Contribution of Oxygen Reduction to Charge Injection on Platinum and Sputtered Iridium Oxide Neural Stimulation Electrodes

Abstract

The extent to which oxygen reduction occurs on sputtered iridium oxide (SIROF) and platinum neural stimulation electrodes was quantified by cyclic voltammetry and voltage-transient measurements in oxygen-saturated physiological saline. Oxygen reduction was the dominant charge-admittance reaction on platinum electrodes during slow-sweep-rate cyclic voltammetry, contributing ~12 mC/cm² (88% of total charge) to overall cathodal charge capacity. For a 300-nm-thick SIROF electrode, oxygen reduction was a minor reaction contributing 1.3 mC/cm², ~3% of total charge. During current pulsing with platinum electrodes, oxygen reduction was observed at a level of 7% of the total injected charge. There was no indication of oxygen reduction on pulsed SIROF electrodes. A sweep-rate-dependent contribution of oxygen reduction was observed on penetrating SIROF microelectrodes (nominal surface area 2000 μm²) and is interpreted in terms of rate-limited diffusion of oxygen in electrolyte that penetrates the junction between the insulation and electrode shaft. For typical neural stimulation pulses, no oxygen reduction could be observed on penetrating SIROF microelectrodes. Based on the in vivo concentration of dissolved oxygen, it is estimated that oxygen reduction on platinum microelectrodes will contribute less than 0.5% of the total injected charge and considerably less on SIROF electrodes.

I. Introduction

COMMONLY used platinum and iridium oxide neural stimulation electrodes inject charge into tissue by reduction and oxidation reactions occurring at the surface of an electrode or, in the case of iridium oxide, within a porous 3-D film. For iridium oxide, reduction–oxidation (redox) processes are dominant, while for platinum, capacitive charging and discharging at the metal–electrolyte double-layer is also a significant contributor to charge injection. Since most neural stimulation protocols employ biphasic current or voltage pulses in which the leading phase is cathodal, electrodes are polarized negatively to potentials at which oxygen reduction can occur. The chemical products of oxygen reduction are reactive and have been identified in studies by Morton et al. [1] and Wang et al. [2], as potentially injurious to tissue. In the present paper, the extent to which oxygen reduction is a charge-injection process on platinum and sputtered iridium oxide (SIROF) stimulation electrodes has been examined in vitro under the near-equilibrium conditions of slow-sweep-rate cyclic voltammetry and during high-current-density, cathodal-first pulsing.

Based on the present results and earlier reports with SIROF electrodes [2], it is evident that the extent of oxygen reduction depends on the morphology and composition of the SIROF, but is substantially absent during a charge injection pulse. Similarly, although oxygen reduction is clearly observable on platinum electrodes, the extent of this reaction in vitro during current pulsing is modest. The extent to which oxygen reduction might occur in vivo is discussed in terms of the physiological oxygen concentration and diffusion rate, both of which favor a deceased contribution of oxygen reduction to charge admittance compared with the present in vitro measurements.

II. Experiment

A. Multielectrode Arrays

Both planar and penetrating SIROF-coated multielectrode arrays were used in the study. The planar microelectrodes were fabricated on flexible polyimide film using approximately 1.5 μm thick, thermally evaporated gold metallization at the electrode sites, which were coated with SIROF using deposition methods that have been described earlier [3]. Two array geometries, one comprising 15400-μm-diameter electrodes and one comprising four sets of four electrodes with diameters of 50, 100, 200, and 400 μm were employed. Scanning electron micrographs (SEMs) of these arrays and a 400-μm diameter, SIROF-coated site are shown in Fig. 1. Only SIROF electrodes with diameters of 50 μm (geometric surface area, GSA = 1960 μm²) and 400 μm (GSA = 125 600 μm²) were investigated. Unless otherwise noted, the SIROF was 300 nm thick, which results in a film that is smooth when imaged at low magnification (see Fig. 1), although a slightly nodular morphology is apparent at higher magnification [3].

Fig. 1.

Scanning electron micrographs showing the geometry of the planar SIROF electrode arrays employed in the study. Upper left array has four 400-, 200-, 100-, and 50-μm diameter sites. Upper right array has 15 400-μm diameter sites. The SIROF is 300 nm thick on both arrays and is deposited to a slightly larger diameter than the via that exposes the underlying gold contact creating a lip of SIROF on the top polyimide encapsulation (lower image). Both arrays employ a-SiC encapsulation of the gold traces.

