Direct measurement of oxygen reduction reactions at neurostimulation electrodes

Journal of Neural Engineering PAPER • OPEN ACCESS Direct measurement of oxygen reduction reactions at neurostimulation electrodes To cite this article: Jií Ehlich et al 2022 J. Neural Eng. 19 036045 View the article online for updates and enhancements. You may also like Extension of the ECRH operational space with O2 and X3 heating schemes to control tungsten accumulation in ASDEX Upgrade H. Höhnle, J. Stober, A. Herrmann et al. - Measurement of the O2O3O4 and O3O4O5 super Coster–Kronig rates in tungsten via asymmetric diffraction spectrometry John Seely, Jack L Glover, Lawrence Hudson et al. - LIGO detector characterization in the second and third observing runs D Davis, J S Areeda, B K Berger et al. - This content was downloaded from IP address 130.236.88.56 on 27/06/2022 at 15:15 J. Neural Eng. 19 (2022) 036045 https://doi.org/10.1088/1741-2552/ac77c0 Journal of Neural Engineering OPEN ACCESS RECEIVED 23 December 2021 REVISED 1 June 2022 ACCEPTED FOR PUBLICATION 10 June 2022 PUBLISHED 27 June 2022 Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. PAPER Direct measurement of oxygen reduction reactions at neurostimulation electrodes Jiˇ rí Ehlich1,4, Ludovico Migliaccio1,4, Ihor Sahalianov1, Marta Niki´ c1,2, Jan Brodský1, Imrich Gablech1, Xuan Thang Vu3, Sven Ingebrandt3and Eric Daniel Głowacki1,∗ 1Bioelectronics Materials and Devices Laboratory, Central European Institute of Technology CEITEC, Brno University of Technology, Purkyˇ nova 123, 61200 Brno, Czech Republic 2Institute of Neuroelectronics, Technical University of Munich, Munich, Germany 3Institute of Materials in Electrical Engineering 1, RWTH Aachen University, 52074 Aachen, Germany 4These authors are contributed equally. ∗Author to whom any correspondence should be addressed. E-mail: eric.daniel.glowac[email protected].cz Keywords: bioelectronics, neurostimulation, faradaic reactions, hypoxia, reactive oxygen species Supplementary material for this article is available online Abstract Objective. Electric stimulation delivered by implantable electrodes is a key component of neural engineering. While factors affecting long-term stability, safety, and biocompatibility are a topic of continuous investigation, a widely-accepted principle is that charge injection should be reversible, with no net electrochemical products forming. We want to evaluate oxygen reduction reactions (ORR) occurring at different electrode materials when using established materials and stimulation protocols. Approach. As stimulation electrodes, we have tested platinum, gold, tungsten, nichrome, iridium oxide, titanium, titanium nitride, and poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate). We use cyclic voltammetry and voltage-step amperometry in oxygenated versus inert conditions to establish at which potentials ORR occurs, and the magnitudes of diffusion-limited ORR currents. We also benchmark the areal capacitance of each electrode material. We use amperometric probes (Clark-type electrodes) to quantify the O2and H2O2concentrations in the vicinity of the electrode surface. O2and H2O2concentrations are measured while applying DC current, or various biphasic charge-balanced pulses of amplitude in the range 10–30 µC cm−2/phase. To corroborate experimental measurements, we employ finite element modelling to recreate 3D gradients of O2and H2O2.Main results. All electrode materials support ORR and can create hypoxic conditions near the electrode surface. We find that electrode materials differ significantly in their onset potentials for ORR, and in the extent to which they produce H2O2 as a by-product. A key result is that typical charge-balanced biphasic pulse protocols do lead to irreversible ORR. Some electrodes induce severely hypoxic conditions, others additionally produce an accumulation of hydrogen peroxide into the mM range. Significance. Our findings highlight faradaic ORR as a critical consideration for neural interface devices and show that the established biphasic/charge-balanced approach does not prevent irreversible changes in O2concentrations. Hypoxia and H2O2can result in different (electro)physiological consequences. 