Comparative study of a thin-layer cell for analysis of stationary and streaming solutions

0039-9140/88$3.00+ 0.00 Copyright 0 1988Pergamon Press plc

Tahta, Vol. 35, No. 11, pp. 855-860, 1988 Pnnted in Great Bntam All nghts reserved

COMPARATIVE STUDY OF A THIN-LAYER CELL FOR ANALYSIS OF STATIONARY AND STREAMING SOLUTIONS G. FARSANG*, T. DANKHAZI, J. GRANTH and L. DARUHAZI Institute of Inorganic and Analytical Chemistry of L. E&vi% University, Muzeum krt. 4/B, 1088 Budapest VIII, Hungary (Received

22 January

1988. Rewed

18 July 1988. Accepted

21 July 1988)

Summary-A comparative study with a home-built Kissinger-type twin-electrode thin-layer cell has been made, with the cell either filled with quiescent solutions or used as a flow-through detector. In both modes the cell was polarized by a slow linear potential sweep. The current-voltage curves recorded showed a limiting current region in both cases, in spite of the different types of transport process (steady-state diffusion caused by the redox cycling effect of the counter-electrode in the quiescent solution and convective diffusion in the flowing solution). The sensitivities and also the lower detection limits were found to be nearly the same for the two different modes of operation of the same cell, for which an explanation is given. The main analytIca advantage of the thin-layer twin-electrode cell filled with quiescent solution is that only a few pi of analyte solution are necessary for the measurement. It was proved experimentally that this type of cell, used as a flow-through detector, gives the expected limiting current, but this current is dependent on the cube root of the linear flow-rate, in agreement with Weber’s theoretical prediction, rather than on the square root of the flow-rate (commonly quoted in the literature).

The twin-electrode thin-layer cell is a voltammetric signal transducer which is a popular flow-through detector in high-performance liquid chromatography,‘,* flow-injection analysis (FIA)3and concentration monitoring techniques.4 The micrometer-type twin-electrode thin-layer cell was introduced by Reilley and co-workers.%’ The cell volume was adjusted by means of a micrometer screw and the electrodes were polarized by a bipotentiostat, which enabled them to be polarized independently to the desired potentials relative to a reference electrode connected to the cell by a Luggin capillary. The theory of this type of cell shows that if the solution film contains one form of a reversible system, then a steady-state limiting current region is developed on the current-potential curve, although both the electrodes and the solution layer are stationary during the recording of the polarization curve. According to this treatment, the steady-state limiting current is proportional to the concentration. This concept was proposed for precise measurement of diffusion coefficients and electrode reaction kinetic parameters. A twin-electrode thin-layer cell with extremely large glassy-carbon electrodes with an area of several cm’ was suggested as a coulometric detector for HPLC by Lankelma and Poppe.8.9 The first modern twin-electrode thin-layer cells having identical and opposing glassy-carbon electrodes of small area (a few mm*), with the possibility of iR compensation, were made by Blank” and by Brunt and Bruins.” Kissinger et .1.‘2.‘3 constructed the first such cell ‘Author

to whom correspondence

should be addressed.

suitable for commercial production (Bioanalytical Systems Inc). This type of cell required not a bipotentiostat, but a simple and inexpensive potentiostat with a built-in current-voltage converter, and its application spread extremely rapidly. One of the twin electrodes is kept at a constant potential against the reference electrode and the potential of the other (counter) electrode is polarized to the potential at which the electrode reaction of the solvent or base-electrolyte component takes place. The aims of the present paper are as follows. (i) To demonstrate the applicability of a homebuilt Kissinger-type thin-layer twin-electrode cell for the analysis of a few ~1 of analyte by linear sweep voltammetry. (ii) To compare the analytical parameters of the cell when used with quiescent and streaming analyte solutions. (iii) To prove experimentally that with the cell geometry used, the current is proportional to the cube root of the linear streaming velocity of the solution, in accordance with Weber’s theoretical prediction,14 as opposed to the generally accepted square root relationship* based on the treatment by Blaedel et al. for streaming in a tubular electrode.‘5,‘6 EXPERIMENTAL Apparatus