The gold leads on the planar substrates shown in Fig. 1 were encapsulated in thin films of amorphous silicon carbide (a-SiC) deposited by plasma-enhanced chemical vapor deposition (PECVD), as shown schematically in Fig. 2. The use and properties of a-SiC as an encapsulation layer for implantable neural electrodes has been described previously by Cogan et al. [4] and by Hsu et al. [5]. The a-SiC is effective in preventing electrolyte leakage between the conductive traces and insulation, eliminating any contribution of gold underlying the polyimide encapsulation to the measured electrochemical properties of the electrodes. The a-SiC encapsulation was accomplished by depositing a 0.5-μm-thick a-SiC film onto polyimide and patterning and depositing the gold metallization directly onto the a-SiC. A second layer of 0.5-μm a-SiC was deposited over the metallization and a spun polyimide film applied over the a-SiC to complete the encapsulation. Vias to the charge injection sites were opened by reactive-ion etching through the top polyimide and a-SiC. The SIROF was deposited into the vias by reactive sputtering using a thin titanium film (50 nm) as an adhesion layer between the gold and SIROF. The SIROF deposition was limited to the vias by a patterned layer of photoresist. The effect of SIROF thickness on oxygen reduction was investigated with a different array design that has been described in detail earlier [3]. Four SIROF thickness levels of 300, 600, 900, and 1200 nm were investigated using 1960-μm² sites on these arrays.

Fig. 2.

Schematic cross section of a planar polyimide array showing the encapsulation of the gold contact metallization by a-SiC.

The penetrating microelectrodes were based on the Utah Array design and were fabricated by Blackrock Microsystems [6], [7]. The microelectrodes have a faceted, pyramidal electrode tip with a nominal exposed GSA of 2000 μm². The SIROF on the penetrating arrays is deposited directly onto the conductive, highly doped silicon used as the structural element of the array shafts. After SIROF deposition, the arrays are coated with Parylene-C and the SIROF electrode sites exposed by etching in an oxygen plasma. SEM images of the array and electrode tip are shown in Fig. 3. The SIROF on the penetrating arrays was deposited by reactive sputtering under the same conditions as used with the planar polyimide arrays and has a nominal thickness of 300 nm, although the thickness on the pyramidal tips will not be the same as that on the planar substrates. A thin (50 nm) sputtered titanium film was also used as an adhesion layer between the SIROF and silicon. Similar penetrating arrays were obtained from Blackrock Microsystems with sputtered platinum on the electrode tips. The nominal exposed GSA of the platinum was also 2000 μm², although the masking process for exposing the platinum or SIROF on the penetrating arrays leads to variability in the electrode area, which results in variability in the electrochemical and charge-injection properties of these electrodes.

Fig. 3.

Scanning electron micrograph of a penetrating SIROF-coated array with a 3 × 4 arrangement of nominally 2000-μm² electrodes. The insert shows the pyramidal shape of the exposed microelectrode tip.

For comparison with the platinum microelectrode arrays, a Bioanalytical Systems’ large-area planar platinum disk electrode (MF-2013) was employed in some studies. The disk electrode had a GSA of 0.02 cm² and was cleaned by polishing with 1-μm alumina followed by rinsing with deionized water.

B. Cyclic Voltammetry

Cyclic voltammetry was performed with a Gamry PC4 potentiostat in a standard three-electrode configuration using a large-area platinum counterelectrode and a Ag|AgCl (4M KCl) reference electrode. The electrolyte employed was phosphate-buffered saline (PBS) having a composition of 126 mM NaCl, 22 mM NaH₂PO₄-7H₂O, and 81 mM Na₂HPO₄-H₂O at pH 7.2–7.4. The PBS was either saturated with oxygen by bubbling oxygen gas through the electrolyte for at least 30 min or the electrolyte was deoxygenated by bubbling argon through the electrolyte for at least 60 min. The gas flows were maintained during the cyclic voltammetry measurements by retracting the gas supply tube from the electrolyte and allowing the gas to flow continuously into the headspace above the electrolyte.

Cyclic voltammetry measurements for both SIROF and platinum electrodes were made at a sweep rate of 50 mV/s between potential limits of −0.6 and 0.8 V, beginning at open-circuit potential and sweeping in the positive direction first. The CV response stabilized rapidly in these in vitro electrolytes and the data reported were acquired on the fifth CV cycle. In a few studies, a rapid sweep rate of 50 000 mV/s was employed in an effort to distinguish between reactions at the exposed tip and those resulting from electrolyte leakage under the Parylene-C insulation on the penetrating microelectrodes [8]. The cathodal charge storage capacity (CSC_c) of the electrodes was calculated from the 50- or 50 000-mV/s CVs as the time integral of the negative current during a full CV cycle. Typically, the CSC_c is used as an estimate of the total capacitive and Faradaic charge available between the potential limits of the CV cycle. For SIROF at a sweep rate of 50 mV/s, the CSC_c is primarily a measure of the charge available from the Ir³⁺/Ir⁴⁺ reduction-oxidation couple, while on platinum, the CSC_c comprises double-layer charging, oxide reduction, and hydrogen deposition. Since oxygen reduction can occur within the potential range of the CV sweep, the CSC_c also includes any contribution from oxygen reduction at the electrode.

