Anal. Chem. 2002, 74, 1294-1301
Hot Microelectrodes Andrzej S. Baranski*
Department of Chemistry, University of Saskatchewan, 110 Science Place, Saskatoon, Saskatchewan, Canada, S7N 5C9
Heat generation at disk microelectrodes by a highamplitude (few volt) and high-frequency (0.1-2 GHz) alternating voltage is described. This method allows changing electrode temperature very rapidly and maintaining it well above the boiling point of solution for a very long time without any indication of boiling. The size of the hot zone in solution is determined by the radius of the electrode. There is no obvious limit in regard to the electrode size, so theoretically, by this method, it should be possible to create hot spots that are much smaller than those created with laser beams. That could lead to potential applications in medicine and biology. The heatgenerating waveform does not electrically interfere with normal electroanalytical measurements. The noise level at hot microelectrodes is only slightly higher, as compared to normal microelectodes, but diffusion-controlled currents at hot microelectrodes may be up to 7 times higher, and an enhancement of kinetically controlled currents may be even larger. Hot microelectrodes can be used for end-column detection in capillary electrophoresis and for in-line or in vivo analyses. Temperature gradients at hot microelectrodes may exceed 1.5 × 105 K/cm, which makes them useful in studies of Soret diffusion and thermoelectric phenomena. Electrically heated microelectrodes were introduced and developed as analytical tools by Grundler.1 In his original design, two ends of a cylindrical microwire electrode (typically 25 µm in diameter) were connected to an alternating current power source. AC current passing through the wire allowed for uniform and wellcontrolled heating of an electrode surrounded by an electrolytic solution. At the same time, the electrode was also connected to a potentiostat, and standard electrochemical measurements were carried out at the hot microwire. However, to avoid distortion of the analytical signal, a high-frequency (∼100 kHz) current generator had to be used. Further reduction of the electrical noise associated with the ohmic polarization of the microwire was achieved by connecting the microwire to the potentiostat in the middle point between two power line contacts. It was possible to maintain the surface temperature of the hot-wire electrode below the boiling point of the solution for an indefinite time or well above the boiling point for a short time2 (e.q., in aqueous solutions, a surface temperature of 250 °C was maintained for ∼5 ms without boiling the solution). However, to provide a large temperature change, high currents up to 6 A had to be used.3 It was shown * E-mail:
[email protected]. (1) Gru ¨ ndler, P. Fresenius’ J. Anal. Chem. 1998, 362, 180-183. (2) Gru ¨ ndler, P.; Degenring, D. Electroanalysis 2001, 13, 755-759.
1294 Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
that hot-wire electrodes offer very significant improvement in the mass transport rates of analytes as the result of a drop in the solution viscosity in the diffusion layer and some contribution from convection.4,5 Dramatic improvements in the rates of kinetically controlled electrode processes were also reported.6 Later, Grundler and co-workers developed a sensor with auxiliary, reference, and working electrodes that were made of thin metallic layers screen-printed on low-temperature cofired ceramics.7 The working electrode was made of gold and had an arc shape (nearly full circle) 4.4 mm in diameter and 0.13 mm in width. Such a sensor was more robust and much more convenient to use than a cylindrical microelectrode, but it had a relatively large size and could not be polished more than 2-3 times without damaging the electrodes. The surface of microelectrodes can also be heated by a laser beam8 or focused microwave radation.9 However, these methods require cumbersome experimental setups, and they have morelimited analytical applications. In this paper, a new approach to making hot microelectrodes will be presented. Here, an excitation waveform of low power but very high frequency (100 MHz to 2 GHz) is used. The heat is produced in the solution adjacent to the electrode (but not in the electrode material itself). This allows raising surface temperature of ordinary disk microelectrodes (which can be prepared and polished in the usual ways). All connections to the working electrode are made through a single lead. The size of the hot zone in solution is determined by the radius of the electrode and can be 15 min) without any indication of bubble formation. EXPERIMENTAL SECTION Reagents. All solutions were prepared in double-distilled deionized water (Corning Mega-Pure System, MP-6A and D-2) using ACS grade chemicals. Ru(NH3)6Cl3 was purchased from Alfa Aesar. All electrochemical measurements were carried out without removal of dissolved oxygen. (3) Gru ¨ ndler, P. Fresenius’ J. Anal. Chem. 2000, 367, 324-328. (4) Beckmann, A.; Schneider, A.; Gru ¨ ndler, P. Electrochem. Commun. 1999, 1, 46-49. (5) Flechsig, G.-U.; Korbut, O.; Gru ¨ ndler, P. Electroanalysis 2001, 13, 786788. (6) Gru ¨ ndler P. Fresenius’ J. Anal. Chem. 2000, 367, 324-328. (7) Voβ, T.; Kirbs, A.; Gru ¨ ndler, P. Fresenius’ J. Anal. Chem. 2000, 367, 320323. (8) Chen, P. H.; McCreey, R. L. Anal. Chem. 1996, 68, 3958-3965. (9) Marken, F.; Matthews, S. L.; Compton, R. G.; Coles, B. A. Electroanalysis 2000, 12, 267-273. 10.1021/ac015659h CCC: $22.00
© 2002 American Chemical Society Published on Web 02/07/2002
Figure 1. Electronic setup used in all experiments.