1. Introduction Electrical neurostimulation is an established part of numerous implantable bioelectronics devices like deep brain stimulators [1], peripheral nerve stimulation devices [2], spinal cord stimulators [3], and retinal prosthetics [4,5]. Electrical stimulation involves charge injection from an electrode surface into physiological electrolyte. The safety and reliability of this process is a vital consideration. Mechanisms of charge exchange at the electrode/electrolyte interface are the subject of extensive research, and can be divided into capacitive, pseudo-capacitive, and faradaic [6–8]. Capacitive charge injection © 2022 The Author(s). Published by IOP Publishing Ltd J. Neural Eng. 19 (2022) 036045 J Ehlich et al involves the charging and discharging of electrolytic double-layers, and no charge is transferred to species in solution. The pseudo-capacitive mechanism, also known as pseudo-faradaic, features redox reactions of the electrode material itself. These reactions can result in high density of charge transfer, and this transfer can be highly reversible. The third category is faradaic, where charge is transferred to a species in solution via a redox process occurring at the electrode/electrolyte interface. This process may or may not be reversible, depending on kinetic factors like activation barriers (aka overpotentials) for a given reaction as well as diffusion of reactants/products. A stimulation electrode works by injecting current into a physiological medium and the resultant electric fields modulate the membrane potential of nearby excitable cells. For instance, a cathodic pulse will efficiently depolarize cell membranes and activate voltage-gated sodium channels, triggering action potentials. Cathodic, i.e. negative currents, have been long recognized as being most efficient at eliciting action potentials. However, it follows that charge artificially injected into a physiological environment should be subsequently removed, that is no net charging of the system should occur. This is because net charge remaining would correspond to polarization of the electrode, or to electrochemical changes in the biological surrounding itself. These changes could potentially be toxic or otherwise spurious. For this reason, virtually all neurostimulation protocols for both basic research and implanted biomedical devices rely on biphasic operation [6]. Charge-balanced, cathodic-leading pulses are the standard [9]. This way, the total injected cathodic charge in the first phase is equalized by an equal-charge anodic phase. In principle, any electrochemical reaction products formed on the electrode during the leading phase should be reoxidized during the second, anodic, phase [7,10]. Determining safe limits for charge injection is the topic of debate and several empirical norms have been suggested for various in vivo applications. A number of studies have considered the reversibility of faradaic electrochemistry during biphasic pulsing [11]. Nearly all of these studies focus on testing corrosion of the metal electrode itself, or voltage excursions beyond the water-splitting window, and thus hydrogen evolution or oxygen evolution reactions [7,11,12]. The possibility of oxygen reduction, occurring during the cathodic stimulation phase, has received relatively little attention [13,14], despite the fact that thermodynamically it is much morefavoured than hydrogen evolution (by at least 1.23 V). Oxygen is present in all physiological fluids. While it is transported in heme-bound form in the vasculature, in excitable neural tissues, oxygen travels through the extracellular and intracellular space by passive diffusion. Neural tissue oxygenation levels vary depending on species, anaesthesia, location. The highest possible equilibrium concentration of dissolved O2is about 250 µM. This is the same level of oxygenation as what is expected for a container of water open to ambient 21% atmospheric oxygen. This condition also applies for most in vitro experiments. We therefore perform this study with this 21% reference point, since it applies to in vitro conditions and with respect to in vivo this assumption mirrors the ‘best case’ scenario. To