Anal. Chem. 2000, 72, 187-192
Ceramic-Based Multisite Microelectrodes for Electrochemical Recordings Jason J. Burmeister,† Karen Moxon,‡ and Greg A. Gerhardt*,†
Departments of Anatomy and Neurobiology, and Neurolology, and the Center for Sensor Technology, University of Kentucky, Chandler Medical Center, 800 Rose Street, Lexington, Kentucky 40536-0098, and Department of Neurobiology and Anatomy, MCP Hahnemann University School of Medicine, Philadelphia, Pennsylvania 19129
This paper describes the development and characterization of ceramic-based multisite arrays for electrochemical recordings in biological systems. These electrodes represent a parallel technology to the design of microelectrodes using silicon substrates. The ceramic substrates are stronger than silicon and are nonconducting, which makes them better suited for in vivo electrochemical measurements. The current designs are based on formation of four-site (50 × 50 µm with 200 µm spacing) electrodes on ceramic wafers using photolithography. The recording sites and connecting lines are made of Pt with a polyimide coating to insulate the connecting lines. The resulting electrodes are cut from the wafers producing a 1 cm length microelectrode that tapers to a ∼2-5 µm tip. Electrochemical measures of dopamine and hydrogen peroxide support that the sensitivity, selectivity, and response characteristics of the electrodes exceed those of previously published silicon substrate-based microelectrodes. This is the first demonstration of microarrays formed from ceramic substrates, and the data presented support the hypothesis that these microelectrodes may be useful for a variety of neurochemical and electrophysiological applications. Preliminary in vivo electrochemical recordings are presented.
and available multisite electrode that would give all laboratories similar neurochemical and electrophysiological recording capabilities. There have been incremental improvements in the performance characteristics of silicon-based multisite microelectrodes.5-15 Using silicon-based electrodes supplied by the Center for Integrated Sensors and Circuits, at the University of Michigan, Ann Arbor, we previously investigated the properties of five-channel microelectrodes sputter-coated with carbon and then dip-coated with Nafion to increase the selectivity of the recording sites for cationic neurotransmitters.16,17 High-speed electrochemical recordings were performed in vitro and in vivo which demonstrated that these multisite microelectrodes were capable of monitoring the evoked overflow of monoamines in selected brain regions of the rat. In addition, when the microelectrodes were coated with Nafion, action potentials from Purkinje cells in the rat cerebellum, identified electrophysiologically, were recorded from different sites on the same microelectrode. More recently we developed a modified silicon substrate-based multisite array in an attempt to further improve the semiconductor-based electrodes for in vivo electrochemical and electrophysiological recordings.18 These microelectrodes have improved insulation by using silicon nitride. However, our prior studies have shown that the silicon-based multisite microelectrodes have decreased selectivity for dopamine
Over the past 20 years, several research groups have been developing semiconductor-based, multisite microelectrodes.1-4 The potential advantages for many scientists in chemistry, pharmacology, neuroscience, physiology, and cell biology to use such electrodes include the following: (1) reproducible recording sites, (2) the ability to record from potentially 4-16 neurons or brain sites from a single microelectrode, (3) spatially defined recording sites for recording from layered brain structures, (4) the ability to stimulate and record single unit or chemical activity from the same recording sites, and (5) production of an affordable
