wehmeyer and Wightman" was of 2.5-fimradius and sealed in glass. The resulting mercury microelectrodesof hemispherical geometry had radii in the range 5.1-5.4 pm. A 1-mm-diameter platinum wire served as counter electrode (CE). The cell and all electrode assemblies were mounted in a steel Faraday cage. For measurementswith conventionalelectrodes,cyclic voltammograms were recorded by using a PAR 174A polarographic analyzer combined with on Omnigraphic 2000 XY recorder (Houston Instruments). The polarographic analyzer was coupled with a PAR 129 lock-in amplifier and an Optimation RCD 10 sine-wave generator via a PAR 174/50 interface for the alternating current measurements (drop time of 5 s). Positive feedback circuitry for minimizing ohmic IR drop effects was introduced via the PAR 174/50 interface. For microelectrodework, cyclic voltammograms were recorded with a Keithley 480 picoammeter and a Houston 100 X-Y recorder. Alternating current measurements used the same instrumentation as for conventional electrodes except that the picoammeter was used as a preamplifier.l* Because of the slow rise time of the picoammeter, ac measurements were limited to frequencies of less than 50 Hz. A three-electrodepotentiostated system with positive feedback circuitry for IR compensation was (17)Wehmeyer, K.R.;Wightman, R. M. Anal. Chem. 1985,57,1989. (18)Bixler, J. W.; Bond, A. M.; Lay, P. A.;Thormann, W.; van den Bosch, P.; Fleischmann, M.; Pons, B. S. Anal. Chim. Acta 1986,187,67.
Langmuir, Vol. 4, No. 2, 1988 343
Adsorption Processes in the Absence of Electrolyte
Table I. Effect of Sodium Fluoride Concentration on the OHP Potential b2 at Eel@= -1.000 V vs SCE and on the Value of Eel= at Which E - 62 = -1.000 V NaF concn, 429 v Eelec,V vs SCE mol dm-3 (Eelw = -LOO0 V vs SCE) (E - 62 = -LOO0 V) 1
lo-’ 10-2 10-3 10-4
-0.044 -0.090 -0.143 -0.197 -0.252
-1.538 -1.593 -1.652 -1.711 -1.770
used with conventional working electrodes whereas the microelectrode work was undertaken in a two-electrode mode. Solution conductances were measured with a Radiometer CDM 80 conductivity meter. The concentrations of coumarin in ita saturated solutions were determined spectrophotometrically (at 278 and 308 nm) with the aid of a Unicam SP800 spectrophotometer.
Results and Discussion Figure 1 shows conventional ac polarograms at a conventional DME for a saturated solution of coumarin in the presence of various concentrations of NaF at 15 “C. In 0.9-0.01 M NaF instrumental problems are minimal and reliable double-layer differential capacitances can be measured directly. This contrasts with the measurements at lower concentrations where unreliable and nonreproducible capacitance values are obtained with a conventional DME because of the high-resistance component (“best” values shown in Figure 1 for 0-0.001 M NaF). The data in Figure 1 show four distinct regions labeled I-IV. Region I corresponds to an adsorbed monolayer of coumarin with the molecules lying flat on the mercury surface. In region I1 a quasi-solid monolayer is formed at the mercury electrode with the molecules in a compact end-on perpendicular orientation, whereas region I11 corresponds to the side-on perpendicular orientation of the molecules in the adsorbed monolayer. Finally, region IV corresponds to the onset of the faradaic reduction of coumarin. Further details concerning the adsorption of coumarin on mercury electrodes are available in ref 16. At all concentrations of fluoride, including zero added, the sharp transitions between the various structures (I and II; I1 and 111)can be readily observed (see Figure 1). This demonstrates that the structures of the three different adsorbed phases are independent of the added fluoride concentration. However, it must be noted that even with no added fluoride some ions are present. These arise from the dissociation of water and any trace impurities in the coumarin (e.g., organic acids). The specific conductance of saturated solutions of coumarin in water was in the range 10-16.2 $3 cm-’, and the pH of the deaerated solutions was 6.5 f 0.1. Although the sharp discontinuities in the CE curves do not occur at the equilibrium potential values of the surface phase transitions,16 a definite trend to more negative potentials is noted as the concentration of fluoride decreases. The negative shift of the domain of potential stability of the condensed phase I1 is as expected from an extension of Grahame’s classical ~ o r k The . ~ tabulated ~ ~ ~ data show that the calculated values of $z (potential of OHP) a t a given electrode potential increase as the concentration of electrolyte decreases. If it is assumed that the potential difference across the compact double layer is the dominant parameter in the determination of the transition potential, then the observed trend can be explained. That is, as the concentration of electrolyte is decreased, the magnitude (19)Grahame, D.C. Chem. Reu. 1947,41, 441. (20) Russell, C.D.J. ElectroanaL Chem. 1963, 6, 486.