our knowledge, there are four published studies which consider oxygen reduction reactions (ORR) on neurostimulation electrodes, primarily on Pt and Au [13–16]. These all rely on some form of transient electrochemical measurement techniques to estimate irreversible charge transfer to O2, and all these studies agree that a substantial fraction of cathodic current can be irreversibly transferred to oxygen (reported ranges between 5% and 80%). None of these studies quantified the generation of peroxide via two-electron reduction, or actual oxygen concentration changes. In contrast to these works, we have designed our study to probe both O2and H2O2concentrations in the vicinity of the electrode surface to quantify the effects of ORR directly, and have used the same technique to compare eight different neurostimulation electrode materials we have prepared in thin-film from: Ti, TiN, Au, IrOx, Pt, W, NiCr, and poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate), shortened as PEDOT:PSS. This list is chosen it represents common electrode materials used in both neuroscience/electrophysiology research as well as biomedical devices. In addition, a sample of commercial high-surface area TiN was tested [17]. This way, we can establish to what extent irreversible ORR can occur at neurostimulation electrodes, and compare and contrast different electrode materials. To corroborate experimentally-measured O2and H2O2 concentrations, we have also established finite element simulations to understand the geometry of concentration gradients over time. 2. Materials and methods 2.1. Model stimulation electrode preparation Microscope slides (3 ×1) inch2were cleaned according to established methods, treated with oxygen plasma, and then sputter coated with a 100 nm layer of Ti using a Kaufman ion-beam source (IBS). The Ti acts as the common conducting layer below all studied samples, as it has excellent adhesion on glass and is a suitable underlayer for all the studied materials. Platinum (60 nm) is deposited using DC magnetron sputtering. W (60 nm) was deposited using the same sputtering system. TiN (60 nm) is reactively sputtered from a Ti target using two Kaufman IBSs, and will be referred to in this article as IBS_TiN. The primary IBS is used for sputtering from Ti target employing Ar and N2plasma, while the secondary 2 J. Neural Eng. 19 (2022) 036045 J Ehlich et al IBS is used for substrate bombardment with ions from pure N2plasma. Au is also deposited using primary IBS, to a thickness of 60 nm. NiCr (60 nm) was prepared by the same primary IBS, using pure Ar. IrOxwas obtained via DC reactive magnetron sputtering in an Ar/O2plasma (100 nm) according to previous published methods [18]. PEDOT:PSS (PH1000 formulation from Clevios, plus 5 wt% ethylene glycol, 0.1 wt% 4-dodecylbenzenesulfonic acid and 1 wt% (3-glycidyloxypropyl)trimethoxysilane) was spin-coated at 3000 rpm and annealed for 1 h at 130 ◦C (giving roughly 100 nm thickness as measured by stylus profilometry). As a reference, we characterize commercial TiN used in multielectrode arrays (Multichannel Systems GmbH). These films are about 600 nm thick. To distinguish this TiN from our in-house prepared samples, we will refer to it as MCS_TiN. 2.2. Electrochemical cell A homemade custom electrochemical cell allowed correct positioning and characterization of the electrodes under investigation. The cell was made from clear 5 mm thick acrylic sheets and a microscope glass slide as a front window. The internal volume of the cell was 9.3 ml. Microscope slides with deposited model stimulation electrodes were cut to (1 ×1) inch2pieces and horizontally inserted through a tight opening on the side of the cell. Polydimethylsiloxane was used to fix and seal the sample in position and to prevent any electrolyte leakage. The exposed area of each electrode was masked using a 70 µm thick polyvinylchloride foil (Minitronic elektronik GmbH). The circular opening in the foil had a diameter of 3 mm, defining the electrode under test area (active electrode area =0.0706 cm2). The cell was equipped with a Pt wire coil as counter (active area ∼7 cm2) and an Ag/AgCl as reference electrodes, two openings provided access for the