Activated iridium oxide film (AIROF) electrodes for neural tissue stimulation

Abstract

Objective.

Iridium oxide films are commonly used as a high charge-injection electrode material in neural devices. Yet, few studies have performed in-depth assessments of material performance versus film thickness, especially for films grown on three-dimensional (instead of planar) metal surfaces in neutral pH electrolyte solutions. Further, few studies have investigated the driving voltage requirements for constant-current stimulation using activated iridium oxide (AIROF) electrodes, which will be a key constraint for future use in wirelessly powered neural devices.

Approach.

In this study, iridium microwire probes were activated by repeated potential pulsing in room temperature phosphate buffered saline (pH 7.1–7.3). Electrochemical measurements were recorded in three different electrolyte conditions for probes with different geometric surface areas (GSAs) as the AIROF thickness was increased.

Main results.

Maintaining an anodic potential bias during the inter-pulse interval was required for AIROF electrodes to deliver charge levels considered necessary for neural stimulation. Potential pulsing for 100–200 cycles was sufficient to achieve charge injection levels of 2.5 mC cm⁻² (50 nC/phase in a biphasic pulse) in PBS with 2000 μm² iridium probes. Increasing the electrode surface area to 3000 μm² and 4000 μm² significantly increased charge-injection capacity, reduced the driving voltage required to deliver a fixed amount of charge, and reduced polarization of the electrodes during constant-current pulsing.

Significance.

This study establishes methods for choosing an activation protocol and a desired GSA for three-dimensional iridium electrodes suitable for neural tissue insertion and stimulation, and provides guidelines for evaluating electrochemical performance of AIROF using model saline solutions.

1. Introduction

Iridium oxide is known for its high charge storage capacity and is an attractive material for neural stimulation applications [1]. Iridium oxide electrode coatings are corrosion resistant [2], are not toxic to cells [3, 4], and also exhibit charge injection capacities (measured by constant-current pulsing voltage transients (VTs)) higher than many other electrode materials used in neural devices [5–7]. There are multiple methods for producing an iridium oxide surface layer on electrodes, most commonly by sputter deposition (SIROF) or electrodeposition (EIROF) of an iridium oxide film, or by potential cycling activation of iridium metal (AIROF). Activated iridium oxide film electrodes, sometimes referred to as anodic iridium oxide films in electrochemistry research, are commonly used in applications where polymer-insulated electrodes are used to penetrate tissue during device implantation. Previous studies have reported on the electrochemical properties and stability of AIROF under a range of conditions, both in vivo [8] and in saline electrolytes [9, 10].

The use of AIROF electrodes in neural devices is typically targeted at neural stimulation applications, although the use of AIROF for neural recording has been employed as well [11]. AIROF electrodes were used for neural stimulation after implantation acutely in zebra finch [12], in cat for up to 415 d [11, 13, 14], in macaque visual cortex for one year [15], in human visual cortex [16], and in human auditory brain stem [17]. There is considerable potential for these electrodes to transition to clinical studies, with expectations for use in cortex [18] and in peripheral nerve [19]. In anticipation of future clinical studies, our work intends to expand upon prior studies [1, 6, 12, 15, 20–24] and more clearly establish methods for developing iridium activation protocols.

While much is known about iridium oxide, prior work does not adequately establish the effect of film thickness, as measured by charge storage capacity, on electrochemical properties of AIROF in physiologically relevant aqueous solutions (pH 7.1–7.4). Additionally, research is sparse for AIROF electrodes fabricated with a 3-dimensional conical geometry where the electrode itself is intended to penetrate tissue and facilitate implantation. Early work investigated properties of AIROF electrodes with high charge storage capacities reaching 30 to 85 mC cm⁻² [6, 25]. However, thick AIROF coatings typically do not utilize the entire surface area of the three-dimensional film structure during stimulation pulsing. Also, increasing activation to reach charge storage capacities above 100 mC cm⁻² may make AIROF more susceptible to delamination [26].

When activating iridium metal with potential cycling, the electrode potential is swept from the anodic limit $\left(E_{\mathrm{a}}\right)$ to the cathodic limit $\left(E_{\mathrm{c}}\right)$ and back to $E_{\mathrm{a}}$, typically at or below a rate of 100 mV s⁻¹. Potential pulsing methods step the electrode potential to the anodic limit and hold at $E_{\mathrm{a}}$ for a set time, then step directly to the cathodic limit and hold at $E_{\mathrm{c}}$ for the same time (as in repeated 2-step chronoamperometry or square wave potential pulsing). For both potential cycling and potential pulsing, the process is repeated as many times as necessary, continually increasing the thickness of the iridium oxide film, until the desired charge storage capacity is obtained. Charge storage capacity will increase as the thickness of the oxide layer increases [27], and the desired capacity will set the number of repetitions required for either activation method.

Prior investigations of AIROF electrodes do not provide sufficient guidance for establishing the number of activation cycles required to deliver clinically relevant levels of charge for neural stimulation applications. An FDA-approved early feasibility clinical trial that will use AIROF electrodes is in the early participant recruitment phase. And so here, we investigate the potential use of AIROF for neural stimulation in an intracortical vision prosthesis [20, 28] where the maximum necessary clinical charge injection is expected to be 0.4 nC/phase to 16 nC/phase for cathodal-first current pulsing [16, 28–30]. The AIROF electrodes investigated in this study are identical in design to those anticipated for use in the clinical trial. This work establishes the changes in the impedance, charge storage capacity, and charge injection $\left(Q_{\mathrm{i}\mathrm{n}\mathrm{j}}\right)$ capacity for AIROF electrodes in neutral pH solutions as a function of changing the number of activation cycles (increasing film thickness proportional to number of repetitions of potential pulses), the electrode surface area, and the use of an anodic potential bias applied during the interpulse period during constant-current pulsing. Table 1 below lists several abbreviations used throughout this text for various electrochemical terms and parameters.

Table 1.

List of Abbreviations.

Term	Abbreviation
Phosphate Buffered Saline	PBS
Air Equilibrated	PBSair
Argon Sparged	PBSargon
Model Interstitial Fluid	mISF
Open Circuit Potential	E oc
Anodal Potential Limit	E a
Cathodal Potential Limit	E c
Cyclic Voltammetry	CV
Electrochemical Impedance Spectroscopy	EIS
Voltage Transient	VT
Charge Storage Capacity	CSC
Cathodic Only, + 0.8 V Limit	CSCc
Cathodic Only, + 0.6 V Limit	CSC600
Max Current	Imax
Max Charge	max
Interpulse Bias	Vbias
Access Voltage	Vacc
Driving Voltage	Vdrive
Electrode Polarization	Emc

2. Methods

2.1. Test solutions

Two electrolytes were chosen for testing: phosphate buffered saline (PBS) and an inorganic model of interstitial fluid (mISF). Both electrolytes were used extensively in prior testing of materials for neural stimulation and recording. PBS is used for activation of iridium to iridium oxide and an extensive body of prior work exists for AIROF electrodes measured in PBS, facilitating electrode performance comparisons. Additionally, the conductivity and buffering capacity of PBS is relatively high, allowing detailed observations of electrochemical reactions. By comparison, the conductivity and buffering capacity of mISF is much lower than PBS and closer to the range of in vivo conductivity and buffer capacity [23].