C. Voltage Transients

Voltage transients were measured with a custom-built Sigenics (Chicago, IL) stimulator that provides monophasic cathodal current pulses with the electrode biased in the interpulse period [8], [9]. In the present study, measurements were made with the SIROF and platinum biased at either 0.0 or 0.6 V (AglAgCl). Charge balance was obtained by reestablishing the interpulse bias using an anodic recharge current that was sufficient to establish the bias within a few milliseconds after the end of the cathodal current pulse. The stimulator is designed to limit the recharge current so that the microelectrode cannot be polarized more positively than the 0.8 V water oxidation limit observed with iridium oxide or Pt electrodes [10]. A 1.9-ms delay was imposed between the end of the cathodal current pulse and the anodic recharge phase to facilitate analysis of the electrode polarization. The electrode is at open-circuit in the interphase period and, with no imposed current, there is no ohmic or activation overpotential contribution to the measured potential. The maximum negative potential excursion (E_mc) was taken as the potential at which the current becomes zero after the end of the cathodal current pulse. The time for the current to decay to zero was about 18 μs.

D. Statistical Treatment

When measurements of CSC_c or voltage-transient parameters were compared for the same set of electrodes in argon or oxygen-saturated PBS, the results are reported as the mean ± standard-deviation for each gas saturation. The null hypothesis that the means are not different was tested with a two-sided, paired t-test using a significance level of α = 0.05.

III. Results

A. SIROF Cyclic Voltammetry

Cyclic voltammograms of six 400-μm-diameter planar SIROF electrodes from a single array, acquired at 50 mV/s in argon and oxygen-saturated PBS, are compared in Fig. 4. Oxygen reduction is apparent from the higher currents observed in oxygen-saturated PBS at potentials slightly more negative than −0.3 V, as the potential is swept in the negative direction. The effect of oxygen persists on the positive sweep to a potential of about 0.0 V. The CSC_c increased from 43.9 ± 0.2 mC/cm² in argon-saturated PBS to 45.2 ± 0.1 mC/cm² in oxygen-saturated PBS. The 1.3-mC/cm² difference in means is significant with p < 0.0001 (n = 6). The results shown in Fig. 4 suggest that oxygen reduction occurs, for this particular SIROF thickness and electrode geometry, as a minor charge-admittance process, contributing about 3% of the total CSC_c at potentials between −0.3 and −0.6 V (Ag|AgCl), the latter being the potential for the onset of water reduction on iridium oxide at neutral pH and a limit for reversible injection of charge [8], [11]. A similar degree of oxygen reduction was observed with 1960 μm² planar SIROF electrodes. The average CSC_c of four 50-μm-diameter electrodes increased from 61.4 ± 4.2 mC/cm² in Ar-saturated PBS to 64.6 ± 3.8 mC/cm² in O₂-saturated PBS, which is significantly different (p = 0.0009). Similar to the 400-μm-diameter SIROF electrodes, the effect of oxygen was observed at potentials negative of −0.3 V with a resultant 5% increase in CSC_c in the presence of oxygen. A higher level of oxygen reduction might be expected on the 50-μm-diameter microelectrodes, since oxygen transport has a significant contribution from hemispherical diffusion, as opposed to linear diffusion, which dominates oxygen transport to the larger electrodes.

Fig. 4.

Cyclic voltammograms of six SIROF electrodes in Ar-saturated and O₂-saturated PBS at a 50-mV/s sweep rate. The average CSC_c capacities are 43.9 ± 0.2 mC/cm² in argon and 45.2 ± 0.1 mC/cm² (mean ± S.D.) in oxygen.

Oxygen reduction was also investigated with 50-μm-diameter planar SIROF electrodes cycled at a 50 000-mV/s sweep rate. To ensure that the response was steady state, the electrodes were subjected to 50 consecutive CV cycles and the CSC_c calculated from the fiftieth cycle. The CSC_c values were 11.7 ± 2.0 mC/cm² and 11.3 ± 1.3 mC/cm² in argon-and oxygen-saturated PBS, respectively. The difference in the means is not significant (p = 0.46, n = 4). High sweep rates were not useful for investigating 400-μm diameter electrodes because of the concentration and ohmic overpotential contribution to the measured potentials at the high current densities encountered with large electrodes at the 50 000-mV/s sweep rate. The large overpotentials result in an overestimate of the true electrode potential, which in the case of a negative CV sweep, may not be sufficiently negative to reduce oxygen. In addition, the current associated with the oxygen reduction reaction, which is kinetically slow, is masked by the much larger SIROF reduction and double-layer charging currents.