(5) BeMent, S. L.; Wise, K. D.; Anderson, D. J.; Najafi, K.; Drake, K. L. IEEE Trans. Biomed. Eng. 1986, 33 (2), 230-241. (6) Najafi, K.; Ji, J.; Wise, K. D. IEEE Trans. Biomed. Eng. 1990, 37 (1), 1-11. (7) Najafi, K.; Hetke, J. F. IEEE Trans. Biomed. Eng. 1990, 37 (5), 474-481. (8) Ji., J.; Najafi, K.; Wise, K. D. IEEE Trans. Biomed. Eng. 1991, 38 (1), 7581. (9) Carter, S. J.; Linker, C. J.; Turkle-Huslig, T.; Howard, L. L. IEEE Trans. Biomed. Eng. 1992, 39 (11), 1123-1128. (10) Ang, S. S.; Sreenivas, G.; Brown, W. D.; Naseem, H. A.; Ulrich, R. K. J. Electronic Materials 1993, 22, 344-352. (11) Hooger, A. C.; Wise, K. D. IEEE Trans. Biomed. Eng. 1994, 41 (12), 11361146. (12) Chen, J.; Wise, K. D.; Hetke, J. F.; Bledsoe, S. C., Jr. IEEE Trans. Biomed. Eng. 1997, 44 (8), 760-769. (13) Chen, J.; Wise, K. D. IEEE Trans. Biomed. Eng. 1997, 44 (8), 770-774. (14) Akin, T.; Najafi, K.; Smoke, R. H.; Bradley, R. M. IEEE Trans. Biomed. Eng. 1994, 41 (4), 305-313. (15) Bradley, R. M.; Cao, X.; Akin, T.; Najafi, K. J. Neurosci. Methods 1997, 73 (2), 177-186. (16) van Horne, C.; Hoffer, B. J.; Stromberg, I.; Gerhardt, G. A. J. Pharmacol. Exp. Therap. 1992, 263, 1285-1292. (17) Moore, P. A.; Zimmer-Faust, R. K.; BeMent, S. L.; Weissburg, M. J.; Parrish, J. M.; Gerhardt, G. A. Biol. Bull. 1992, 183, 138-142. (18) Sreenivas, G.; Ang, S. S.; Fritsch, I.; Brown, W. D.; Gerhardt, G. A.; Woodward, D. J. Anal. Chem. 1996, 68, 1858-1864.
* Corresponding author: (phone) (303) 315-5628; (fax) (303) 315-5347. † University of Kentucky. ‡ MCP Hahnemann University School of Medicine. (1) Wise, K. D.; Angell, J. B.; Starr, A. IEEE Trans. Biomed. Eng. 1970, 17 (3), 328-347. (2) Pochay, P.; Wise, K. D.; Allard, L. F.; Rutledge, L. T. IEEE Trans. Biomed. Eng. 1979, 26 (4), 199-206. (3) Prohaska, O. J.; Olcaytug, F.; Pfunder, P.; Dragaun, H. IEEE Trans. Biomed. Eng. 1986, 33 (2), 223-229. (4) Blum, N. A.; Carkhuff, B. G.; Charles, H. K.; Edwards, R. L.; Meyer, R. A. IEEE Trans. Biomed. Eng. 1991, 38 (1), 68-74. 10.1021/ac9907991 CCC: $19.00 Published on Web 12/03/1999
© 2000 American Chemical Society
Analytical Chemistry, Vol. 72, No. 1, January 1, 2000 187
over ascorbic acid, higher background currents, and poorer detection limits as compared to typical single carbon-fiber electrodes.19,20 In addition, the signal-to-noise ratios of electrophysiological recordings of cerebellar Purkinje neurons or hippocampal pyramidal neurons were poorer than those achieved by glass micropipets. Thus, the multisite arrays were still not as good as single carbon-fiber electrodes for chemical recordings and glass micropipets for electrophysiological measures. Our initial studies18 showing the high background current and poor signal-to-noise characteristics of the multisite array microelectrodes were likely due to cross talk between the recording sites and a shunt capacitance between the metallization layer and the silicon substrate.6 This shunt capacitance is due to the fact that the silicon substrate is a good conductor and must first be insulated with silicon nitride to protect it from the saline environment of the brain before the metallization step. The cross talk between the lines may be due to a similar capacitance created by the poor insulating quality of the silicon nitride. Thus, the silicon substrates themselves may not be the optimal material to use as a substrate for the fabrication of microarrays used in high-salt environments. A potential parallel technology involves the use of other electrode substrates, such as ceramic, silica, or glass. In particular, ceramic materials can be formed into thin substrates that can be used to manufacture multisite microelectrodes.21 Rather than building silicon-based microelectrodes, we propose a microelectrode made from inert ceramic. There are several potential advantages of such ceramic-based electrodes. First, they may have increased strength due to the ceramic substrate. Second, there