Whereas mISF is a closer approximation for in vivo conditions than PBS, properties measured in vivo will be different than saline measurements [12]. In vivo, the charge injection capacity will be lower and access voltage will be higher, due to a combination of factors such as decreased electrolyte conductivity and adsorption of biomolecules onto the electrode surface. Thus, the maximum charge injection capacities reported in mISF likely are not achievable in the animal [12]. So, data recorded in mISF is used to establish the baseline properties of the electrodes and provide a baseline comparison for future studies of these AIROF probes in vivo.

The electrolyte solutions used for testing (PBS and mISF) contained the salt concentrations listed in table 2. Electrochemical measurements were recorded at 37 °C in three different solution conditions: air-equilibrated PBS (PBSair), argon-sparged PBS (PBSargon), and mISF sparged with a gas mixture of (5% oxygen, 6% carbon dioxide, 89% nitrogen) to maintain the pH within the desired range. The pH range for all three test conditions was 7.1–7.4. The conductivity was approximately 20–25 mS cm⁻¹ for PBS and 14–17 mS cm⁻¹ for mISF. PBSair provided a baseline assessment of electrochemical reactions at the electrode-electrolyte interface for comparison to prior publications. PBSargon provided an assessment of the effects of dissolved oxygen on reactions at the electrode surface by comparison with PBSair data. mISF provided an assessment of electrochemical performance that more closely approximates expected in vivo electrochemistry due to the lower conductivity and buffering capacity of the solution compared to PBS.

Table 2.

Electrolyte Compositions.

Ionic Component	PBS (mM)	mISF (mM)
NaH2PO4 • H2O	22.00	0.50
Na2HPO4 • 7H2O	81.05	2.00
NaCl	125.94	109.99
NaHCO3	—	28.00
KHCO3	—	7.50
MgCl2	—	0.50
MgSO4	—	0.50
CaCl2	—	0.50

2.2. Electrodes and electrochemical cleaning

Iridium microwires with 100 μm diameter were coated with Parylene-C insulation and the tips exposed by laser ablation as described previously [31] (MicroProbes for Life Science, Gaithersburg, MD). The exposed tips were fabricated to have a rounded conical geometry of approximately 5 μm diameter at the tip, 15 μm diameter at the laser cut, 65 to 120 μm exposure length, and tip angle of 10 degrees. This geometry is intended to avoid a sharp electrode tip and minimize non-uniform currents, during activation or neural stimulation, that might otherwise result in degradation of the AIROF coating. The fabrication methods for these electrodes and their use in devices for intracortical stimulation, including activation of the iridium over a wireless link, has been described previously [19, 32].

Electrodes were electrochemically cleaned in PBSair at room temperature. The electrode surface was cleaned by sweeping the potential at 4 mV s⁻¹ from + 1 V to + 2 V and then back to + 1 V vs. a Ag|AgCl (3 M NaCl) reference electrode. The surface was evaluated immediately before and after cleaning by cyclic voltammetry at sweep rates of 0.05 V s⁻¹ and 50 V s⁻¹. Some probes were evaluated with the additional sweep rates 0.10 V s⁻¹, 0.50 V s⁻¹, 1 V s⁻¹, 3 V s⁻¹, 5 V s⁻¹, and 10 V s⁻¹. Potential limits for all cyclic voltammograms were −0.60 V and + 0.80 V vs. Ag|AgCl.

2.3. Iridium activation

The anodic and cathodic limits for activation are typically set at the potential for water oxidation and reduction, respectively, which will depend on both the electrolyte solution and the reference electrode used in the electrochemical cell [33, 34]. The two methods generally used to vary electrode potential for iridium activation, linear sweep potential cycling or square wave potential pulsing, are shown in figure 1(a). All iridium electrodes in this study were activated at 37 °C in PBSair using square wave potential pulsing at 0.05 Hz and 50% duty cycle holding at + 0.80 V and −0.60 V (10 s hold at each potential) vs. a Ag|AgCl reference electrode using a Reference 1000E Potentiostat (Gamry Instruments, Warminster, PA). Electrode groups, activation, and testing steps are described in table 3. The electrochemical measurements used to determine AIROF properties at each level of activation are described in section 2.4 below.

Figure 1.

Study methods. (a) Shows the potential pulsing method used for activation of all electrodes in this study (bottom) compared to potential cycling methods (top). Both methods start the electrode at its open circuit potential $\left(E_{\mathrm{o}\mathrm{c}}\right)$, increase the potential to the anodic limit $\left(E_{\mathrm{a}}\right)$, decrease the potential to the cathodic limit $\left(E_{\mathrm{c}}\right)$, and then repeat cycling between $E_{\mathrm{a}}$ and $E_{\mathrm{c}}$. (b) Shows values extracted from voltage transient $\left(\mathrm{V}\mathrm{T}\right)$ measurements: charge per phase $\left(Q_{\mathrm{p}\mathrm{h}}\right)$, maximum current $\left(I_{\mathrm{m}\mathrm{a}\mathrm{x}}\right)$, bias potential ${(V}_{\text{bias}})$, access voltage $\left(V_{\text{acc}}\right)$, driving voltage $\left(V_{\text{drive}}\right)$, and electrode polarization $\left(E_{\mathrm{m}\mathrm{c}}\right)$. (c) Shows the two different calculations used in this study for cathodic charge storage capacity $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\right)$ over the entire test range −0.6 to + 0.8 V (left) and over the partial range −0.6 to + 0.6 V (${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$, right).

Table 3.

Study Groups: Electrode Activation and Electrochemical Tests.

Group ID	No. ofElectrodes	GSA (μm2)	Activation Steps (Total Pulses)	Tests After Cleaning and After Each Activation Step	Test Electrolytes
I	4	2000	20 500 50 600 100 700 150 800 200 900 250 1000 300 1200 400	Cyclic Voltammetry:  0.05 and 50 V s−1 Electrochemical Impedance Spectroscopy:  DC bias: Eoc and + 0.3 V Voltage Transient Measurements:  + 0.0 V bias: 5 μA (1 nC/phase),       10 μA (2 nC/phase),       Imax  + 0.6 V bias: Imax	PBSair PBSargon mISF
II	4	2000	20 50 100 150 200	Cyclic Voltammetry:  0.05, 0.50, 5, and 50 V s−1 Electrochemical Impedance Spectroscopy:  DC bias: Eoc, −0.4, 0.0, + 0.3, + 0.6 V Voltage Transient Measurements:  + 0.0 V bias: 10 μA (2 nC/phase)  + 0.6 V bias: 10 μA (2 nC/phase),       20 μA (4 nC/phase)	PBSair mISF
III	4 4 4 (12 Total)	2000 3000 4000	100	Cyclic Voltammetry:  0.05, 0.50, 5, and 50 V s−1 Electrochemical Impedance Spectroscopy:  DC bias: Eoc, + 0.3, + 0.6 V Voltage Transient Measurements:  + 0.0 V bias: 5 μA (1 nC/phase),       10 μA (2 nC/phase),       Imax  + 0.6 V bias: 10 μA (2 nC/phase),       20 μA (4 nC/phase),       40 μA (8 nC/phase),       63 μA (12.6 nC/phase),       80 μA (16 nC/phase),       100 μA (20 nC/phase),       Imax	PBSair PBSargon mISF
IV	4	2000	1000	Cyclic Voltammetry:  0.05, 0.10, 0.50, 1, 3, 5, 10, and 50 V s−1 Voltage Transient Measurements:  + 0.6 V bias: 40 μA (8 nC/phase),       Imax	PBSair mISF

E_oc = Open Circuit Potential. GSA = Geometric Surface Area.