Cyclic voltammograms of SIROF-coated penetrating microelectrodes (see Fig. 3) exhibited a different behavior, as shown by the representative data from one electrode in Fig. 5. The cyclic voltammograms in Fig. 5 were acquired at 50 mV/s in PBS saturated with argon and oxygen. In the oxygen-saturated PBS, there is a significant increase in current at potentials more negative than −0.05 V. This additional cathodal current is presumed to be oxygen reduction. The CSC_c for this electrode increased from 22 mC/cm² in argon-saturated PBS to 42 mC/cm² in oxygen-saturated PBS, suggesting that oxygen reduction is contributing about 20 mC/cm², or 48% of the charge admittance during the negative potential sweep. Six electrodes from the same array exhibited an increase in CSC_c from 20.4 ± 12.0 mC/cm² to 43.2 ± 9.9 mC/cm² in argon- and oxygen-saturated PBS, respectively. While the large standard deviations reflect a considerable variation in the GSA of the electrodes, the difference in means was significant (p = 0.008, n = 6). When the same electrode was evaluated in argon- and oxygen-saturated PBS at a sweep rate of 50 000 mV/s, the cyclic voltammograms were essentially identical as shown in Fig. 6. At 50 000 mV/s, the CSC_c values of ten electrodes from this array were 7.59 ± 4.43 mC/cm² and 7.57 ± 4.43 mC/cm² in argon and oxygen, respectively, which is not statistically different (p = 0.4, n = 10). The difference in the degree to which oxygen reduction occurs on the penetrating electrodes compared with the planar 1960 μm² electrodes at the 50-mV/s sweep rate is attributed to electrolyte ingress between the Parylene-C insulation and the silicon shaft of the penetrating electrodes. The apparent lack of oxygen reduction on the penetrating electrodes at the 50 000-mV/s sweep rate may in part be due to overpotential effects, but it is reasonable to presume that exposed surfaces under insulation are not accessed at high sweep rates, and therefore, do not contribute to oxygen reduction [8].

Fig. 5.

Cyclic voltammograms of a 2000-μm² SIROF penetrating microelectrode in argon- and oxygen-saturated PBS at a 50-mV/s sweep rate.

Fig. 6.

Cyclic voltammograms of a 2000-μm² SIROF penetrating microelectrode in argon- and oxygen-saturated PBS at a 50 000-mV/s sweep rate.

The effect of SIROF thickness on oxygen reduction was determined over a thickness range of 300–1200 nm using 1960-μm² planar electrodes [3]. Four electrodes of each thickness were characterized by cyclic voltammetry in argon- and oxygen-saturated PBS. The mean CSC_c values for each thickness and gas saturation are listed in Table I. Oxygen reduction increased from 5% of CSC_c at a thickness of 300 nm to over 10% of CSC_c in the 600–1200 nm thickness range.

TABLE I.

Mean CSC_c Calculated from 50-mV/s Cyclic Voltammograms of Planar, 1960-μm² SIROF Electrodes as a Function of Thickness

Thickness nm	CSCc O2 mC/cm2	CSCc Ar mC/cm2	Difference %	P
300	51.8±2.1	49.2±2.0	5	0.012
600	103.6±3.7	87.9±3.6	15	0.001
900	169±6.8	147±3.9	13	0.0004
1200	223±7.1	194±6.5	13	0.0036

p-values for comparison of means in a paired, two-sided t-test (n = 4).

B. Platinum Cyclic Voltammetry

Electrochemically driven reduction of oxygen on platinum is well known and this is shown in Fig. 7 by a comparison of 50 mV/s CVs of a planar platinum disk electrode in argon-saturated and oxygen-saturated PBS. The onset of oxygen reduction is evident at a potential of 0.2 V versus Ag|AgCl. As is clear from Fig. 7, oxygen reduction is the principal charge-admittance process in the oxygen-saturated electrolyte, increasing the CSC_c from 1.7 mC/cm² in argon-saturated PBS to 14.0 mC/cm² after oxygen saturation. A similar response was obtained with the penetrating platinum microelectrodes as shown in Fig. 8 for 50 mV/s CVs. The onset of oxygen reduction, from Fig. 8, is also apparent at potentials just negative of 0.2 V. The mean charge CSC_c values in argon and oxygen-saturated PBS were significantly different, 11.5 ± 2.6 mC/cm² and 183 ± 41 mC/cm², respectively (p = 0.0007, n = 5). Similar to the size difference observed with SIROF electrodes, the larger oxygen reduction current density at the platinum microelectrode, compared with the large-area platinum-disk electrode, is due to the contribution of spherical rather than planar diffusion of oxygen. Unlike the SIROF penetrating microelectrodes shown in Fig. 6, however, the effect of oxygen reduction was also apparent in 50 000 mV/s CVs of platinum microelectrodes as shown by the representative comparison in Fig. 9. The mean CSC_c calculated from 50 000 mV/s CVs of seven electrodes was 1.1 ± 0.5 mC/cm² and 1.4 ± 0.5 mC/cm² in argon and oxygen, respectively, suggesting that ~0.3 mC/cm² or 21% of the charge admittance is due to oxygen reduction when the PBS is saturated with oxygen gas. The difference in means is significant (p = 0.0001, n = 7).

Fig. 7.

Cyclic voltammograms of a 0.02 cm² platinum disk electrode in argon- and oxygen-saturated PBS at a 50-mV/s sweep rate.

Fig. 8.

Cyclic voltammograms of a 2000-μm² platinum penetrating microelectrode in argon- and oxygen-saturated PBS at a 50-mV/s sweep rate.

Fig. 9.

Cyclic voltammograms of a 2000-μm² platinum penetrating microelectrode in argon- and oxygen-saturated PBS at a 50 000-mV/s sweep rate.