I
100 n A
- 0.5
-1.0
E (VvsSCE)
Figure 2. Cyclic voltammograms for saturated solutions of coumarin at 15 O C in the presence of (a) no added NaF, (b) lo4 M NaF, and (c) M NaF. Scan -0.1 -1.4 -0.1 (V). Scan rate 100 mV s-l. HMDE area 2.94 mm2. Scan 2 plotted (scans 2-10 identical).
- -
0
- 0.5
-1.0
E ( V v o SCE)
Figure 3. Effect of scan rate on the cyclic voltammograms of saturated coumarin solutions in lo4 M NaF at 15 OC. Scan rates. (-) 20 mV s-*; (--) 50 mV s-l; (---) 100 mV s-l; (-*-)200 mV s-l. HMDE area 2.94 mm2. Scan 0 -1.4 0 (V). Scan 2 plotted. No IR compensation applied.
- -
of the electrode potential must be increased to produce a given charge and/or potential must be increased to produce a given charge and/or potential difference across the compact part of the double layer.20 An alternative explanation to the potential shifts may be the variable junction potential present as the electrolyte concentration is decreased, although (for the reasons given below) this is not considered likely. The limitations in the classical polarographic method at low electrolyte concentrations are clearly revealed in Figure 1and arise from uncompensated resistance. At the high resistances associated with low or zero added electrolyte concentrations, positive feedback circuitry is unreliable,21~22 providing incomplete IR compensation and (21)Brown, E.R.;Smith, D.E.;Booman, G.L. A w l . Chem. 1968,40, 1411.
Bond and Thomas
344 Langmuir, Vol. 4, No. 2, 1988
A
/
50 PA
I
I
1
1
- 0.5
1
,
,
1
1
1
1
1
,
-1.0
/
Figure 4. Cyclic voltammogramsof saturated coumarin solutions in M NaF at 15 "C. Scan rates: (-) 20 mV 8 ;(--) 50 mV s-l; (- - -) 100 mV s-l; (- -) 200 mV s-l. Scan -0.1 -1.4 -0.1 (V). Scan 2 plotted. HMDE area 2.94 mm2. Maximum IR
.
I
I
~
- -
compensation applied.