I_max = maximum cathodic current applied before electrode polarization (E_mc) exceeded −0.6 V.

All potential values are listed with respect to a Ag|AgCl reference electrode.

Four different groups of electrodes were used to test different aspects of AIROF probes. Group I probes were used to find charge storage capacity and charge injection capacity $\left(Q_{\mathrm{i}\mathrm{n}\mathrm{j}}\right)$ changes when increasing film thickness in multiple steps. Group II probes were used to further investigate $Q_{\mathrm{i}\mathrm{n}\mathrm{j}}$ properties of AIROF electrodes at and below 200 total activation pulses. Group III probes were used to investigate the effects of varying electrode surface area on charge storage capacity, max $Q_{\mathrm{i}\mathrm{n}\mathrm{j}}$, and electrode polarization $\left(E_{\mathrm{m}\mathrm{c}}\right)$. Group IV electrodes were used to further investigate sweep rate dependent features of the CV profile and to compare charge storage capacity and $Q_{\mathrm{i}\mathrm{n}\mathrm{j}}$ properties of electrodes activated in one step to electrodes activated in multiple steps (Group I).

2.4. Electrochemical measurements

2.4.1. Cyclic voltammetry (CV)

All CV measurements were recorded over the potential range of −0.60 V to + 0.80 V vs. Ag|AgCl with varying scan rates. A total of three cycles were recorded for all CVs with a scan rate of 0.05 V s⁻¹ or 0.50 V s⁻¹; five total cycles for 1 V s⁻¹, 3 V s⁻¹, or 5 V s⁻¹ CVs; and ten total cycles for 10 V s⁻¹ or 50 V s⁻¹ CVs. Three cycles was sufficient to obtain a near-steady-state CV response.

2.4.2. Electrochemical impedance spectroscopy (EIS)

Potentiostatic EIS was recorded with a ± 10 mVrms sinusoid around the open circuit potential $\left(E_{\mathrm{o}\mathrm{c}}\right)$ or around one of four chosen DC bias levels: −0.4 V, 0.0 V, + 0.3 V, and + 0.6 V vs. Ag|AgCl. EIS is typically recorded around the open circuit potential of the electrode in solution. However, during our experiments we observed fluctuations in $E_{\mathrm{o}\mathrm{c}}$ of approximately ± 200 mV at the beginning of EIS recordings due to changes in the oxide film with each set of activation pulses and slight remnant polarization of the electrode due to CV measurements immediately preceding EIS. AIROF charge-transfer properties are highly dependent on the potential at the electrode surface, and so we recorded additional impedance spectra while potentiostatically controlling the DC bias potential of the electrodes [35, 36].

This method was used to avoid any influence of the fluctuating $E_{\mathrm{o}\mathrm{c}}$ when comparing changes in the impedance of AIROF probes as the total number of activation pulses was increased. Four different bias levels were selected to measure the impedance of the AIROF electrodes in order to understand the role of different iridium valence states in changing the overall conductance of the film. The accessible valence states of the AIROF are pH dependent and may include primarily Ir⁵⁺ at + 0.6 V bias, Ir⁴⁺ at + 0.3 V bias, and Ir³⁺ at 0.0 V and −0.4 V bias vs. Ag|AgCl [37, 38]. Bias levels were chosen based on these prior studies and the 0.05 V s⁻¹ CV profile we recorded for AIROF probes in PBSair.

2.4.3. Voltage transients (VTs)

Both CV and EIS measurements were recorded in all test solutions first, before returning to each test solution to record VT measurements. All VTs were recorded using a custom-built stimulator (Sigenics, Inc. Chicago, IL) that applies cathodal-first, constant-current pulses. Charge-balance is obtained by poteniostatically controlling the interpulse potential to a set bias versus the reference electrode. In the present study, VT measurements were recorded with the interpulse bias $\left(V_{\mathrm{b}\mathrm{i}\mathrm{a}\mathrm{s}}\right)$ controlled to either 0.0 V or + 0.6 V vs. Ag|AgCl. All VTs were recorded during pulsing at a frequency of 50 Hz with a 200 μs pulse width.

The maximum negative potential $\left(E_{\mathrm{m}\mathrm{c}}\right)$ of the electrode was determined from VT measurements using the potential of the electrode at the end of the cathodic current pulse, excluding overpotentials (access voltage, $V_{\mathrm{a}\mathrm{c}\mathrm{c}}$). The maximum charge injection capacity $\left(Q_{\mathrm{i}\mathrm{n}\mathrm{j}}\right)$ was determined by increasing the amplitude of cathodic current during pulsing until the electrode potential $\left(E_{\mathrm{m}\mathrm{c}}\right)$ reached the water reduction potential of −0.60 V vs. Ag|AgCl. ${\mathrm{I}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ is the applied cathodic current resulting in an electrode potential $\left(E_{\mathrm{m}\mathrm{c}}\right)$ equal to −0.60 V. Example VT data indicating the parameters described above are shown in figure 1(b).

2.5. Calculations and data analysis

2.5.1. Charge storage capacity

As a thicker layer of iridium oxide is formed, the current associated with charge-transfer processes between the film and the surrounding electrolyte as well as valence changes within the film will increase. This increase in current amplitude causes an overall increase in the area within the CV curve, and thus the calculated charge storage capacity. Charge storage capacity is considered to be a good measure of the maximum amount of electrical charge that can be stored within the entire volume of the porous AIROF. Charge storage capacity is influenced by scan rate dependent characteristics of the CV profile and the different charge transfer phenomena occurring within the electrochemical cell, and so the measurement is dependent on the applied potential sweep rate being slow enough to allow all electrochemical reactions to approach equilibrium during the CV measurement. Different approaches can be used to determine charge storage capacity from cyclic voltammetry data, but the most common approach is to report the cathodic charge storage capacity $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\right)$ of AIROF electrodes by recording a CV at 0.05 V s⁻¹ from −0.6 V to + 0.8 V (vs. Ag|AgCl) and integrating a partial area under the curve that includes only cathodic currents, rather than the entire CV curve.