The cell used was a home-built Kissinger-type twinelectrode thin-layer cell slightly modified as follows. The cell body was built from two “Plexiglas” rectangles. Tokai glassy-carbon plates, 5 x 7 mm, were embedded in the cell

856

G. FARSANGet al.

body with epoxy resin and positioned exactly opposite each other. In the upper half of the cell body a hole was drilled for connection of the salt bridge solution of the reference electrode to the cell solution by means of a porous Vycor”’ rod. The reference electrode was a chloride-coated silver wire immersed in I M sodium chloride saturated with silver chloride. The stainless-steel inlet and outlet tubes were led through two “Plexiglas” screws positioned at equal distances from the centre of the working electrode, which was always the electrode in the lower cell body. These connections were made leak-proof by use of PTFE ferrules. The two surfaces of the cell body with the embedded electrodes were polished to a mirror finish, the final polishing being done with Buehler alumina, of 0.01 pm average grain-size, suspended in doubly distilled water. A 65 pm thick Teflon spacer with a 6.0 x 30 mm channel cut in it was positioned between the two halves of the cell and the whole sandwich was held together with stainless-steel screws at the four comers. In some cases, to decrease the volume of the cell, a 25 pm thick poly(vinyl chloride) spacer was used, with a channel of the same size as before. For measurements of quiescent solutions, the cells were flushed and then filled with the analyte solution by means of an Agla micrometer syringe with its glass needle fitted to the inlet tube through a very short polyethylene tube (bore 0.25 mm). Both the inlet and outlet tubes were then closed with polyethylene caps. For measurements of streaming solutions an MMG AM 0885 pneumatic low-pressure control valve fed high-purity (99.98%) nitrogen into a stainless-steel buffer chamber, to give a pressure which could be varied precisely and continuously in the range O-2.5 bar. On the lid of the chamber up to three liquid-stream connections were mounted with soft copper and rubber O-ring gaskets. In this way the liquid pumping lines could be used simultaneously, the main advantage being fine control of flow-velocity and pulsationfree liquid stream(s) in the streaming solution line(s). The carrier solutions were contained in glass beakers in the pressurized chamber, each eqmpped with a polyethylene tube with one end near the bottom of the solution, and the other end attached to the stream connection in the lid of the vessel. The flow apparatus and cell are shown in Fig. 1. A Radelkis OH-105 type Universal Polarograph was used to polarize the cell and record the signal. Either am-

perometric (E, = constant) or potentiodynamic

[i =f(E,)]

curves were recorded (E, = working electrode potential). In

each case, a three-electrode arrangement was used. The instrument had a built-in potentiostat providing k50 V output voltage for compensating the iR drop. A further instrument used was a Radelkis OP 208/l precision digital pH-mV meter, with a Metrohm EA-120X combined glass electrode, to check the pH of the base electrolytes used. Reagents Chlorpromazine (IO-(3’-dimethylaminopropyl)-2-chlorophenothiazine hydrochloride], EGIS Pharmaceutical Factory, Pharmacopeia Hungarica VII quality, dissolved in 0.01 M hydrochloric acid; Variamine Blue. HCI, 4-amino4’-methoxydiphenylamine hydrochloride, freshly dissolved in 0.2M pH 4.6 acetic acid/sodium acetate buffer; dopa

(3,4_dihydroxyphenylalanine), EGIS, Pharmacopeia Hungarica VII quality, dissolved in O.lM phosphoric acid adjusted with disodium hydrogen phosphate to pH 2.40. Unless otherwise stated, all chemicals were of analytical reagent grade (Reanal) and used without further purification.

RESULTS AND DISCUSSION Voltammetric measurements quiescent solutions

with

cells filled

with

One of the most significant characteristics of the measuring cell is the concentration-dependence of the signal, which was determined in the 10-7-10-3M analyte range. With a thin quiescent layer of solution between the twin electrodes, the product formed at the working electrode is converted back into its original oxidation state at the counter-electrode if the system being investigated is electrochemically reversible. During polarization, the diffusion profiles formed will be similar to those in Anderson and Reilley’s special thin-layer ~ell.~~’ It is expected, therefore, that in spite of the solution being quiescent, a limiting current region will be found on the i =f(E,)