O2or H2O2 sensor and a teflon tube for O2/N2purging. The sensor was placed in vicinity of the exposed active electrode material (distance =200 µm). The counter electrode is roughly 5 mm away from this area, at the top of the cell. Care should be taken that this distance is sufficient to ensure that any eventual products on the counter electrode do not affect the measurement. The cell was mounted on a submicrometer-precision XYZ stage (ThorLabs). Using a digital microscope (Q-SCOPE 20200-P), the sensor tip was positioned using XYZ stage to the point of light contact in the middle of the sample and then moved to the 200 µm distance in the Zdirection (figures 1(b) and S1). The cyclic voltammetry (CV) electrochemical characterization was carried out in a larger cell (from Redox.me) using an electrochemical active area of 1 cm2, in a three-electrode configuration having Ag/AgCl as reference electrode and Pt wire as counter electrode in a range of potentials using 0.1 M phosphate buffer saline (PBS) solution as electrolyte. 2.3. Electrochemical measurements—DC and AC conditions CV and chronoamperometry (DC conditions) were applied using an Ivium PocketSTAT2 potentiostat. For biphasic pulsing, a Digitimer DS4 biphasic constant current isolated stimulator was used, waveform was driven by a PicoScope 3404D oscilloscope with a built-in function generator. Transient voltages were recorded during biphasic pulsing by measuring voltage using the oscilloscope input (1 MΩinput impedance) between the electrode under test and an Ag/AgCl reference electrode. Data were collected in oxygenated and deoxygenated conditions. The relatively low input impedance is selected in order to provide a shunt resistance to prevent potential ratcheting and thus prevent the introduction of a DC bias offset voltage during AC pulsing experiments. 2.4. Clark electrode O2and H2O2quantification Local oxygen and peroxide concentrations were measured in situ during electrochemical measurements described above in section 2.2 using a four-channel microamperometric amplifier system (TBR4100, World Scientific Instruments), with fourchannel analog-digital converter board (LabTrax, World Scientific Instruments). The respective sensor probes used were ISO-HPO-2 and ISO-OXY-2. The O2sensor was kept constantly polarized at a bias of 700 mV, meanwhile the H2O2specific sensor at 450 mV. The sensors were always calibrated before the measurement of an individual material following the procedure reported in the instruction manual. The drop in O2concentration or increase in H2O2 was tracked by LabScribe software (World Scientific Instruments). It should be noted that the O2sensor functions along the classic Clark-electrode mechanism where oxygen is reduced at the sensing electrode, and oxygen reaches the sensor via an oxygenpermeable membrane. The peroxide sensor, on the other hand, operates via the oxidation of H2O2, and is therefore also cross-sensitive to dissolved H2(via the H2oxidation reaction). For experiments where cathodic polarization of the electrode under test results in H2evolution, the sensor signal can register a false positive H2O2signal. Caution should therefore be taken to test electrodes in deoxygenated electrolytes to establish the cathodic water-splitting onset. Peroxide sensor readings should only be accepted within the range before cathodic water splitting begins. 2.5. Finite element analysis of ORR at stimulation electrodes Simulation of oxygen and peroxide diffusion was conducted with finite element method, implemented in COMSOL 5.5 software package and transport of diluted species module (www.comsol.com/productdownload). The experimental setup was reproduced within an axisymmetric 2D model. After the revolution, the 3 J. Neural Eng. 19 (2022) 036045 J Ehlich et al Figure 1. (a) Schematic of capacitive and faradaic charge transfer that can occur at an electrode interface during the cathodic phase. At a cathodically-polarized electrode, charge can accumulate in a capacitive double layer (Cdl), or can be transferred in a faradaic reaction to a species in solution (Zfx). Oxygen reduction reactions (ORR), Zf1−Zf3, are thermodynamically favoured over the hydrogen evolution reaction