Charge storage capacity calculations including only cathodic currents over the range of −0.6 V to + 0.6 V rather than the full CV test range of −0.6 V to + 0.8 V will also more accurately represent the total charge available for stimulation since AIROF probes are typically pulsed from a controlled positive bias of + 0.6 V (vs. Ag|AgCl). Therefore, we employed a finite Riemann sum (Equation 1) to estimate the cathodic charge contributions over the range of −0.6 V to + 0.6 V, designated as ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$, and over the full CV potential limit range of −0.6 V to + 0.8 V, designated as ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ (figure 1(c)).

$$\begin{gathered} \text{CSC}=\sum _{\text{j}={\text{E}}_c}^{{\text{E}}_a-\mathrm{\Delta }\text{v}}0.5\cdot \left({\text{Ic}}_j+{\text{Ic}}_{j+1}\right)\cdot \mathrm{\Delta }\text{t} \\ -\sum _{\text{j}={\text{E}}_{{}_c}}^{{\text{E}}_{\text{a}}-\mathrm{\Delta }\text{v}}0.5\cdot \left({\text{Ia}}_j+{\text{Ia}}_{j+1}\right)\cdot \mathrm{\Delta }\text{t} \\ \forall Ic,Ia \end{gathered}$$ (1)

Where ${\mathrm{E}}_{\mathrm{c}}$ is −0.6 V, ${\mathrm{E}}_{\mathrm{a}}$ is either + 0.8 V (for ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$) or + 0.6 V (for ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}),{\mathrm{I}\mathrm{c}}_{\mathrm{j}}$ is the current value at potential $\mathrm{j}$ for the cathodic sweep, ${\mathrm{I}\mathrm{a}}_{\mathrm{j}}$ is the current value at potential $\mathrm{j}$ for the anodic sweep, $\mathrm{\Delta }$ is the voltage step size for the CV scan, and $\mathrm{\Delta }\mathrm{t}$ is $\mathrm{\Delta }$ divided by the CV scan rate.

For AIROF, if current contributions from leakage pathways are present (e.g. damaged insulation around the electrode) then ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ calculated from slow sweep rate (≤ 0.05 V s⁻¹) CV profiles will be skewed to larger charge values that do not accurately reflect the real charge storage capacity of the electrode. In these instances, it may be more appropriate to calculate ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ from CVs recorded at higher sweep rates, up to 50 V s⁻¹, as increasing the CV sweep rate will reduce contributions from leakage pathways [39]. Further, ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ calculated from 50 V s⁻¹ CVs is more indicative of the charge injection capabilities of the AIROF electrode than ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ calculated from 0.05 V s⁻¹ CVs, as the total possible stored charge will not be transferred to the surrounding tissue during the time scale of a stimulation pulse (typically 100–400 μs).

2.5.2. Charge injection capacity and utilization

The maximum charge injection capacity per unit area ${(Q}_{\text{inj}})$, reported in units of mC cm⁻², was calculated by multiplying the maximum current measured from the VT $\left(I_{\mathrm{m}\mathrm{a}\mathrm{x}}\right)$ by the pulse width (200 μs) and dividing by the electrode’s nominal geometric surface area (GSA) of either 2000 μm², 3000 μm², or 4000 μm². Applied charge per phase $\left(Q_{\mathrm{p}\mathrm{h}}\right)$ values are reported in units of nC/phase (applied current multiplied by the pulse width). Percent utilization of the charge storage capacity was calculated as $Q_{\text{inj}}$ (with $I_{\mathrm{m}\mathrm{a}\mathrm{x}}$ measured from + 0.6 V bias VTs) divided by ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$.

2.5.3. Statistical analysis

A sample size of $\mathrm{n}=4$ electrodes was chosen for each study group based on the methods presented in prior AIROF research which used $n=5$ per group [40]. Statistical tests were not completed on data with repeated measures at increasing levels of activation or on impedance spectra. The Shapiro-Wilk test for normality was performed on ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ and $Q_{\text{inj}}$ data for Group III electrodes and failed to reject the hypothesis that the data is from a normal distribution. Statistical analyses were performed using a 1way ANOVA on ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ data (figure 5(a)) as well as $Q_{\text{inj}}$ and $Q_{\mathrm{p}\mathrm{h}}$ data (figures 5(b)–(c)). Multiple regression analysis was used to investigate the relationship between $V_{\text{drive}}$ and $E_{\mathrm{m}\mathrm{c}}$ data (figures 6(a)–(b)) and the area of the electrode at different levels of charge per phase. Regression analyses were performed using Stata (StataCorp, College Station, TX) and coefficients are reported with 95% confidence intervals and adjusted $R^2.V_{\text{drive}}$ and $E_{\mathrm{m}\mathrm{c}}$ data for individual probes are reported as mean ± standard deviation for each level of $Q_{\mathrm{p}\mathrm{h}}$ investigated.

Figure 5.

Comparison of 2000 μm², 3000 μm², and 4000 μm² iridium probes (Group IIII) after 100 activation pulses. Each data point represents one measurement for one of the $n=4$ probes in each group. The dashed lines represent the average value and the solid lines represent standard deviation for the $n=4$ electrodes. Cathodic charge storage capacity $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\right)$ in air-equilibrated phosphate buffered saline (PBSair) and model interstitial fluid (mISF) is shown for values calculated from cyclic voltammograms at (a) 0.050 V s⁻¹ and (b) 50 V s⁻¹. Charge injection capacity from voltage transients with 0.0 V and + 0.6 V interpulse bias is shown in (c) as a function of surface area ($Q_{\text{inj}}$, mC cm⁻²) and in (d) as a function of the total charge applied within each pulse ($Q_{\text{ph}}$, nC/phase).

Figure 6.

Comparison of the (a) electrode polarization and (b) driving voltage in response to increasing levels of applied charge for 2000 μm², 3000 μm², and 4000 μm² AIROF probes after 100 activation pulses (Group IIII). 200 μs pulses were applied at 50 Hz with the cathodal current set at 10, 20, 40, 63, 80, and 100 μA. The charge density corresponding to each value of applied current is shown in (c). Error bars represent standard deviations. For clarity in figures 6(a)–(b), $Q_{\text{ph}}$ on the abscissa is plotted categorically and the $Q_{\text{ph}}$ for the 2000 μm² and 4000 μm² GSAs are shifted by − 0.7 nC/phase and 0.7 nC/phase, respectively. In the regression analyses $Q_{\text{ph}}$ is treated as a continuous variable.

3. Results

3.1. Electrode cleaning

Similar electrochemical cleaning methods employing an anodic potential hold have been used in prior investigations of iridium oxide tested in sulfuric acid [41, 42]. The electrode surface observed after electrochemical cleaning is similar to previously reported characteristics for electrodes generated by the same manufacturing techniques [31]. Cyclic voltammetry profiles at 50 V s⁻¹ showed peaks that imply the presence of some oxide on the electrode surface, but not a complete monolayer as evident by lack of significant changes in 0.05 V s⁻¹ CV profiles. Apparent changes to the electrode surface due to electrochemical cleaning did not obviously affect the overall activation of iridium to iridium oxide. Figure 2 shows example SEM images of one 2000 μm² iridium probe before and after the cleaning procedure.

Figure 2.