C. Other Electrode Materials

Besides platinum and iridium oxide, thin films of gold, titanium, and titanium nitride (TiN) are employed as interconnect metallization and adhesion layers in electrode fabrication, and in the case of porous TiN, as a low-impedance electrode coating. The extent to which oxygen reduction occurs on these materials was surveyed by comparing CSC_c values in argon- and oxygen-saturated PBS at 50 mV/s, and these results are presented in Table II, which includes a compendium of the CSC_c measurements from all electrode types and materials investigated.

TABLE II.

Compendium of CSC_c Values and Calculated Contribution of Oxygen Reduction to CSC_c for SIROF and Platinum Electrodes Measured in PBS

Electrode	GSA	Rate mV/s	CSCc Ar mC/cm2	CSCc O2 mC/cm2	O2 Reda %	Geometry - Comments
SIROF	125,600 μm2	50	43.9	45.2	3	planar – 300 nm thick
SIROF	1960 μm2	50	61.4	64.6	5	planar – 300 nm thick
SIROF	2000 μm2	50	20.4	43.2	53	penetrating electrodeb
SIROF	2000 μm2	50,000	7.59	7.57	0	penetrating electrode
Platinum	0.02 cm2	50	1.7	14.0	88	planar
Platinum	2000 μm2	50	11.5	183	94	penetrating electrodeb
Platinum	2000 μm2	50,000	1.1	1.4	21	penetrating electrodeb
Gold	0.02 cm2	50	0.83	8.44	90	planar
TiN	4 cm2	50	6.6	7.4	11	planar, −0.6/0.8 Vc
TiN	4 cm2	50	11.2	12.8	13	planar, −0.9/0.8 Vd
Ti	20,000 μm2	50	0.16	0.33	52	planar

Representative data for gold, titanium nitride, and titanium metal are included for comparison.

^aPercent contribution of oxygen reduction to the CSCc measured in oxygen-saturated PBS.

^bIncludes oxygen reduction contribution from conductors underlying electrode encapsulation.

^cStandard −0.6 to 0.8 V potential range for TiN CV.

^dExpanded negative potential limit for TiN CV [23].

D. SIROF Voltage Transients

The voltage-transient response of 12 400-μm diameter SIROF electrodes was compared in argon-saturated and oxygen-saturated PBS in response to 2-mA constant current pulsing with 1-ms cathodal pulses at a frequency of 20 Hz, corresponding to a charge injection of 2 μC/phase (1.6 mC/cm²). The SIROF electrodes were biased at 0 V versus Ag|AgCl in the interpulse region and the charge was sufficient to polarize the electrodes to about the −0.6 V limit for avoiding water reduction. This charge level ensures that the SIROF is polarized over the full potential range at which oxygen reduction might occur but without water reduction. While earlier studies have shown that a significantly higher charge-injection capacity is obtained by biasing the SIROF to 0.6 V versus Ag|AgCl in the interpulse period [3], this strategy could not be employed with the 125 600-μm² SIROF electrodes, because the current output of the stimulator was insufficient to polarize the electrodes to the −0.6 V water reduction limit. A comparison of the voltage-transient response in oxygen- and argon-saturated electrolyte is shown in Fig. 10. From these voltage transients, the maximum driving voltage (V_drv), the access voltage (V_a), the maximum negative potential excursion (E_mc), and the potential at the end of the zero-current interphase period (E_ip) were determined. For all parameters, there was no significant difference in oxygen- or argon-saturated PBS as detailed in Table III.

Fig. 10.

Comparison of the voltage-transient response of a planar 125 600-μm² SIROF electrode in argon- and oxygen-saturated PBS, and corresponding current waveform. V_drv = maximum driving voltage, V_a = access voltage, E_mc = maximum negative potential excursion, E_ip = potential at the end of the zero-current interphase period.

TABLE III.

Comparison of Voltage-Transient Parameters of Planar 125 600-μM² SIROF Electrodes in Argon- and Oxygen-Saturated PBS (n = 12)

	PBS-Argon, V	PBS-Oxygen, V
Vdrv	2.16±0.03	2.16±0.03
Va	1.84±0.03	1.83±0.03
Emc	−0.59±0.01	−0.59±0.01
Eip	−0.32±0.01	−0.32±0.01

Voltage versus Ag|AgCl.

The voltage-transient response from a 1960-μm² planar SIROF electrode in oxygen- and argon-saturated PBS is compared in Fig. 11. The SIROF was subjected to 1 ms, 237-μA pulses (237 nC/phase, 11.9 mC/cm²) at a repetition rate of 20 Hz from an interpulse bias of 0.6 V versus Ag|AgCl. The transient waveform shown in Fig. 11 is an average over 64 pulses. The maximum negative potential excursion (E_mc) and the 1.9-ms interphase period in which the imposed current is zero are also indicated in Fig. 11. For four 1960-μm² electrodes on one array, the E_mc in oxygen- and argon-saturated PBS was —0.55 ± 0.08 V and −0.54 ± 0.07 V, respectively, which is not significantly different (p = 0.39, n = 4). Pulsing the 1960-μm² electrodes at a slightly lower current (206 μA) produced the same result with E_mc values of −0.44 ± 0.06 V and −0.45 ± 0.07 V in oxygen- and argon-saturated electrolyte, respectively (p = 0.82, n = 4). Other waveform parameters, V_a, V_drv, and E_ip were also not significantly different at either current level.