Zf4. Oxygen can undergo a four-electron, Zf1, or 2 +2 electron reduction pathway (Zf2+Zf3) to yield water as a product. The two-electron pathway Zf2produces hydrogen peroxide, while Zf3consumes hydrogen peroxide. All ORR pathways Zf1−Zf3, lead to depletion of oxygen concentration near the electrode, and Zf2can result in net accumulation of hydrogen peroxide. (b) Experimental setup for testing ORR occurring at an electrode interface. In this configuration, the voltage/current of the electrode under test is controlled in a three-electrode configuration with a potentiostat, or with a biphasic current stimulator. A Clark electrode sensor is used for probing the oxygen or peroxide concentration at a fixed position near the electrode under test (200 µm from the surface). The whole chamber is enclosed to allow experiments under air (21% O2), 100% O2, or 100% N2. Photographs of the setup with the adjustable position between the electrode under test and sensor are shown below the schematic. (c) DC mode protocol used for testing electrodes involves steps of cathodic potentials from 0 to −0.7 V vs. Ag/AgCl. These example results are for an Au electrode. Potential is held for 600 s and current is registered. ORR currents start at highly cathodic values and decrease to diffusion-limited equilibrium values within tens of seconds. Over the course of 600 s, the sensor is used to register the local O2or H2O2concentration. (d) AC mode involves charge-balanced cathodic-leading current pulses 250 µs per phase, 10 Hz modulation. Interpulse time is varied 0, 20, 50 µs; and three amplitudes are tested: 10, 20, 30 µC cm−2/phase. These example traces are recorded for Au electrodes. computational model obtained a cylindrical shape filled with water-based electrolytes (figure 5(a1)). Faradaic reactions occur on the area of a cathodic pixel located on the bottom. The diameter of pixel d=3 mm was adopted from the experiment. The water electrolyte domain spreads on 5 mm from the stimulation electrode, thus making the diameter of the electrolyte cylinder 13 mm and its height 5 mm (figure 5(a1)). The model contained two variables: cO2 and cH2O2, which goes for a concentration of dissolved 4 J. Neural Eng. 19 (2022) 036045 J Ehlich et al oxygen and hydrogen peroxide. Initial values were assigned as csaturated O2=300µM and cH2O2=0µM. We simulated both twoand four-electron reaction pathways separately. In case of two-electron case, we considered the possibility of further reduction of peroxide into water, or so called 2 +2 reaction pathway. ORR was modelled via the introduction [19] of the empirical faradaic efficiency function (1), which modifies fluxes of oxygen and peroxide: feff =f1+f2 Cnearpixel O2 Csaturated O2 ,(1) where Cnearpixel O2is oxygen concentration, measured at 10 nm from the stimulation electrode. The values of constants f1and f2range between 0 and 1 (f1+f2=1) and depend on the type of cathode. If the constant f2is bigger than zero, the resulting flux of hydrogen peroxide will decrease with oxygen depletion. The diffusion equations governed the change of oxygen and peroxide concentration dci dt+∇ · Ji=0,(2) Ji=−Di∇ci,(3) where igoes for O2and H2O2and Jiare fluxes of oxygen and peroxide. Diffusion coefficients at 23 ◦C are: DH2O2=1.8 ×10−9m2s−1;DO2=2.5 ×10−9m2s−1 [20,21]. Production and consumption of diluted O2and H2O2molecules were carried out through boundary fluxes and can be found with all other boundary conditions in figure 5(a2). The model considered oxygen reduction into hydrogen peroxide in ratio 1:1 with a possible correction by faradaic efficiency function. Boundary fluxes RO2and RH2O2were defined as in the recent work of Abdullaeva et al [19]: RH2O2=FI 2A×feff,(4) RO2=−FI 2A×(feff +1−feff 2),(5) where Iis applied current (different in case of DC or AC simulations), Fis Faraday constant, A=3.14 ×1.52mm2is the stimulation electrode area, feff is dimensionless faradaic efficiency function, defined by constants f1and f2and measured O2concentration Cnearpixel O2on the 10 nm distance from a production pixel. In the case of the four-electron ORR reaction pathway, H2O2does not participate in the process, and diffusion equations (2) and (3) is solved only for a concentration of oxygen with fluxes of O2molecules. Assuming the absence of the reversed reaction, oxygen consumption is defined by RO2=−FI 4A.