Electrochemical Cell. A standard electrochemical cell equipped with a water jacket was used. The temperature of the solution was controlled by means of a circulating bath (Haake FE2). The temperature of the electrode surface was determined from the Nernst equation (for details, see eq 2 in the next section). The auxiliary electrode was made of a platinum plate ∼1 cm2 in surface area. The reference electrode was Ag/AgCl in ′saturated KCl. Working electrodes were prepared by sealing Pt or Au wires 12.5 µm in radius (Goodfellow Metals Ltd.) into Corning Kovar sealing glass tubing no. 7052 (World Precision Instruments). A lead was made by connecting copper wire with microwires with the aid of silver conductive adhesive paste (Alfa Aesar). Then the electrode tip was cut, and the electrode was polished with 400and 600-grade carborundum paper. The final, mirrorlike polishing was accomplished using 3- and 0.3-µm aluminum oxide finishing films (TrueView Products Inc.). Electronic Circuit. The electronic setup used in this work was originally designed for carrying out faradaic rectification10 experiments with ultramicroelectrodes. The diagram is shown in Figure 1. The high-frequency sinusoidal waveform was produced by a Hewlett-Packard Signal Generator (model HP8648B) equipped with a high power option (1EA). This generator could produce a signal with frequency ranging from 10 kHz to 2 GHz and voltage amplitude up to 2.6 Vrms (that limit actually varied somewhat with frequency). This waveform (later called the excitation signal) was applied to an AC input of the potentiostat marked as EAC in Figure 1. The connection between the HP8648B and the potentiostat was made with a short (30 cm) coaxial cable terminated with a 50 Ω resistor (RT). The main pathway for the excitation signal was from input EAC to capacitor C1, the working ultramicroelectrode, the solution, the auxiliary electrode, and capacitor C2 to ground. Other pathways were limited by inductors and resistors in filters F1, F2, and F3. (10) Agarwal, H. P. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1974; Vol. 7, pp 161-211.
The circuit also included a standard current follower (OPA1) and a potentiostat (OPA2). The potentiostat bandwidth was limited to ∼1000 Hz by filters F2 and F3. The scanning waveform was applied to input E(t). The current follower included a variable feedback resistor, Rf (10 kΩ to 1 GΩ) and a bandwidth limiting capacitor Cf. The role of filter F1 was to provide high-input impedance for high-frequency currents and to prevent highfrequency currents from passing to the current follower. The center frequency for this filter was ∼20 kHz. The current transfer ratio (Iout/Iin) was 0.98-0.2j at 1000 Hz (where j ) x-1) and (6.6-1.8j) × 10-5 at 100 kHz; the transfer ratio at higher frequencies was negligibly small. All capacitors shown in Figure 1 were high-quality mica capacitors with very low frequency dispersion. Measurements were carried out with a custom design data acquisition system based on a microcontroller (Microchip PIC16F877) that was interfaced with a computer via serial link. The scanning waveform was generated by a 16-bit digital-to-analog converter (Linear LTC1655). The potential steps in the scanning waveform were always 0 (curve c in Figure 10), and consequently, the steady-state limiting current increases. The opposite is true for D′i < 0 (curve d in Figure 10). Results presented in Figure 10 may be useful in the qualitative explanation of current overshoots observed during switching the excitation waveform “on” and “off ” (see Figure 6). When heating is switched on, the temperature near the electrode surface rises rapidly, but the concentration profile initially remains the same as it was at the low temperature (i.e., very steep). Due to the combination of low viscosity and a high concentration gradient, the current is larger (in absolute terms) than it should be under new steady-state conditions. After some time, a new, more blunted concentration profile is established, and the current drops. This process takes a relatively long time, because the depletion layer is relatively thick. When heating is switched off, the situation is the opposite: the temperature near the electrode drops and viscosity rises, but the concentration gradient is initially the same as it was at the high temperature. Consequently, the current falls for a few seconds below the steady-state value. Conclusions arising from theoretical considerations