Fig. 11.

Comparison of the voltage-transient response of a planar 1960-μm² SIROF electrode in argon- and oxygen-saturated PBS with corresponding current waveform.

The effect of oxygen on the voltage transients of penetrating SIROF microelectrodes was evaluated by pulsing five electrodes to a maximum negative potential excursion (E_mc) of approximately −0.5 V versus Ag|AgCl using a 1-ms pulsewidth and 0.6-V interpulse bias at a frequency of 20 Hz. Because of variations in electrode surface area, different electrodes required different currents to achieve the same polarization level. The test protocol for these electrodes involved setting a current level to achieve the desired polarization and then changing between oxygen and argon gas saturation of the PBS without changing the current setting on the stimulator. There was no statistically significant difference between the voltages transients in argon- or oxygen-saturated PBS with values as follows: E_mc(Ar) = −0.50 ± 0.01 V, E_mc(O₂) = −0.52 ± 0.01 V (p = 0.07, n = 5); and E_ip(Ar) = −0.30 ± 0.04 V, E_ip(O₂) = −0.30 ± 0.03 (p = 1, n = 5). The current levels across the five electrodes varied from 120 μA to 240 μA, reflecting the variation in electrode area.

E. Platinum Voltage Transients

A comparison of voltage transients of 2000-μm² platinum penetrating microelectrodes pulsed at 20 Hz in argon- and oxygen-saturated PBS revealed a small but significant contribution of oxygen to charge admittance. The electrodes were subjected to 19.6-μA, 1-ms pulses (19.6 μC/phase, 1 mC/cm²) from a bias of 0.6 V, which is approximately 0.3 V positive of the open-circuit potential of these electrodes in PBS. A comparison of the voltage transients is shown in Fig. 12. The high charge capacity of the platinum electrodes is not only due to the use of the positive bias, but also the likelihood that the exposed geometric area of the electrode tips exceeds the nominal 2000 μm². The average E_mc decreased from −0.43 ± 0.08 V in Ar-saturated PBS to −0.41 ± 0.08 V in oxygen-saturated PBS. While the difference in these means is small, 0.02 V, it was significant (p = 0.045). Similarly, E_ip decreased from −0.17 ± 0.06 V to −0.14 ± 0.06 after saturating the PBS with oxygen. The difference in means of 0.03 V is also significant (p = 0.005). As would be expected, V_a, which is determined primarily by electrolyte conductivity, was not significantly different in oxygen- or argon-saturated PBS.

Fig. 12.

Comparison of the voltage-transient response of a penetrating platinum microelectrode (nominal GSA = 2000 μm²) in argon- and oxygen-saturated PBS with corresponding current waveform.

The quantity of charge admitted by the oxygen reduction reaction on these platinum microelectrodes was estimated by pulsing three electrodes from the previous group in argon-saturated PBS with 19.6 μA pulses, and for each electrode determining the decrease in current that results in a 0.02-V reduction in electrode polarization (i.e., a 0.02-V positive shift in E_mc). For all three electrodes, decreasing the current by 1.4 μA resulted in a 0.02 V decrease in polarization, suggesting that in oxygen-saturated PBS about 7% of the charge admittance, under the particular pulsing conditions of the study, is by reduction of oxygen.

IV. Discussion

Oxygen reduction has been identified as a charge-admittance reaction during current pulsing with gold electrodes [1]. In their study, Morton et al. [1] employed the pulse-clamp method in which the electrode potential following a cathodal current pulse is clamped to the potential at the start of the pulse. The technique effectively controls the interpulse potential to a value close to the open-circuit potential of the gold in their saline electrolyte. The corresponding voltage and current waveforms are similar to those obtained with stimulation protocols that employ cathodal-first current pulsing with interpulse bias control to maintain charge balance [9], [10]. By calculating the difference between the anodal charge necessary to reestablish the starting potential and the cathodal charge in the current pulse, Morton et al. [1] defined an unrecoverable charge attributable to irreversible oxygen reduction during the cathodal pulse. The charge was significant, about 10%-35% of the total charge for a 100-μ/cm² pulse. The unrecoverable charge increased with the injected charge density and with the time delay between the end of the cathodal pulse and the initiation of the potential clamp.

The voltage-transient response for the platinum penetrating electrode, shown in Fig. 12, is consistent with the results from Morton et al. [1]. The observation that E_mc on platinum is slightly less negative in the presence of oxygen, suggests that oxygen reduction is contributing to charge admittance during the 1-ms cathodal current pulse. During the 1.9-ms interphase period, the potential of the platinum electrode in oxygen-saturated PBS also shifts more positive than in the argon-saturated electrolyte. Although the difference is again small, a 0.03-V shift, the result was quite significant. The greater positive shift in oxygen-saturated PBS is attributed to chemical oxidation of the reduced platinum surface in the presence of dissolved oxygen. Chemical oxidation due to dissolved oxygen also explains the observation by Morton et al. [1] of a higher charge-imbalance for longer interphase delays with gold electrodes; the increase in the imbalance is due to chemical oxidation, which decreases the anodal charge required to reestablish the interpulse potential when the voltage clamp is applied.