(6) More details on the model construction can be found in the supplementary information appendix 1. 3. Results 3.1. CV characterization of electrodes CV was used to characterize both capacitive charging and faradaic process occurring at each electrode material in PBS solution (figure 2). The system has been oxygenated and de-oxygenated (N2gas flow) to monitor the electrochemical behaviour of the materials under three different conditions (21% O2 (atmospheric pressure), 100% N2and 100% O2). Comparing CV curves from de-oxygenated conditions versus oxygenated ones can reveal which peaks are originating from ORR. The faradaic/capacitive charging behaviour differs markedly depending on the electrode material, nevertheless in all cases it is possible to observe an increase in cathodic current corresponding to oxygenation. ORR is apparent in CVs for voltages lower than +100 mV. For each material, we chose a smaller voltage window to measure in a region where capacitive charging dominates, allowing estimation of the double-layer capacitance of each electrode material (table 1; supplementary figure S2). CV scans for capacitance determination were obtained using a 1 mV step and current averaging according to methods described by Weltin and Kieninger [22]. 3.2. DC chronoamperometry and direct measurements of ORR processes While CV is useful to screen for possible ORR and other faradaic reactions, it does not reveal the magnitude of diffusion-limited oxygen reduction currents, or quantify the concentration of ORR products. To accomplish this, we combine DC chronoamperometry experiments with simultaneous recording of oxygen and hydrogen peroxide concentrations at a fixed point near the stimulation electrode surface (at a height of 200 µm). In these experiments, we measure current over time during the application of a constant potential (from +0.1 V to −0.9 V, by 0.1 V steps, depending on the material investigated) over a defined time (600 s) on each different stimulation electrode. The protocol and example results are plotted in figure 1(C). At potentials which are too anodic to reduce oxygen at a given electrode material, no sustained chronoamperometric currents are measurable. Once the onset potential is reached, sustained cathodic ORR current is clear. Onset potentials for ORR processes can be found in table 1. For reference, 5 J. Neural Eng. 19 (2022) 036045 J Ehlich et al Figure 2. Cyclic voltammetry of neurostimulation electrodes in PBS solution. CVs of Ti, IBS_TiN, MCS_TiN, Pt, Au, NiCr, W, IrOxand PEDOT:PSS in 100% oxygenated (blue), 21% oxygenated (ambient air, black), and 0% oxygenated (100% N2purged, red) conditions. Scan rate =100 mV s−1; Potential, E, range +0.7 V to −0.9 V versus Ag/AgCl. Due to large differences between materials in terms of their capacitive and faradaic currents, the respective plots have different current density y-axes. Table 1. ORR process onset potentials for each electrode materials in PBS solution, and electrochemical double-layer capacitance estimated for CVs measured in a nonfaradaic potential window (CVs shown in figure S2). Onset potentials for ORR and cathodic water-splitting are estimated from the chronoamperometry voltage-step profile and corresponding amperometric confirmation of oxygen reduction/peroxide generation. Potentials are given versus Ag/AgCl, [Cl−]=0.12 M. Material Oxygen reduction onset (mV vs. Ag/AgCl) Hydrogen peroxide evolution onset (mV vs. Ag/AgCl) Hydrogen evolution reaction onset (mV vs. Ag/AgCl) Double-layer capacitance (µF cm−2) Ti −700 −700 −900 21 IBS_TiN −500 −600 −1400 22 MCS_TiN −500 −500 −1400 669 Pt +100 0a−700 97 Au −300 −300 −700 56 NiCr −400 −400 −1000 24 W−500 −600 −1000 69 IrOx0 0 −700 530 PEDOT:PSS −600 −600 −1600 197 aOnly trace amounts of peroxide detected. each sample is measured also in deoxygenated electrolyte to establish the cathodic onset potential for water splitting/H2evolution. These H2evolution onsets, also given in table 1, are always more negative than measured ORR potentials. Chronoamperometry reveals a peak cathodic ORR current