may be decreased cross talk between recording sites due to the inert ceramic substrate. Third, there is great versatility afforded by using a substrate that can be machined or laser-cut. Fourth, the use of ceramic substrates can potentially lower the cost of developing and producing a variety of multisite microelectrodes. It has been our goal to develop a multisite microelectrode that is sufficiently inexpensive to be disposable and can be applied to in vivo electrochemical measurements of neurotransmitters over a period of a few days. The microelectrode fabrication methods and general recording characteristics of the first ceramic-based microelectrodes are the focus of this paper. EXPERIMENTAL SECTION Reagents. All chemicals were used as received unless otherwise stated. Nafion (5% in a mixture of aliphatic alcohols and water) and ascorbic acid were obtained from Aldrich. Potassium nitrate was purchased from Fluka. H2O2 (30%), dopamine, potassium ferricyanide, sodium chloride, sodium glutamate, glutaraldehyde, bovine serum albumin, dibasic sodium phosphate, and monobasic sodium phosphate were obtained from Sigma. LGlutamate oxidase was purchased from Seikagku America, Inc. All solutions were prepared using distilled water, which was deionized using a Barnstead/Thermolyne D8922 ion-exchange column. H2O2 (30%) was diluted to 8.8 mM upon receiving and stored at 4 °C. Glutaraldehyde was stored in a freezer at -4 °C. (19) Hebert, M. A.; van Horne, C. G.; Hoffer, B. J.; Gerhardt, G. A. J. Pharmacol. Exp. Ther. 1996, 279, 1181-1190. (20) Hoffman, A. F.; Lupica, C. R.; Gerhardt, G. A. J. Pharmacol. Exp. Ther. 1998, 287, 487-496. (21) Coors Ceramics Co. Thin Film Substrates Technical Specifications 10-2-0194, Golden, CO.
188 Analytical Chemistry, Vol. 72, No. 1, January 1, 2000
Figure 1. Schematic diagram illustrating the fabrication steps of the ceramic-based multisite microelectrodes. Two masks are required if the polyimide is patterned onto the microelectrodes while only one is required if the insulating layer is applied using a brush.
Electrode Fabrication. Ceramic Substrates. The basic fabrication scheme for the ceramic-based microelectrodes is seen in Figure 1. The fabrication of the ceramic wafers was carried out in conjunction with Thin Film Technology, Inc. (Buellton, CA). The 0.005 ( 0.0005 in. thick, 1 in. square Superstrate 996 ceramic substrates used for these studies were obtained from Coors Ceramics Co. (Golden, CO). Initial attempts were made to use thinner, 25-50 µm thick, ceramic substrates. However, it was discovered that the resulting substrates were very fragile and difficult to use. The 0.005 in. substrates were cleaned with a mixture of sulfuric acid and chromium trioxide and then rinsed with deionized water. Thin Film Technology complies with all OSHA requirements for safe handling of dangerous compounds. The substrates were then dried at 120 °C followed by O2 plasma ashing. Photoresist was evenly applied to the cleaned ceramic substrate using a photoresist spinner. The wafers were then covered with a custom-designed mask to pattern the recording sites, lines, and connecting pads. The resulting microelectrodes were composed of four 50 × 50 µm recording sites and connecting lines and bonding pads that extended ∼1 cm from the recording sites. The recording sites were 200 µm apart center to center. Upon collimated UV light exposure, the ceramic plates were developed to expose the bonding pads, connecting lines, and recording sites of the individual microelectrodes. The substrates were then sputter-coated with an adhesion layer of 500 Å of titanium followed by a 1500 Å thick layer of platinum. Oxide formation on the adhesion layer was prevented by not removing
Figure 2. Schematic diagram of the rectangular (A) and tapered (B) multisite ceramic-based microelectrodes. Included is a magnification of the tapered microelectrode tip (C).