Scanning electron microscopy images of the same 2000 μm² iridium probe before (a)–(b) and after (c)–(d) electrochemical cleaning with a 4 mV s⁻¹ potential sweep from +1 V to +2 V to +1 V vs. Ag|AgCl. Residual carbon debris is indicated by the arrow in (b). Targeted dimensions for 2000 μm² probes are labeled in (a).

Carbon debris is sometimes redeposited on the electrode surface during laser removal of Parylene-C insulation [31]. Higher laser fluence is used to minimize redeposition of carbon debris as the material properties of iridium make it resistant to damage induced by high laser power [24]. In prior work, residual carbon deposits were partially removed by ultrasonic cleaning in a chemical solution [43] and by oxygen plasma treatments [44]. Our use of an anodic potential hold with the electrode submerged in electrolyte causes oxygen evolution at the iridium metal surface that likely removes most of the residue from the fabrication process, possibly by acidification or mechanically due to gas bubble nucleation on the electrode surface. Example CVs before and after the cleaning procedure are shown in supplemental information figure S1 available online at stacks.iop.org/JNE/17/056001/mmedia.

3.2. Impedance

Impedance of the 2000 μm² AIROF electrodes decreased with increasing total number of activation pulses, as was observed in prior studies [36]. Changes in impedance and phase spectra with increasing activation were consistent across all three test solutions. Initial activation of iridium probes resulted in a significant reduction in impedance magnitude and was accompanied by changes in the phase angle. Continued activation resulted in the same trends in changing impedance magnitude and phase as initial activation. Activation to 1000 pulses, where some probes exhibited an additional CV peak we believe to be associated with fracturing of the iridium oxide film, showed distinct changes in the impedance behaviour. High frequency (1 kHz to 10 kHz) impedance magnitude increased compared to measurements of thinner iridium oxide films. Impedance and phase spectra for EIS measured around a fixed DC bias of + 0.3 V vs. Ag|AgCl are shown in figures 3(a)–(b) for Group I probes and in supplemental information figure S2 for Group II probes.

Figure 3.

(a–b) Impedance and phase spectra for 2000 μm² Group I probes measured in PBSair as a function of the number of total activation pulses. (c–d) Impedance and phase spectra measured in PBSair as a function of an applied potential bias of − 0.4, 0.0, + 0.3, or + 0.6 V vs. Ag|AgCl for 2000 μm² Group II probes after 200 activation pulses. The shaded region represents the standard deviation for $n=4$ probes.

Varying the DC bias at which EIS are recorded can significantly alter the impedance of the iridium oxide film [21, 34, 36]. EIS measurements with the AIROF in the reduced state at −0.4 V bias exhibited impedance magnitude and phase spectra similar to that of iridium probes before activation. EIS at + 0.3 V and + 0.6 V, where iridium within the AIROF is at a higher oxidation state, were not significantly different from each other. EIS at 0.0 V bias was almost identical to measurements at + 0.3 V and + 0.6 V, except for differences in the phase profile in the 10 Hz to 1 kHz range. Figures 3(c)–(d) show impedance and phase spectra measured around −0.4 V, 0.0 V, + 0.3 V, and + 0.6 V vs. Ag|AgCl for Group II probes after 200 activation cycles $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\right.5.26\pm 0.86\mathrm{m}\mathrm{C}\mathrm{c}\mathrm{m}{}^{-2})$. Increasing electrode surface area had no effect on impedance spectra for AIROF probes at low levels of activation. However, the effect of increasing GSA on the EIS of probes with varying AIROF thicknesses will need to be investigated further in future work. Impedance and phase spectra measured around + 0.3 V for different GSAs (Group III probes) activated to 100 total pulses each are shown in supplemental information figure S2.

3.3. Cyclic voltammetry profiles

3.3.1. Electrolyte-dependent characteristics

PBSair.

Cyclic voltammetry in PBSair at 0.05 V s⁻¹ showed increasing growth of iridium oxide film (AIROF) with increasing number of activation pulses. Figure 4(a) shows the 0.05 V s⁻¹ CV profile of a 2000 μm² Group I iridium probe before activation with a profile typical of iridium metal, and after activation to varying levels with reduction and oxidation (redox) peaks typical of AIROF. The primary redox peaks are assigned to transitions between valence states Ir³⁺/Ir⁴⁺ and Ir⁴⁺/Ir⁵⁺ and continue to increase in amplitude with additional activation pulses due to increasing AIROF thickness [21, 25, 42, 45]. Cyclic voltammetry at a higher sweep rate of 50 V s⁻¹ also showed the emergence of redox peaks associated with AIROF (figure 4(d)).

Figure 4.

Results from increasing activation of 2000 μm² iridium probes (Groups I and IV). Panels a, b, d, and e show example cyclic voltammetry (CV) profiles from activation of a single iridium probe (Group I). (a) and (b) show data before activation and after 100, 500, and 1000 total activation pulses measured at 0.05 V s⁻¹ in air-equilibrated phosphate buffered saline (PBSair) and model interstitial fluid (mISF), respectively. (d) and (e) show data before activation and after 100 total activation pulses measured at 50 V s⁻¹ in PBSair and mISF, respectively. Cathodic charge storage capacities calculated for all Group I ($n=4$, avg. ± std. dev.) and Group IV ($n=4$, avg. ± std. dev.) probes are shown in (c) for 0.05 V s⁻¹ CVs and (f) for 50 V s⁻¹ CVs. The maximum charge injection capacities obtained from voltage transient measurements are shown in (g) and (h) where the interpulse bias was set at 0.0 V vs. Ag|AgCl (g) or at + 0.6 V vs. Ag|AgCl (h). Error bars represent standard deviation.

PBSargon.

The CV profile in PBSargon (supplemental information figure S3) did not differ greatly from PBSair. Typically, differences in oxygen reduction reactions in PBSargon vs. PBSair will be more apparent in CV profiles of an iridium metal surface than an AIROF surface, so changes in AIROF CV profiles were not present in PBSargon, as expected.

mISF.

The CV profile in mISF was markedly different than in PBS due to higher overpotentials associated with the lower buffering capacity and conductivity of the mISF. The position of oxidation peaks $\left(E_{\text{p-ox}}\right)$ were shifted to a more positive potential and the position of reduction peaks $\left(E_{\mathrm{p}\text{-red}}\right)$ to a more negative potential compared to PBSair and PBSargon. The shift in ${\mathrm{E}}_{\mathrm{p}}$ also increased with increasing activation. Figure 4(b) shows the 0.05 V s⁻¹ CV profile and figure 4(e) shows the 50 V s⁻¹ CV profile observed in mISF for the same probe at the same activation levels shown for PBSair.