In the present study, no effect of oxygen reduction was observed during pulsing of SIROF electrodes with 1 ms pulsewidths. This observation is consistent with the cyclic voltammetry measurements, which indicate that oxygen reduction is a minor charge-admittance reaction for the 300-nm SIROF electrodes investigated. While the planar SIROF electrodes exhibited very little oxygen reduction, oxygen reduction has been identified as a charge-admittance reaction on thicker and more porous, 2-μm-thick SIROF electrodes in the study by Wang et al. [2]. They report significant oxygen reduction at potentials more negative than −0.3 V (SCE) during cyclic voltammetry in a combined phosphate-carbonate-buffered physiological saline saturated with gas of composition 5% CO₂, 6.1% O₂, and 88.9% N₂. By calculating the difference in cathodal and anodal charge-storage capacity in the absence and presence of oxygen during cyclic voltammetry, a mean cathodal imbalance, attributable to oxygen reduction, of ~3 mC/cm² was observed for SIROF with a CSC_c of 34 mC/cm². This 10% contribution of oxygen reduction to charge admittance, while higher than the 3% contribution observed in the present study with 300-nm-thick SIROF in oxygen-saturated PBS, is consistent with our observation of 13%–15% contribution with 600–1200-nm-thick SIROF reported in Table I.

The notable increase in oxygen reduction observed when SIROF film thickness is increased from 300 to 600 nm suggests that the reduction reaction occurs on the iridium oxide and is not caused by the underlying titanium or gold metallization. Any contribution from the titanium or gold is expected to increase as the SIROF thickness decreases, since the thinner SIROF is less limiting of oxygen diffusion. Saturation of the oxygen reduction reaction in the 600–1200 nm range is attributed to the limited diffusion rate of dissolved oxygen into the pore structure of the SIROF. In effect, above a critical SIROF thickness, no additional oxygen can diffuse into the pore structure over the time scale of the cyclic voltammogram, similar to the phenomenon described by Park et al [12] for nanoporous platinum electrodes.

There are several reasons why oxygen reduction in vivo will contribute less to charge admittance than might be estimated from in vitro studies. Measurement of oxygen reduction by Morton et al. [1] on gold electrodes, and in the present study on platinum electrodes, was made in buffered saline solutions that were saturated with oxygen at ambient temperature and pressure. The oxygen concentration under these conditions is about 1 mM, with an accompanying oxygen partial pressure (pO₂) of 736 mmHg, after accounting for the partial pressure of water vapor at 25 °C [13]. The oxygen concentration in tissue is much lower. The reported average pO₂ in brain is about 30 mmHg [14], which corresponds to a dissolved molecular oxygen concentration of 0.04 mM, using an oxygen solubility constant of 1.35 × 10^–3 mM/mm Hg [15]. The extent to which oxygen reduction occurs in vivo, therefore, will be significantly less than observed in these in vitro studies for which the dissolved oxygen concentration is about 25 times higher than in brain. Higher oxygen concentrations have been observed in the vicinity of stimulation electrodes in response to enhanced neuronal activity during pulsing [16]. However, the maximum increase is modest, about 15 mmHg, resulting in an oxygen concentration of about 45 mmHg, which is approximately 6% of the concentration in oxygen-saturated PBS.

The rate of oxygen diffusion to an electrode during current pulsing may also limit the current density that can be supported by oxygen reduction. A 19.6-μA current pulse (see Fig. 12) corresponds to a current density of ~1 A/cm², which is considerably higher than the near-steady-state maximum of 6 mA/cm² observed by cyclic voltammetry (see Fig. 8). This latter current density is similar to mass-transport-limited oxygen reduction current densities observed on metal electrodes [17], although somewhat higher due to the spherical microelectrode geometry and mechanical agitation of the oxygen-saturated PBS due to gas flow. Whether mass transport limits the rate of oxygen reduction at an electrode will depend on the imposed current density and pulsewidth, as well as the diffusion rate of oxygen through tissue adjacent to the electrode. In general, diffusional limitations will be most important when the electrolyte oxygen composition at the electrode surface is reduced to near zero. Finally, oxygen reduction is kinetically slow and surface adsorption of biomolecules, and possibly inorganic anions, is likely to impede the reduction reaction, and further reduce the contribution of oxygen reduction to charge admittance during a stimulation pulse [18].

For platinum microelectrodes, an oxygen reduction contribution of about 7% was estimated for pulsing in oxygen-saturated PBS. Considering the much lower in vivo concentration of dissolved oxygen, and the aforementioned factors, it seems likely that oxygen reduction will be a very minor charge-admittance process on platinum electrodes, contributing less than 0.5%, if the extent of the reaction scales with dissolved oxygen concentration. Although the contribution from oxygen reduction is very low, it cannot be excluded as a tissue damage mechanism, since there will be reactive species generated, even on SIROF electrodes, albeit at low concentrations. However, other mechanisms such as pH change [19], neuronal hyperactivity [20], electroporation [21] and mechanical motion [22], appear more likely causes of tissue damage with implanted stimulation electrodes than oxygen reduction.