the substrate from the vacuum chamber between sputter coatings. An ultrasonic acetone liftoff was used to remove the remaining masked areas from the ceramic wafer, thus exposing the recording sites, connecting lines, and bonding pads of each of the 56 microelectrodes/ceramic wafer. A 2-3 min length sputter-etching cleaning step was used to remove any remaining photoresist. The ceramic wafers were then ready to be cut in order to expose the individual microelectrodes. Individual microelectrodes were cut from the bulk ceramic wafer in rectangular sections using a saw with a diamond blade. Rectangular microelectrodes as depicted in Figure 2A were used for in vitro characterization studies. Microelectrodes for in vivo use were further cut into a tapered design using an industrial laser (see Figure 2B and C). The microelectrode surfaces excluding the recording sites and bonding pads were then coated with polyimide ( ∼10 µm) using a proprietary brush to insulate the conducting lines. All multisite arrays were then wire bonded to a printed circuit board carrier that allowed for easy handling and testing of the microelectrodes. An epoxy coating was used on the wire bonding sites to insulate these areas from the printed circuit board substrate. A photograph of one of the laser-cut multisite microelectrodes is seen in Figure 3. The printed circuit board is connected to instruments through a custom-fabricated connector (Quanteon, L. L. C., Denver, CO) for electrode characterization. The cutting and mounting of the microelectrodes were performed in conjunction with Hybrid Circuits, Inc. (Sunnyvale, CA). The anionic polymer Nafion was applied onto the electrodes to repel anions such as ascorbate.16,18,22 The electrode tips were gently dipped and swirled in the Nafion suspension (5% in aliphatic alcohols). The electrodes were then dried at 200 °C for 4 min. Nafion in conjunction with high-temperature drying has been shown to enhance selectivity for cationic neurotransmitters, such as dopamine, over anionic interferences.19,20 Electrode Testing and Characterization. (a) Cross Talk Tests. There was no electrode pretreatment prior to characteriza(22) Gerhardt, G. A.; Oke, A. F.; Nagy, G.; Moghaddam, B.; Adams, R. N. Brain Res. 1984, 290, 390-395.
Figure 3. Photograph of a multisite ceramic microelectrode tip. The transparent polyimide coating allows inspection of the recording sites and connecting lines. Calibration bar equals 200 µm.
tion. The cross talk between recording sites was measured by sequentially lowering recording sites into solution. A four-channel electrochemical recording system (IVEC-5/4; Quanteon) was used to record the background current from all four sites simultaneously at an applied potential of +0.7 V. Using a MO-10 (Narishige) microdrive, the first site was lowered into a 250 µM ascorbate solution prepared in 0.1 M phosphate-buffered saline (PBS, pH 7.4). Each additional site was then successively lowered into solution. Increases in background current observed for the sites other than the one being lowered into solution were considered to be due to electrode cross talk. Electrode impedance was measured using an Omega Tip impedance meter (World Precision Instruments, Sarasota, FL) operating at 500 Hz in 0.9% saline. (b) In Vitro Tests. In vitro testing to investigate the electrochemical behavior of these new microelectrodes was carried out using cyclic voltammetry, chronoamperometry,and constantpotential amperometry. Cyclic voltammetry studies were performed using the test molecule potassium ferricyanide (K3Fe(CN)6), prepared in potassium nitrate electrolyte (1 M). Cyclic voltammetry was performed using an EC-225 Voltammetric analyzer (IBM Instruments). Quantitative measurements of dopamine and H2O2 were performed using high-speed chronoamperometry (5 Hz) and constant-potential amperometry, respectively. Chronoamperometry and constant-voltage amperometry were performed using an IVEC-10 high-speed electrochemistry instrument (Harvard Apparatus, Holliston, MA) using FAST-12 software (Fast Analytical Sensor Technology, Quanteon, L. L. C.). Calibrations were performed in 0.1 M PBS by increasing the concentration of analyte using four additions of stock solutions. A handheld mixer was used to stir the solutions after each addition. The selectivity, sensitivity, detection limit, and linearity were calculated for the multisite microelectrodes. The selectivity of each Nafion-coated electrode was tested prior to each calibration using a 250 µM ascorbate challenge. Selectivity is denoted as the ratio of the signals of a given concentration of analyte to an equal concentration of ascorbate. High selectivity ratios (no detected change following ascorbate addition) were included in the average as 1000. The sensitivity of each electrode was described as the slope of the current versus concentration Analytical Chemistry, Vol. 72, No. 1, January 1, 2000
189