poor reproducibility, both factors making quantitatively useful capacitance data impossible to obtain. However, the transition potentials observed for coumarin in water do not depend significantly on this instrumental limitation. To investigate possible ways of obtaining reliable adsorption data in high-resistance solutions, dc voltammetry at stationary electrodes was explored. Figure 2 shows dc cyclic voltammograms a t a conventional HMDE (area of 2.94 mm2) in the presence of various concentrations of electrolyte. With this technique, the potential domains of the three adsorbed phases and the foot of the faradaic wave are readily identified. Figure 3 shows the scan rate dependence with M NaF being present ( K = 36.3 pS cm-') and Figure 4 the influence of positive feedback circuitry for the same solution. The presence of uncompensated resistance in solutions having low electrolyte concentrations is revealed by a broadening of the adsorption peaks and the influence of positive feedback circuitry. When these experiments were repeated in the absence of added electrolyte ( K = 16.2 pS cm-'), similar behavior to that shown in Figure 3 was observed even when maximum positive feedback available with the PAR 174/50 was applied to the potentiostat. Therefore, dc cyclic voltammograms at a conventional HMDE provide only qualitatively correct information in dilute electrolyte solutions. The results from both the ac and dc techniques at conventional sized electrodes were found to be in qualitative agreement. In dilute or zero electrolyte, contamination via leakage from the reference electrode is a probable source of error. The experiments described in Figures 3 and 4 were therefore repeated with a platinum wire quasi-reference electrode. Within experimental error, identical cyclic voltammograms were obtained, but shifted to more negative potentials by 130 mV compared with data obtained relative to the SCE. Variable junction potential effects can also be expected with a change in the electrolyte concentration. However, the fact that the same relative effects in the measured potentials of the adsorption processes were obtained with two distinctly different reference (22) Brown, E. R.; Hung, H. L.; McCord, T. G.; Smith, D.E.; Booman, G. L. Anal. Chem. 1968, 40, 1424.
~
-1.0
-0.5
E ( V vs S C E )
'
l
~
~
~
-1.5
i v v s PRE) Figure 5. Cyclic voltammograms of saturated solutions of coumarin at 15 "C using a hemispherical mercury microelectrode (area 1.83 X mm2 for a and b; 1.63 X mm2 for c): (a) water only; (b) lo4 M NaF; (c) 0.1 M NaF. Scan rate 100 mV s-l. Scan -0.3 -1.6 -0.3 (V) vs PRE. First two (b) or three (a and c) scans reported. E
- -
electrodes suggests that this is not a problem. Furthermore, the shifts in the potentials of the transitions at the different electrolyte concentrations relative to the ferrocene/ferrocinium redox couplez3were found to be similar to those observed with the SCE and the PRE. The above data demonstrate that conventional sized mercury electrodes in either dropping or stationary formats are unsuitable for quantitative adsorption studies in the absence of added electrolyte. In view of the success of microelectrodes for studying faradaic processes in the absence of added electrolyte, the use of these electrodes for studying adsorption processes was investigated. Adsorption currents are much smaller than faradaic currents, so that instrumental problems in measuring pA or fA currents need to be kept in mind. As a first step toward exploring the use of microelectrodes in the study of adsorption processes, measurements at a small HMDE (area 0.19 mm2) were undertaken. In a study of condensed thymine films at mercury electrodes in 1 M NaC1, de L e ~ i obtained e~~ excellent agreement between the double-layer capacitances measured with small mercury electrodes (area 0.04 mmz) and those measured with a conventional HMDq (area 3 mmz). In the absence of added electrolyte, the ac measurementswith an electrode of area 0.19 mm2 were not significantly improved. However, sharper adsorption peaks were produced in the cyclic voltammograms, suggesting that a minor improvement in the minimization of IR drop effects had been achieved. The use of vastly smaller mercury-coated platinum mm2 microelectrodes with areas of (1.63-1.83) X provided substantial improvement. The 1000-fold decrease in area leads to the high-quality dc cyclic voltammetric data shown in Figure 5. For solutions containing 0.1, lo4, and 0 M added NaF, all regions I-IV are very clearly defined. The peaks for the transitions are now very sharp even in the absence of added electrolyte. The transitions (23) Bond, A. M.; McLennan, E. A.; Stojanovic, R. S.;Thomas, F. G. Anal. Chem. 1987, 59, 2853. (24) Sridharan, R.; de Levie, R. J. Phys. Chem. 1982, 86, 4489.