presented in this section may be applicable to disk electrodes, because from the point of view of transport characteristics, a disk electrode resembles a hemispherical electrode with a radius π/2 times smaller.20 On the basis of this analogy, it can be shown that eqs 15 and 16 are valid for spherical, hemispherical, and disk electrodes. The relationship between ∆T and ∆V predicted by eq 16 is shown in Figure 5 (dashed line). Calculations were carried out with F ) 2.4 Ωcm (measured value for 2.5 M (NH4)2SO4) and κ ) 5.64 × 10-3 W cm-1 K-1 (same as the thermal conductivity of pure water.17) Calculated values are in surprisingly good agreement with experimental observations; however, eq 16 contains two major oversimplifications. First of all, a change of the solution resistance with temperature was neglected: when the temperature of the solution increases, the solution resistance decreases, and Joule heating, given by ∆V 2/Rs, increases. Omitting this fact should lead to a large negative error. On the other hand, at disk electrodes, heat generation occurs only in the solution, but heat dissipation takes place in both the solution and an insulator surrounding the electrode. The most common insulating material (glass) has a higher thermal conductivity than water; therefore, (20) Oldham, K. B. J. Electroanal. Chem. 1981, 122, 1-17.
the heat transfer through glass is very significant. In addition, heat transfer through the microwire cannot be totally ignored. This extra cooling through the electrode body was neglected in the derivation of eq 16, and that should lead to a large positive error. It is possible that by lucky coincidence, these two types of errors compensate each other. Soon, an attempt will be made to develop a more rigorous mathematical model for disk electrodes. CONCLUSIONS Hot-disk microelectrodes are smaller and more convenient to use than hot-wire electrodes. Hot-disk microelectrodes could be used in systems in which application of hot-wire electrodes is not possible (e.g., electrochemical detection in capillary electrophoresis or in vivo analysis). Another very important advantage of hotdisk microelectrodes is the possibility of maintaining a very high surface temperature (exceeding the boiling point of the solution by more than 100 °C) for a very long time without boiling the solution. Consequently, it is possible to obtain very significant improvements of analytical signals and signal-to-noise ratios for both diffusion-controlled and kinetically controlled processes. The size of the hot zone is determined by the size of the electrode. There is no obvious limit in regard to the electrode size; however, smaller electrodes will require larger excitation frequencies to overcome larger capacitive reactants of the double layer. In principle, by this method it should be possible to create hot spots that are much smaller than those created with laser beams. That could lead to potential applications in medicine and biology. Other important applications may include studies of the kinetics and mechanisms of electrode processes at temperatures well above the boiling points of solutions. Estimated temperature gradients in experiments presented in this work were approaching 1.5 × 105 K/cm; with the use of smaller electrodes, obtaining much higher temperature gradients should be possible. These extremely high temperature gradients may be very useful in studies of Soret diffusion and thermoelectric phenomena in electrolytic solutions. Hot disk microelectrodes have one disadvantage in comparison with hot-wire electrodes. In the former case, the temperature change is inversely proportional to the specific resistance of the electrolyte, whereas in the latter case, the solution resistance is irrelevant. Simple calculations indicate that in order to work with hot-disk electrodes in 0.01 M electrolytes, excitation amplitudes approaching 20-30 Vrms are needed. Fortunately, power requirements are very low (only ∼20 mW is needed to raise the temperature of a 12.5-µm disk electrode by 200 °C); therefore, the required high amplitudes can be easily obtained by means of small and inexpensive high-frequency transformers. ACKNOWLEDGMENT The financial support of the Natural Sciences and Engineering Research Council, Canada (NSERC), through the research grant is gratefully acknowledged.
Received for review October 28, 2001. Accepted January 15, 2002. AC015659H Analytical Chemistry, Vol. 74, No. 6, March 15, 2002
1301