plot in units of pA/µM. The limit of detection was defined as the concentration that corresponds to a signal-to-noise level of 3. Rootmean-squared noise levels were calculated over 10 points of the baseline. The linearity of each electrode’s response was described by the Pearson correlation coefficient (R2). All errors are reported as standard error of the mean. In Vivo Testing. Male Fischer 344 rats were anesthetized and prepared for electrochemical recordings as previously described.23 A glutamate sensor was constructed based on procedures modified from Hu and Wilson.24 A 0.5 µL drop of a solution containing glutamate oxidase (2%), bovine serum albumin (2%), and glutaraldehyde (0.12%) in water was placed on a Nafion-coated microelectrode site and allowed to dry. After calibration at 37 °C for glutamate, this microelectrode was placed into the prefrontal cortex region (AP +2.7 mm, ML 1.5 mm, and DV -3.0 mm) of a male Fischer 344 rat based on the atlas of Paxinos and Watson.25 RESULTS AND DISCUSSION Basic Microelectrode Design. Parts A and B of Figure 2 are schematic diagrams of the rectangular and the tapered multisite microelectrodes, respectively. For clarity, a magnification of the tapered design microelectrode tip is depicted in Figure 2C showing the recording sites and connecting lines. The smallest features on these electrodes are the interconnect lines, which are 10 µm. However, Thin Film Technology is capable of patterning features as small as 5 µm with strict clean room procedures. Figure 3 is a photograph of the tip of a ceramic multisite microelectrode. A total of over 80 microarrays were successfully assembled using the aforementioned methods. The resulting electrodes were seen to have recording sites that were very uniform in geometric size. The manufacturing process produces microelectrodes with recording sites that are indented from the surface of the polyimide film. This microwell may aid in the adhesion of coatings or immobilized enzymes onto the electrode surfaces. Preliminary studies (see below) supported that the Nafion coatings readily adhered to these surfaces. Wire-type electrodes or single carbon-fiber-type microelectrodes required four to five dip coats of Nafion, while the ceramic-based multisite electrodes needed only one coat to achieve the same selectivity. The ceramic-based multisite electrodes have many similarities in design to previously reported silicon-based electrodes.16 Similarities include the overall tapered design, shank width, and microelectrode tip width. The surface areas of some silicon-based microelectrodes are 1000, 2000, 4000, and 8000 µm2 compared to 2500 µm2 for our ceramic-based microelectrodes. The areas of the recording sites can be adjusted for measuring compounds in various concentration ranges. The substrate thickness (125 µm) is the main difference between these and previously reported microelectrodes. In our hands, the thin silicon substrates (15 µm) with long (∼1 cm) electrode shanks were very flexible, which made them difficult to control when placing into tissues. The stiffer ceramic substrate of the present design allows for greater control when placing the multisite electrodes into tissues. Although only thicker (125 µm) substrates have been used thus far, thinner (23) Hebert, M. A.; Gerhardt, G. A. Brain Res. 1998, 797 (1), 42-54. (24) Hu, Y.; Mitchell, K. M.; Albahadily, F. N.; Michaelis, E. K.; Wilson, G. S. Brain Res. 1994, 659, 117-125. (25) Paxinos, G.; Watson, C. The Rat Brain Stereotaxic Coordinates; Academic Press: New York, 1986.
190
Analytical Chemistry, Vol. 72, No. 1, January 1, 2000
substrates (25-50 µm) are under development. Although it is generally thought that smaller electrodes cause less tissue damage, to our knowledge, there has not been a systematic study on the thickness of wafer-based electrodes and the resulting effects on tissues when placed in vivo. Tests of the Microarrays. Three different measures were used to test the electrodes: (1) cross talk analysis in 0.9% saline to determine whether the individual recording sites of the microelectrodes were insulated, (2) cyclic voltammetry, and (3) calibrations using amperometry and chronoamperometry to study the electrochemical characteristics of the microelectrodes. A major issue of concern with the microelectrodes is potential electrical cross talk between adjacent recording sites and adjacent metal conducting lines. In cell-containing media or in brain tissue, the width of spacing between interconnect lines determines the coupling capacitance. As the space decreases, the capacitance increases. By using an excellent insulating material, polyimide, to insulate the recording sites, the cross talk between conducting lines was greatly decreased.6,16 Impedance was used as a general measure of electrode surface area. Individual recording site impedance was 17.8 ( 2.8 MΩ (n ) 8 sites) for electrodes that were not cross talking after a 24 h soak in 0.9% saline. However, analysis of individual impedance values was not found to provide a reliable method of measuring cross talk. To dynamically measure cross talk, the peak current was recorded at each individual site from a multisite microelectrode as the sites were sequentially lowered into a 0.1 M PBS solution containing 250 µM ascorbic acid. The data from individual sites were normalized, and the relative change in current was calculated. This change in the baseline for the recording sites was used as a measure of cross talk. Because the background current gradually decreases over time, negative changes were included in the average as zero. Lowering the electrode into solution disturbed the double layer of the microelectrode and caused a small increase (