3.3.2. Observations from cyclic voltammetry

Early investigations of iridium oxide observed the same general CV profile described in this work and in prior work by several others [46–48], with minor differences that can be ascribed to differences in electrolyte compositions and testing conditions. In our study, CV measurements of some probes showed the appearance of an additional oxidation peak when probes were at or above 150 activation pulses, without the appearance of an additional corresponding reduction peak. When present, the additional oxidation peak was readily observed in PBSair and PBSargon but was not typically discernible in mISF as a distinct peak. The appearance of the oxidation peak at 0.3 to 0.6 V vs. Ag|AgCl was observed in CV profiles of AIROF probes at varying levels of activation and was not obviously correlated with a specific number of activation pulses. However, SEM images of probes that exhibited this change in CV profile suggest that the appearance of the additional oxidation peak coincided with the occurrence of fracturing and sometimes flaking of the iridium oxide film (see supplemental information figure S4 and S5).

Similar observations of CV profile changes were reported previously for AIROF [20, 26, 49], and multiple mechanisms could potentially lead to fracture of the hydrous iridium oxide film. Fracturing, for example, could be caused by the evolution of H₂ gas from the underlying iridium metal surface in locations where polarization exceeds the water reduction limit (−0.6 V vs. Ag|AgCl) during pulsing due to a non-uniform distribution of the potential over the probe surface. Future work could usefully investigate mechanisms that lead to fracturing of the AIROF, and whether or not fracturing is a normal change in film micro-structure or leads to delamination of the film over time.

3.4. Charge storage capacity

Group I cathodic charge storage capacity $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\right)$ continuously increased with increasing number of activation pulses. No differences were seen between ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ in PBSargon and ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ in PBSair, implying minimal contribution of oxygen reduction to the overall charge transfer in pH ~ 7.2 PBS solution. ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ in mISF calculated from 0.05 V s⁻¹ CVs was unexpectedly higher than in PBSair and PBSargon. ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ in mISF calculated from 50 V s⁻¹ CVs, however, was lower than ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ in PBSair and PBSargon as expected. Changing the order of testing (mISF then PBS versus PBS then mISF) did not change this trend. Typically, peak CV current amplitudes are expected to be lower in mISF than PBS due to the lower conductivity and lower buffering capacity of the mISF, which also account for shifts in peaks for both oxidation $\left(E_{\text{p-ox}}\right)$ and reduction $\left(E_{\text{p-red}}\right)$ with increasing activation. Figure 4(c) and figure 4(f) show ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ vs. number of activation cycles for Group I probes in PBSair, PBSargon, and mISF calculated from 0.05 and 50 V s⁻¹ CV measurements, respectively. Group II probes exhibited the same trend as Group I probes for ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ at 0 to 200 total activation pulses.

Group IV probes reached higher ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ values than Group I probes activated for the same number of total pulses. Group I probes were activated in multiple stages while Group IV probes were activated to 1000 total pulses in one step. Group IV (2000 μm² GSA probes) ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ values calculated from 0.05 V s⁻¹ CVs were 27.8 ± 6.4 mC cm⁻² in PBSair and 33.3 ± 4.8 mC cm⁻² in mISF (figure 4(c)). Group IV ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ values calculated from 50 V s⁻¹ CVs were 21.0 ± 1.0 mC cm⁻² in PBSair and 16.8 ± 0.5 mC cm⁻² in mISF (figure 4(f)).

Group III charge storage capacities, which are normalized to GSA, generally did not change with increasing surface area, and statistical analysis did not find a significant difference between ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ values versus electrode GSA. Figures 5(a)–(b) show ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}$ calculated from 0.05 V s⁻¹ and 50 V s⁻¹ CVs for 2000 μm², 3000 μm², and 4000 μm² probes after 100 activation pulses. We expect very little effect of GSA on charge storage capacity within the limited range tested here at such a small number of activation pulses. Additional testing of a larger number of electrodes or testing AIROF probes at a higher number of activation pulses may reveal effects of GSA on charge storage.

3.5. Charge injection capacity

The maximum charge injection capacity ${(Q}_{\text{inj}})$ of 2000 μm² (Group I) AIROF electrodes continued to increase with increasing activation up to 1200 total pulses. The $Q_{\text{inj}}$ at 0 V bias was highly variable but showed a generally increasing trend with increasing activation and is shown in figure 4(g). Maintaining a + 0.6 V bias vs. Ag|AgCl between current pulses increased $Q_{\text{inj}}$ from 0.35 mC cm⁻² (with 0 V bias) to more than 2.0 mC cm⁻² (i.e. 40 nC/phase for 200 μs pulses) in PBS and mISF after the first 100 activation cycles as shown in figure 4(h).

At 200 activation pulses, $Q_{\text{inj}}$ measured with + 0.6 V bias was 1.92 ± 0.2 mC cm⁻² in mISF and was significantly lower than $Q_{\text{inj}}$ in PBSair and PBSargon (2.84 ± 0.71 mC cm⁻² and 2.81 ± 0.27 mC cm⁻², respectively). At 300 total activation pulses, $Q_{\text{inj}}$ measured with + 0.6 V bias began to exceed the compliance limit of the stimulator used for data collection, which supplied a maximum of approximately 320 μA. Once $Q_{\text{inj}}$ exceeded the stimulator supply, data were recorded at the stimulator’s maximum current output and the decrease in electrode polarization ${(E}_{\text{mc}})$ was tracked for the remaining activation steps (supplemental information figure S6).

$Q_{\text{inj}}$ was relatively insensitive to changes in the electrode surface area, as shown in figures 5(c)–(d) as a function of GSA for VT measurements with 0 V bias and + 0.6 V bias. As expected, $Q_{\text{ph}}$ increased with increasing surface area. Increasing GSA from 2000 μm² to 3000 μm² increased the maximum charge per phase that could be applied in mISF (with + 0.6 V bias pulsing) from 30.3 ± 4.8 nC/phase to 47.3 ± 9.4 nC/phase (figure 5(d)). Maximum ${\mathrm{Q}}_{\text{ph}}$ for 4000 μm² probes with + 0.6 V bias pulsing was not captured as the value was outside the available compliance of the stimulator.

3.6. Utilization

The percent utilization is a measure of the fraction of total available charge stored by the electrode that can be delivered with a short duration current pulse (here, 200 μs). Conditions where $Q_{\text{inj}}$ was not recorded due to limited compliance of the stimulator were excluded from percent utilization calculations. We found that percent utilization decreased with increasing activation of the iridium probes. Our results are consistent with expected charge-transfer phenomena in porous materials like AIROF, where it becomes more difficult to access charge from deep layers of the film during a 200 μs pulse as film thickness increases. Utilization $\left(Q_{\text{inj}}÷{\mathrm{C}\mathrm{S}\mathrm{C}}_{600}\right)$ was 35% (Group I) to 48% (Group III) in PBSair and 22% (Group I) to 36% (Group III) in mISF for 2000 μm² probes at 100 activation pulses. Increasing activation to 300 total pulses (Group I probes) reduced utilization to 29% in PBSair and 18% in mISF. Further, Group IV probes activated to 1000 total pulses showed utilization of only 23% in PBSair and 4% in mISF.