Fig. 1. Stabilized low-pressure pneumatic-flow apparatus for analysis of streaming soluttons by methods sensitive to fluctuations of the flow-rate, with a voltammetric detector. The system shown includes a sample injector for FIA but results obtained with this method are not given in this paper. 1, N, gas cylinder; 2, pressure control valve; 3, MMG AM (Hungary) 0885 pneumatic pressure controller; 4, buffer vessel for pressure stabilization within *0.5%; 5, the solution to be pumped out from the glass beaker through a 0.25-mm bore soft silicone-rubber tube connected to a stainless-steel tube sealed into the hd of the buffer vessel; 6, a similar silicone rubber tube connected to the outlet of the stainless-steel tube in the lid; 7, Labor MIM (Hungary) OE 320 HPLC injector in “load” position; 8, in FIA operation the filling connection of the injector loop; 9, waste receiver vessel; 10, porous frit for mixing the carrier and Injected solutions; I 1, cross section of the thin-layer cell, where W is the working electrode, C the counter-electrode and R the reference electrode; 12, outlet tube of the cell; 13, Radelkis OH-105 Universal Polarograph.

Analysis of stationary and streaming solutions

4

b

0

3

1

Fig. 2. Voltammetric i =f(E,) curves recorded with the cell filled with quiescent solution. Depolarizer, I x IO-‘M dopa in pH-2.64 phosphate buffer; d = 65 pm (Teflon spacer), iL, the stationary limiting current; ip, anodic peak current. The solid line shows the curve recorded with the redox-cycling counter-electrode and the broken line the curve when the outlet tube of the cell was used as the counter-electrode.

curve, and be completely different from the symmetrical peak-shaped i-E,,, curves of classical thinlayer cells, where the counter-electrodes do not reduce the product. The rate of polarization, u, must be slow enough to allow the development of stationary diffusion profiles between the two electrodes. A further condition needed for this behaviour is electrochemical reversibility of the system under study. Figure 2 shows cyclic voltammetric curves for dopa, recorded in the twin-electrode cell in two ways: (a) with the counter-electrode opposite the working electrode to act as a redox-cycling electrode and (b) with the stainless-steel outlet tube as the counterelectrode, and the glassy-carbon electrode opposite the working electrode acting only as a diffusionlimiting barrier. Hysteresis was found on the reverse sweep, which can be explained by the electrode reaction mechanism of dopa” and its tendency to slight filming on the electrodes. The effect of the reducing counter-electrode is quite unambiguous when the cyclic voltammetric curve is compared with that obtained with the outlet tube as counterelectrode, where the irreversible shape is explained by

857

the presence of a coupled chemical reaction in the electrode reaction of dopa.” Similar measurements were made with two other reversible redox systems with well known electrode reaction mechanisms, Variamine Blue” and chlorpromazine.‘9*” The shape of the signal in each case. showed a limiting current region when the counter-electrode worked as a redoxcycling electrode. The current was a linear function of the concentration in the range 1 x 10-‘-l x 10e4M. The coefficients of the linear calibration curves iL = mC + b measured in the I x 10m6-1 x 10W4Mare summarized in Table 1. From the results of Table 1 and experimental work done in the 1 x lo-‘1 x 10m6M concentration range, the following analytical consequences can be inferred. (a) The thin-layer cell with a redox-cycling counter-electrode and a quiescent analyte layer can be used as a sensitive electrochemical detector (see the slopes obtained). (b) The sensitivity and the lower detection limit depend on the thickness of the spacer used; with the 25 pm spacer the detection limit is found to be approximately 1 x lo-‘M. In the micromolar concentration range, the residual current is reproducible and can be easily compensated. To a first approximation, the relationship derived by Anderson and Reilley’ for the limiting current can be used: nFAD, CT

i, =

(1)

d

where A is the surface area of the working electrode, C, the sum of the concentrations of the oxidized and reduced forms in the diffusion layer, d the thickness of the solution film, n and F have their usual meanings, and D,, is given by D

=

2DJ’r

(2)

ss D,+D,

where D, and D, are the diffusion coefficients of the oxidized and reduced forms of the reversible redox system. For n = 2, F = 96485 C/eq, A = 0.350 cm2, D,=D,=l~lO-~cm~/sec,C,=10-‘Mandd=25 pm, the value of iL calculated by use of (1) is 27 nA at the limit of detection.