V. Conclusion

The extent to which oxygen reduction occurs on SIROF and platinum electrodes has been quantified in vitro by cyclic voltammetry and voltage-transient measurements. Oxygen reduction is the dominant charge-admittance reaction on negatively polarized platinum electrodes during slow-sweep-rate cyclic voltammetry and a minor reaction on SIROF electrodes, even when measured at oxygen concentrations well above physiological levels. During current pulsing with platinum electrodes, oxygen reduction was observed at a level of 7% of the total injected charge. The contribution of oxygen reduction to charge admittance during a cathodal stimulation pulse was too small to be measured on planar or penetrating SIROF electrodes by observing differences in voltage transients between argon- and oxygen-saturated PBS. While the present study suggests that oxygen reduction is a very minor charge-admittance process in vivo, on either pulsed platinum or SIROF electrodes, some oxygen reduction is inevitable. It seems likely, however, that for tissue damage caused by implanted stimulation electrodes, mechanisms other than oxygen reduction will be more dominant.

Biographies

Stuart F. Cogan (M’95) received the B.Sc. degree in mechanical engineering and the M.S. degree in materials science from Duke University, Durham, NC, in 1975 and 1977, respectively, and the Sc.D. degree in materials science from the Massachusetts Institute of Technology, Cambridge, in 1979.

He is currently the Director of Advanced Materials Research, EIC Laboratories, Inc., Norwood, MA. He is currently involved in the research on electrodes for retinal prostheses, vision prostheses using intracortical stimulation, cardiac pacing and defibrillation, and neurotrophin releasing polymers for intracortical recording electrodes. His research interests include thin-film electrochromics for optical switching devices, materials for encapsulating implanted medical devices, and electrode materials for stimulation and recording in prosthetic and pacing applications.

Julia Ehrlich received the M.S. degree in materials science from the Institute of Fine Chemical Technology, Moscow, Russia, in 1991.

She was at ETEX Corporation for one year. Since 1993, she has been at EIC Laboratories, Inc., Norwood, MA. Her current research interests include the development of electrodeposition processes for electroactive coatings, and the characterization of coatings by electrochemical and electron microscopy techniques.

Timothy D. Plante received the B.Sc. degree in chemistry from Providence College, Providence, RI, in 1983.

He is currently with EIC Laboratories, Inc., Norwood, MA. His research has been concerned with the development of thin film deposition processes and the electrochemical characterization of materials for implantable medical electrodes and optical switching devices, with an emphasis on development and testing of electrode coatings for use in neural recording and stimulation.

Marcus D. Gingerich received the B.S. degree in electrical engineering from Michigan Technological University, Houghton, in 1992, the M.S. degree in biomedical engineering in 1994, and the M.S. and Ph.D. degrees in electrical engineering, in 1996 and 2002, respectively, from the University of Michigan, Ann Arbor.

He joined the Boston Retinal Implant Project in 2002 as a Biomedical Research Engineer in the Center for Innovative Visual Rehabilitation, Veterans Affairs Medical Center (VAMC), Boston, MA, where he was involved in the research remotely as a Visiting Scientist at Cornell University utilizing the Cornell NanoScale Science and Technology Facility. He is currently a Health Scientist in the VAMC, where he is involved in the research on the advanced microfabrication technologies related to the development of a retinal prosthesis. His research interests include retinal prostheses, neuroprosthetics, biomicroelectromechanical systems, microfabrication and microelectrode technology, implantable microelectronic systems, and micropackaging.

Dr. Gingerich is a member of Association for Research in Vision and Ophthalmology.

Douglas B. Shire (S’84—M’08) received the B.S. degree from Rensselaer Polytechnic Institute, Troy, NY, in 1984, and the Ph.D. degree in electrical engineering from Cornell University, Ithaca, NY, in 1989.

From 1989 to 1994, he was at Hewlett-Packard company, Optoelectronics Division, San Jose, CA. He was a Postdoctoral Associate at Cornell University during 1994 through 1997. He was an Adjunct Assistant Professor at Electrical Engineering Department, Syracuse University, NY, where he was involved in teaching. In 1997, he joined the Boston Retinal Implant Project team, where he was involved in developing microfabrication processes for creating electrode arrays for retinal neurostimulation, first in a consulting capacity and later as a member of the Center for Innovative Visual Rehabilitation, Veterans Affairs Medical Center (VAMC), Boston, MA, and as a Visiting Scientist at Cornell University. Since 2006, he has been an Engineering Project Manager for the retinal prosthesis development team, VAMC. His current research interests include development of high-reliability neurostimulators with large channel counts, and microfabrication techniques for biocompatible, and high-charge-capacity flexible electrode arrays.

Dr. Shire is a member of Tau Beta Pi, Eta Kappa Nu, and the Association for Research in Vision and Ophthalmology (ARVO).