Our study also found that the measured maximum charge injection capacity ${(Q}_{\text{inj}})$ was 25%–37% of ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$ calculated from high sweep rate 50 V s⁻¹ CVs recorded in PBSair ($Q_{\text{inj}}$ 32%–37% of ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$ in mISF), for probes activated to 1000 total pulses. For probes activated to 100–300 total pulses, the measured ${\mathrm{Q}}_{\text{inj}}$ is typically 44%–58% of ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$ from 50 V s⁻¹ CVs in PBSair and 43%–55% of ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$ from 50 V s⁻¹ CVs in mISF. Therefore, ${\mathrm{C}\mathrm{S}\mathrm{C}}_{600}$ calculated froma 50V s⁻¹ CV can provide a measure of expected $Q_{\text{inj}}$ for AIROF electrodes if the AIROF thickness is similar across probes activated in the same manner.

3.7. Electrode potential

Monitoring maximum negative potential ${(E}_{\text{mc}})$ of the electrodes during constant-current pulsing is important for avoiding water electrolysis in saline and in vivo, where exceeding the potential limits for water oxidation and reduction can damage the electrodes or lead to damage of the surrounding tissue [50]. Utilizing a + 0.6 V interpulse bias vs. Ag|AgCl significantly reduced the $E_{\text{mc}}$ of AIROF electrodes during pulsing. Figure 6(a) shows the increasing magnitude of the $E_{\text{mc}}$ of AIROF probes when applying cathodic current pulses at increasing levels of charge ${(Q}_{\text{ph}})$. For all levels of $Q_{\text{ph}}$, the magnitude of $E_{\text{mc}}$ decreased with increasing GSA. Multiple regression analysis showed a significant difference $(\mathrm{p}<0.01)$ in $E_{\text{mc}}$ for each increase in surface area. However, AIROF probes of all three GSAs remained positive of the −0.6 V limit up to 20 nC/phase when pulsing with a + 0.6 V bias. With a 0.0 V bias, pulsing above 2 nC/phase was not possible as $E_{\text{mc}}$ exceeded the −0.6 V limit for avoiding water reduction at higher charge levels.

Charge densities associated with the applied $Q_{\text{ph}}$ are shown in figure 6(c) for each GSA tested, where the highest charge density was 1.0 mC cm⁻². The ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\text{c}}$ in mISF for Group III probes (shown in figures 5 and 6) was 6.31 ± 1.03, 5.30 ± 1.33, and 7.41 ± 1.12 mC cm⁻² for 2000, 3000, and 4000 μm² probes respectively. Maximum charge injection limits of 2.5 mC cm⁻² and 2.0 mC cm⁻² were reported for AIROF with much higher ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\text{c}}$ (20 to 60 mC cm⁻²) pulsed in buffered saline from + 0.8 V bias vs. SCE [6] and from + 0.6 V bias vs. Ag|AgCl [7] respectively.

3.8. Driving voltage

VT measurements were recorded at increasing levels of charge per phase ${(Q}_{\text{ph}})$ for AIROF electrodes with different surface areas (Group III). The driving voltage ${(V}_{\text{drive}})$ required, which is particularly important in determining power requirements for wireless devices, was less than 1.5 V up to the maximum applied charge of 20 nC/phase (100 μA, 200 μs pulse) for all probes. Increasing the electrode surface area significantly reduced the driving voltage required to deliver the same level of current. Figure 6(b) shows the driving voltage required to deliver 2, 4, 8, 12.6, 16, and 20 nC/phase for three different electrode GSAs using 200 μs pulses at 10, 20, 40, 63, 80, and 100 μA respectively. Multiple regression analysis showed a negative correlation between ${\mathrm{V}}_{\text{drive}}$ and surface area with a coefficient of − 0.19 V/1000 μm² GSA (95% CI: − 0.22, − 0.16; $\mathrm{p}<0.001$; adjusted $R^2=0.909;N=72$ observations). As expected, $E_{\text{mc}}$ was positively correlated with GSA, showing reduced negative polarization with increasing GSA, with a coefficient of 0.14 V/1000 μm² GSA (95% CI: 0.11, 0.16; $\mathrm{p}<0.001$; adjusted $R^2=0.850;N=72$ observations).

4. Discussion

Potential cycling of iridium metal has consistently produced high charge storage capacity AIROF electrodes in a variety of electrolytes [6, 34, 51, 52], but is a relatively slow process. Further, potentiostatically holding iridium electrodes at potentials between $E_{\text{a}}$ and $E_{\text{c}}$ does not significantly contribute to formation of an oxide layer [53], and cycling between sufficiently oxidizing ${(E}_{\text{a}})$ and reducing ${(E}_{\text{c}})$ potentials is necessary for iridium oxide film growth [34, 46]. We employed potential pulsing as a high-rate process for AIROF formation [46, 54]. In addition, we also employed relatively long hold times at each potential limit similar to those employed previously in acidic [46] and neutral [12] pH solutions.

4.1. Anodic biasing during stimulation

AIROF electrodes are capable of providing charge injection levels suitable for most neural stimulation applications only when used with anodic biasing. Iridium oxide films (including SIROF and EIROF) benefit from the use of anodic biasing strategies for improving charge injection and for mitigating excessive electrode polarization that could damage the electrode or contribute to stimulation-induced tissue damage. Increasing the interpulse bias from 0.0 V to + 0.6 V increases the oxidation state of the iridium making the iridium oxide more electronically conductive [9, 49]. And so, the increased electronic conductivity in the film at + 0.6 V bias results in reduced ohmic-voltage drops in the electrolyte or AIROF film (access resistance and $V_{\text{acc}}$) and reduced driving voltage ${(V}_{\text{drive}})$. Our study found that access resistance (up to 200 activation pulses) for probes pulsed from + 0.6 V bias was approximately equivalent to high frequency impedance measured in PBSair (figure 3(c), 2.2 to 2.7 kΩ).

AIROF electrodes observed during overstimulation studies using 200 μs current pulses at 60 nC/phase (3 mC cm⁻²) applied in both saline and in vivo environments showed that when the electrode polarization exceeded the − 0.6 V limit for water reduction there was significant damage to both the electrode as well as the surrounding tissue [26]. And so, the use of anodic biasing is essential to allowing high levels of charge injection with AIROF electrodes. Despite the known benefit of anodically biasing iridium oxide [7, 49, 55], no study has looked directly at the long-term effects of applying a continuous anodic bias either in saline solutions or in vivo. However, McCreery et al [4] used AIROF electrodes in a chronic stimulation study in cats implanted for 450 to 1282 d, with no major pulsing-induced damage of neural tissue within 150 μm of electrodes pulsed with anodic bias at 100% duty cycle, 2 nC/phase (100 μC cm⁻²), 50 Hz, 8 h per day for 30 d, or with the same parameters at 50% duty cycle.

4.2. Influence of electrode surface area

Previous studies with SIROF electrodes have shown a decrease in the area-normalized $Q_{\text{inj}}$ with increasing electrode area [40]. We expect AIROF electrodes tested over the same size range (1960 to 125 600 μm²) would follow a similar trend. So, the ability of AIROF macroelectrodes (on the order of a few mm²) to inject charge is likely to be much less than AIROF microelectrodes on a per-unit-area basis. Therefore, the use of AIROF in applications using large surface area macroelectrodes may not provide the charge injection capacity required for neural stimulation.