Table I. The coefficients of the analytical calibration curves for three reversible systems measured with redox-cycling counter-electrodes at two spacer thicknesses* Compound measured L-Dopa L-Dopa Chlorpromazine Chlorpromazine Variamine Blue

Base electrolyte 0.2M phosphate buffer, pH 2.64 0.2M phosphate buffer, pH 2.64 IO-‘M HCI IO-*M HCI 0.2M acetate buffer, pH 4.60

Concentration range, M

Slope (m), mA.I.mole-‘.mm-2

Intercept (b), nA /mm’

Correlation coefficient

d, p

I x 10-6-I x IO-4

21.2

2.88

0.9956

65

I x 10-6-I x IO-’

37.8

6.77

0.9868

25

0.780 2.35 0.642

0.9989 0.9979 0.9998

65 25 65

I x 10-6-I x IO-4 I x 10-6-I x 10-4 I x 10-6-I x IO-4

0.878 2.24 1.58

*The d values are only approximate: the thickness of the solution layer is always smaller because of deformation of the spacers under pressure.

G.

858

FARSANG et

al.

Chlorpromazinez2 and diazepamz3 in biological fluids have been determined by differential pulse voltammetry. (4 A pulse-free pumping system is needed only when flow analytical methods are to be used. (e) The limiting currents are additive, so this method is more convenient than simple linear-sweep voltammetric batch analysis.” Figure 3 shows that the sensitivity of the method is much higher in the redox-cycling mode than when the thin-layer cell is used with the twin electrode acting only to limit diffusion. 02468'0 Concentration

Voltammetric measurements with twin-electrode thinlayer cells with jlowing sample solutions

(PM)

Fig. 3. The sensitivity of the cell filled with quiescent solutions of dopa dissolved in phosphate buffer of pH 2.64. Curve a, redox-cycling counter-electrode; curve b, nonredox-cyciing counter-electrode.

The charging current, ic. can be calculated*’ from

where C,, is the differential capacity (F/cm*) and o the rate of polarization (Vjsec). With A as above, C, = 20 pF/cm2 and v = 8.33 mV/sec, i, = 58 nA/V. According to this the charging current interferes at the detection limit, which emphasizes the need to apply as low a polarization rate as possible during the recording of the i =f(E,) curve. In the course of the measurements there was no difficulty in achieving a detection limit of lo-‘M for all the analytes investigated, which was helped by the fact that the twin electrodes worked not in a circular film of solution (Anderson and Reilley’s construction”) but in a rectangular layer with an area larger than A. Hence besides strictly linear diffusion between the twin electrodes, there was diffusion towards the working electrode from the rest of the film. (c) The method shows outstanding characteristics as a microanalytical technique. Only a Xl-100 ~1 volume of analyte solution is needed for the measurement, including flushing the cell with new solution. This is especially advan~geous when the volume of sample for analysis is extremely small, e.g., some samples of blood and other body fluids, especially when these contain drugs or metabolites. Table 2. Analytical signal-concentration

If the cell is operated as a flow-through detector, two kinds of measurement may be made: (a) the sample ~on~ntration varies continuously in the carrier stream and can be monitored, or (6) samples are injected into the carrier stream in sequence, as in FIA, and are separately analysed. Both techniques are widely used.24 It was surprising to find that the lower detection limits were much the same whether the cell was used as a flow-through detector in streaming solutions or as a detector with quiescent solutions. Table 2 shows results obtained with the cell as a flow-through detector. Comparison of the results in Tables 1 and 2 shows that the most characteristic analytical parameters are at least of the same order of magnitude at the moderate flow-rate applied. The sensitivity values are about five times higher in flowing solutions than in quiescent solution and the correlation coefficients are also better. This can be explained by the cleaning effect of the streaming solution, which reduces the filming effect and so results in better reproducibility of the signals. No dramatic improvement of the analytical performance was observed in flowing solutions, and this can be attributed to two main factors. In agreement with other authors’ observations, when the cell is used as a flow-through detector, there is little, if any, contribution from the redox-cycling caused by the counter-electrode. The relatively small increase in the convective-diffusion limiting current compared to the redox-cycling diffusion limiting current is due to the

dependence obtained with a twin-electrode thin-layer cell as a flow-through detector with d = 65 pm Coefficient of the calibration curve

Compound measured Variamine Blue Variamine Blue Chlorpromazine

Base electrolyte (carrier stream*) 0.2M acetate buffer, pH 4.60 0.2M acetate buffer, pH 4.60 10mZ3vfHCI

*Volume flow-rate 0.8 ml/min. tE, = +0.35 V vs. Ag/AgCl. IjRate of polartzation 0.5 V/min.