~
,
I
Adsorption Processes in the Absence of Electrolyte
h
0
- 0.5
-1.0
-1.5
E ( V V S PRE)
Figure 6. ac voltammetric (quadraturecurrent) curve for saturated coumarin in lo4 M NaF at 15 "C with a mercury microelectrode, area 1.83 X mm2: ac voltage 10 mV p-p, 41.3 Hz; dc scan rate 2 mV d.
from both phases I and I11 to the compact quasi-solid phase I1 show evidence of the stochastic nature of the process even in the absence of electrolyte. That is, the potentials of these transitions vary randomly on successive cycles as shown in Figure 5. This phenomenon is more evident in cyclic voltammograms recorded with microelectrodes than with conventional electrodes where IR drop broadening tends to mask this behavior. There is little evidence of uncompensated resistance in the 10-pA peaks obtained with the microelectrode. The use of a microelectrode also enables ac voltammetric data to be obtained in dilute electrolyte. Figure 6 is the response obtained in lo4 M NaF. The correct features are observed, and the double-layer capacitance is of the magnitude predicted via extrapolation of data obtained at a DME in 0.9 M NaF if one allows for the difference in electrode area. Data obtained at the microelectrode were obtained with a two-electrode cell without positive feedback IR correction. Alternating current measurements obtained with a three-electrode potentiostat and a conventional HMDE are almost featureless in the absence of electrolyte even with positive feedback.
Conclusions Microelectrodes have been shown to be more suitable than conventional electrodes for studying adsorption phenomena in the absence of added electrolyte. In the case of the neutral adsorbate, coumarin, the data show that all three adsorbed phases still prevail in the absence of added
Langmuir, Vol. 4 , No. 2, 1988 345 electrolyte. Shifts are observed in the potentials of the transition between phases I and I11 and the condensed phase. These shifts are consistent with an increase in the size of the diffuse double layer at low electrolyte concentrations. The stochastic nature of the processes is also revealed even without added electrolyte. In the case where charged species are adsorbed and more complex mechanisms p r e ~ a i l the ~ ~ data l ~ ~ in the absence of electrolyte could be distinctly different to those in the presence of electrolyte. The data obtained in the absence of electrolyte, as noted in the Introduction, also will be more relevant to data obtained by nonelectrochemical methods, where the presence of electrolyte has never been regarded as mandatory. The present work demonstrates that quantitative double-layer measurements are possible in aqueous media at microelectrodes in the absence of added electrolyte. However, the measurements of pA currents or pF capacitances in the presence of very high resistance still present substantial instrumental problems that need to be addressed. In the present work, a picommeter has been used on its own for dc work and as a preamplifier for ac measurements. The adsorption processes are only just observable with saturated solutions of coumarin when a 5-pm electrode and the present instrumentation are used. Whilst it is desirable to use even smaller electrodes to further minimize the IR drop effects, these would require even more sensitive instrumentation. The long response time of the picommeter (ca. 10 ms) also restricts the dc scan rate and/or ac frequeecy that can be used to low values. The dc data are as good as can be expected from cyclic voltammetry at low scan rates. The ac data are more valuable for studying adsorption phenomena. This will require improved instrumentation compared to that used in the present study. Subsequent to the above studies being completed, Faulkner et al.n have published a circuit which can be used with microelectrodes for short-timedomain work. Additionally, Cammann and colleagues28 have demonstrated the feasibility of measuring pF capacitances in the presence of up to lOO-MQresistances. Both these instrumental developments could be usefully incorporated into instrumentation for studying adsorption behavior in the absence of electrolyte. Registry No. Hg, 7489-97-6;Pt, 7440-06-4;NaF, 7681-49-4; coumarin, 91-64-5. (25) Laviron, E. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1982; Vol. 12, pp 53-157. (26) Murray, R. W.In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1984; Vol. 13, pp 191-368. (27) Huang, H.-J.;He, P.; Faulkner, L. R. Anal. Chem. 1986,58,2889. (28) Xie, S.-L.; Obert, B.; Cammann, K. Anal. Chem. 1986,58,1506.