Over the narrow range of surface areas investigated in this study we did not observe a dependence of AIROF $Q_{\text{inj}}$ on GSA (figures 5(c)–(d)). We found that increasing the GSA from 2000 μm² to 3000 μm² did not provide a significant increase in either the charge storage capacity ${(\mathrm{C}\mathrm{S}\mathrm{C}}_{\text{c}})$ or the charge injection capacity ${(Q}_{\text{inj}})$ of the activated iridium oxide films. Increasing the GSA did, however, significantly increase the deliverable $Q_{\text{ph}}$ and reduced the overall electrode polarization ${(E}_{\text{mc}})$ and driving voltage ${(V}_{\text{drive}})$ required during pulsing. The effect of GSA on $E_{\text{mc}}$ and $V_{\text{drive}}$ is seen in figures 6(a)–(b), respectively. From the perspective of available stimulator driving voltage, increasing GSA by 1000 μm² reduces the required $V_{\text{drive}}$ by about 0.19 V over a range of charge levels of 2–20 nC/phase, although the data suggest some deviation from linearity between $Q_{\text{phase}}$ and $V_{\text{drive}}$ at the higher charge levels (see figure 6(b)). The lower $V_{\text{drive}}$ at higher GSA levels arises from both a decrease in electrolyte resistance and reduction in electrode polarization. The maximum negative polarization of the AIROF ${(\mathrm{E}}_{\text{mc}})$ is decreased by about 0.14 V for each 1000 μm² increase in GSA from 2–20 nC/phase and this is again expected because of the overall reduction in current density at the electrode. Adopting a GSA of 4000 μm², which maximizes the reduction in $V_{\text{drive}}$ and $E_{\text{mc}}$ for the range of GSA investigated in this study, is attractive from the perspective of stimulation safety (less negative $E_{\text{mc}}$) and avoiding stimulator compliance limits on the maximum deliverable charge (reduced $V_{\text{drive}}$). However the effect of increasing electrode size to 4000 μm² on stimulation selectivity is unknown and would need to be investigated in an in vivo study should the larger GSA electrodes be preferred for their lower $V_{\text{drive}}$ and reduced $E_{\text{mc}}$. For this reason, our future clinical studies with AIROF electrodes will have a nominal GSA of 3000 μm².

With commonly used 2-dimensional circular electrodes, increasing the GSA from 2000 μm² to 3000 μm² requires increasing the electrode diameter by ~ 11 μm. With the 3-dimensional electrodes used in this study, increasing the GSA from 2000 μm² to 3000 μm² is achieved by removing an additional ~ 30 μm of Parylene-C insulation along the length of the probe. For this reason, the volume of tissue stimulated with the conical penetrating probes may be more sensitive to GSA increases than the planar electrodes. In any case, the overall impact of the trade-off between improved electrochemical performance with increasing GSA and any potential loss of spatial selectivity remains to be evaluated.

4.3. Setting an iridium activation protocol

Potential pulsing methods generally produce iridium oxide films at a faster rate than potential cycling methods [46]. Regardless of the method used to grow the iridium oxide film, we have found that charge for storage capacity alone is not sufficient determining the necessary number of iridium activation cycles. While charge storage capacity will increase with increasing activation and consequent thickness of the AIROF, percent utilization of the available charge decreases, at least with short duration stimulation pulses. Continued activation, therefore, does not provide appreciable gains in charge injection during neural stimulation [6, 25] and risks increasing the susceptibility of the iridium oxide film to delamination or mechanical failure. Further, there was very little difference in ${\mathrm{C}\mathrm{S}\mathrm{C}}_{\text{c}}$ for 0.05 V s⁻¹ versus 50 V s⁻¹ CVs for probes at 100–200 activation pulses in PBSair (figure 4(c) vs. 4(f)). This implies that less activation allows utilization of almost all the AIROF for charge transfer in PBS even at the comparatively high 50 V s⁻¹ scan rate.

Our findings are in agreement with prior work investigating charge injection and access resistance in various electrolyte solutions [23], and suggest that charge storage capacity and $Q_{\text{inj}}$ in capacity values measured in mISF are closer (although still not equivalent) to expected in vivo performance than values measured in phosphate-buffered saline solutions. Our study found that AIROF electrodes can provide levels of charge injection sufficient for neural stimulation threshold requirements after only 100–200 activation pulses. Since we expect some decrease in charge injection capacity of electrodes in vivo compared to mISF, future work will need to specifically investigate the charge injection capabilities of AIROF probes with varying amounts of activation during chronic in vivo use.

4.4. Study limitations

Since outcomes of studies on AIROF electrodes are highly dependent on geometry and surface area of the electrode, it may be difficult to directly translate results of this study to other sizes and shapes of electrodes. In particular, behavior of AIROF electrodes at ultramicroelectrode (less than ~ 200 μm²) or macroelectrode (greater than ~ 2 mm²) sizes may differ significantly from that of AIROF microelectrodes. This study also did not examine charge injection capacity and material stability during long-term pulsing or within the in vivo environment. Long-term pulsing in PBS can be useful to predict material stability during chronic device implantation, and future work will assess AIROF probes during chronic pulsing in saline solutions. Performance in vivo can vary from subject to subject, and so similar measurements will need to be repeated while electrodes are in tissue to determine if the same overall relationships hold true. Further, this study did not extensively examine the effect of stopping and starting activation cycles on the structure and long-term stability of AIROF electrodes. And so future work should include an examination of any performance differences between AIROF formed in one activation process compared to AIROF formed by multiple activation steps.

5. Conclusion

Comparatively thin iridium oxide films produced by a low number of activation cycles can achieve $Q_{\text{inj}}$ capacities required for in vivo neural stimulation. However, it remains to be determined how the use of a thin $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\le 12\mathrm{m}\mathrm{C}\mathrm{c}\mathrm{m}{}^{-2}\right)$ AIROF versus a thick $\left({\mathrm{C}\mathrm{S}\mathrm{C}}_{\mathrm{c}}\ge 30\mathrm{m}\mathrm{C}\mathrm{c}\mathrm{m}{}^{-2}\right)$ AIROF may affect chronic stability of the metal-electrode surface during continued constant-current pulsing. Additional work is also needed to fully understand what factors may affect film integrity during chronic in vivo neural stimulation. In particular, additional testing is needed to determine the relationship between film thickness and chronic stability, as well as what stimulation parameters may induce morphological change or fracture of iridium oxide films under high levels of charge injection.

We found that the combined use of charge storage capacity and charge injection capacity is beneficial when determining the required number of activation cycles for iridium electrodes. Additional activation of iridium beyond 400 cycles did not provide significant gains in charge injection capacity in model saline solutions. Electrodes with a slightly larger surface area (3000 μm²) are better suited to deliver clinically relevant levels of charge while mitigating excessive electrode polarization and driving voltage requirements. While increasing the surface area compromises some degree of in vivo selectivity, use of larger electrodes will better maintain electrode parameters within a feasible range for implantable wireless stimulators.