Measuring mode Amperometric monitoringt Linear sweep voltammetry§ Linear sweep voltammetry@

Slope (m),

mA.I.mole-‘.mm-2 7.21

Intercept (b),

nA/mm’ 39.9

Correlation coefficient 0.9998

8.35

2.53

0.9996

4.56

8.16

0.9996

Analysis of stationary and streaming solutions Table 3. The current efficiency (4%) of a thin-layer cell in streaming solution for (a) I x IO-‘M and (b) 6 x IOd4M analyte Variamine Blue in the carrier stream

(a)

(b)

v, mllmin

q, %

v, mllmin

q, %

0.8

8.67 4.80 3.43 2.62 2.16

0.8 2.0 3.3 5.0 6.2

7.90 4.40 3.21 2.51 2.17

2.1 3.5 5.3 6.8

very poor current efficiency of the thin-layer cell as a

flow-through detector, as shown in Table 3. The current efficiency of the cell was measured with streaming Variamine Blue solutions in the 1 x 10e4-6 x 10m4M concentration range. For a given concentration of Variamine Blue at different flow-rates, potential steps of 0.35 V were applied starting at 0 V and the current was recorded as a function of time. After a wait of 2&30 set for the capacitive current to decay, the stabilized current, i, was integrated as a function of time for a given period, t: it = Q. With a known flow-rate, and concentration of Variamine Blue in the carrier stream, the current efficiency of the cell can be readily calculated for the 2-electron reaction. Dependence of the current on the linear flow-rate Kissinger-type thin -layer cell

in a

The literature differs as to the dependence of the current on the flow-rate, but it is accepted that the relationship has the following general form

where i, is the limiting current density (A/cm2), k is a constant, D the diffusion coefficient of the analyte (cm*/sec), v the kinematic viscosity (cm/set), C the analyte concentration in the flowing solution (mM),

t

859

P the linear flow velocity of the solution contacting the electrode surface (cm/set) and a an exponent depending on the geometry of the flow and governed mainly by the cell geometry. According to Levich,24.2S if the flow is not turbulent and the fluid is streaming between two parallel plates in a rectangular flow channel, then a = 0.5. The cell used in this work is stated in the literature2” to behave with ia r”’ but ia V’:3 can also be found.27 According to Weber,14 if the flow profile is completely developed in the cell, i.e., the velocity profile is parabolic in the cell channel, the current is proportional to LP”3 where L is the length of the electrode in the direction of flow, but when the flow-rate profile is unable to develop completely in the cell, then the limiting current i, is proportional to LP”2. Applying Weber’s calculations to a thin-layer cell with flow channel 0.0065 cm high and 0.6 cm wide, with a volume flow-rate of 0.017 ml/set, the distance needed for the full development of a parabolic flow profile is 7.4 pm. This means that most of the thin-layer cells used will show dependence of the limiting current on pIa. The validity of this theoretical treatment was tested for the cell used throughout this work, by measuring the limiting current dependence in the 0.8-6.5 ml/min flow-rate range (P = 3.4-27.8 cm/set) for a carrier stream containing 5 x 10 -5-6 x 10e4M Variamine Blue as depolarizer. From the limiting currents, the current densities were calculated and plotted as functions of rli3 and p112 (Fig. 4). The plots show that the cell behaves in accordance with Weber’s prediction. The erroneous V 1’2dependences reported in the literature may often be due to mechanical application of the Levich theory and the use of peristaltic pumps that do not provide flow-rate changes of wide enough range, such as that which can easily be achieved by pneumatic flow-rate control. With a narrow range of flow-rates it is easy to obtain a plot of i, us. V Ii2that has acceptable linearity. The statistical evaluation of a large number of measured flow-rate dependences shows a clear dependence on pu3

Acknowledgements-The authors express their thanks to Mr. Mihaly Fazekas for his careful work in building the pneumatic flow-rate controller system and the thin-layer cell, and also to the Institute of General and Analytical Chemistry of the Technical University of Budapest for supplying the Variamine Blue. REFERENCES 1.

I

I

I

1

2

3

V-t-1 :

I 4

I 5

)

or Y -3 to)

Fig. 4. The current density as a function of linear flow-rate in a twin-electrode thin-layer cell. Solid line, plotted against P”‘; broken line, plotted against V ‘I*. Depolarizer: 1 x 10m4MVariamine Blue in 0.2M acetate buffer, pH 4.60. E, = +0.35 V us. Ag/AgCl/IM NaCl reference electrode.

P. T. Kissinger, in Laborntory Techniques in Electroanalytical Chemistry, P. T. Kissinger and W. R. Heineman (eds.), pp. 611M30. Dekker, New York, 1984. 2. K. Stulik and V. Padkovl, C.R.C. Crir. Rev. Anal. Chem., 1984, 14, 291. 3. D. C. Johnson, S. G. Weber, A. M. Bond, R. M. Wightman, R. E. Shoup and I. S. Krull, Anal. Chim. Acra, 1986, 180, 187. 4. E. Pungor, Zs. Feher and G. Nagy, Pure Appl. Chem., 1975, 44, 595.

G. FARSANG et al.

860

5. L. B. Anderson and C. N. Reilley, J. Elecfroanal. Chern., 1965, 10, 295. 6. Idem, ibid., 1965, 10, 538. 7. L. B. Anderson, B. McDuffie and C. N. Reilley, ibid., 1966, 12, 477. 8. J. Lankelma and H. Poppe, J. Chromatog., 1976, 125, 375. 9. J. Lankelma, Ph.D. Thesis, University of Amsterdam,

1976. 10. C. L. Blank, J. Chromarog., 1976, 117, 35. II. K. Brunt and C. H. P. Bruins, ibid., 1978, 161, 310. 12. D. A. Roston and P. T. Kissinger, Anal. Chem., 1981,

53, 1695. 13. D. A. Roston, R. E. Shoup and P. T. Kissinger, ibid., 1982, 54, 1417A. 14. S. G. Weber, J. Elecrroanal. Chem., 1983, 145, 1. 15. W. J. Blaedel, C. L. Olson and L. R. Sharma, Anal.

Chem., 1963, 35, 2100. 16. W. J. Blaedel and L. N. Klatt, ibid., 1966, 38, 879. 17. L. H. Piette, P. Ludwig and R. N. Adams, ibid., 1962, 34, 916.

18. G. Farsang, L. Adorjan and L. Ladanyi, Proc. Zhd Co@ Appt. Phys. Chem., Vol. 1, p. 461. Akademiai Kiado, Budapest, 1971. 19. P. Kabasakalian and J. McGlotten, Anal. Chem., 1959. 31, 431. 20. M. D. Hawley, S. W. Tatwawadi, S. Piekarski and R. N. Adams, J. Am. Chem. SOL, 1967, 89, 447. 21. A. M. Bond, Modern Polarographic Methods in Analytical Chemistry, pp. 197-209. Dekker, New York, 1980. 22. T. W. Jarbawi, W. R. Heineman and G. J. Patriarche, Anal. Chim. Acta, 1981, 126, 57. 23. T. P. DeAngehs, R. E. Bond, E. E. Brooks and W. R. Heineman, Anal. Chem., 1977, 49, 1792. 24. E. Pungor, Zs. Feher and M. Varadi, C.R.C. Crir. Rev. Anal. Chem., 1980, 9, 97. 25. V. G. Levich, Physicochemical

Hydrodynamics, Chapters 2 and 12, Prentice-Hall, Englewood Cliffs, 1962. 26. K. Brunt and C. H. P. Bruins, J. Chromatog., 1979. 172,

37. 27. Y. Hirata, P. T. Lin, M. Novotny and R. M. Wightman, ibid., 1980, 181, 287.