which then decays to a steady-state cathodic current. We define the steady-state current value as the current measured at the t=600 s timepoint. The steady-state current as a function of applied potential for each material 6 J. Neural Eng. 19 (2022) 036045 J Ehlich et al Figure 3. DC chronoamperometry with voltage-step protocol and resultant O2drop and H2O2increase for each electrode material. All potentials are versus Ag/AgCl. Each point along the J(V) trace (black) is the equilibrium current at t=600 s. The red trace is the lowest measured %O2at a position 200 µm above the electrode surface over the 600 s period. A drop of −21% is complete deoxygenation, while 0% corresponds to a normal 21% oxygen saturation. The blue traces mark the [H2O2] in mM measured at each potential after 600 s, at a position 200 µm above the electrode surface. Peroxide recordings at voltages where cathodic H2evolution occurs on a given material are not plotted, since H2evolution interferes with the peroxide sensor. is plotted in the black traces shown in figure 3. Simultaneously to recording chronoamperometry, the O2/H2O2amperometric sensor is active and recording the respective O2or H2O2concentration value at a point in the electrolyte directly above the electrode-under-test. The measured values of O2or H2O2concentration at the t=600 s timepoint are shown in figure 3as the red and blue traces, respectively. A change of −21% O2corresponds to a situation where the sensor records 0% oxygen, i.e. the calibrated lowest limit of a fully-deoxygenated solution. In between each voltage-step, the electrolyte in the cell is replaced with fresh, oxygenated electrolyte. It is noteworthy that all electrode materials can produce quite hypoxic conditions, with some such as Pt, Au, and PEDOT:PSS reaching nearly complete deoxygenation at higher cathodic potentials. The materials differ in their ability to produce peroxide from ORR. Au and PEDOT:PSS, for instance, produce concentrations into the millimolar range, while Pt generates barelydetectable trace amounts of peroxide. The recorded chronoamperometric currents and accompanying oxygen and peroxide measurements used to construct the plots in figure 3can be found in supplementary figures S3–S11. 3.3. Direct measurements of ORR processes during AC stimulation protocols Using the same configuration for amperometric sensing of oxygen and peroxide, we next applied charge balanced cathodic-leading pulses to the electrodes under test. The pulse duration per phase of charge balanced cathodic leading pulses used during experiments is kept constant (250 µs) and the interpulse spacing is changed to have three different durations (0, 20, or 50 µs). We use a period of 100 ms (ƒ=10 Hz). We tested three charge density values: 10, 20, or 30 µC cm−2/phase. The total time for each AC experiment is 1200 s, compared with the DC chronoamperometry which was 600 s. This AC stimulation protocol and accompanying examples of measured amperometric transients are given in figure 1(d). The raw data of the amperometric traces of oxygen and peroxide over the course of AC pulsing are shown in figures S3–S11. In all cases, charge-balanced biphasic pulses led to drops in oxygen concentration comparable in magnitude to those found under the most cathodic DC conditions. The generation of peroxide as a byproduct varied based on material, with IrOxand Pt producing only trace quantities, and Au and PEDOT:PSS producing the most. Surprisingly, 7 J. Neural Eng. 19 (2022) 036045 J Ehlich et al Figure 4. Charge-balanced cathodic-leading biphasic pulses 250/x/250, at 10, 20, or 30 µC cm−2result in net oxygen reduction, and in the case of some materials, appreciable H2O2generation. The interpulse spacing, x, of 0, 20, or 50 µs has little effect on the outcome. The charge density also has minimal impact. The error bars represent one standard deviation after taking the mean of all conditions (three different charge densities ×three different interpulse spacings). (a) Peak O2change during the 1200 s period; (b) Total integrated area-under-curve of consumed O2over the course of the 1200 s period; (c) Peak peroxide concentration recording over the 1200 s period; (d) Total produced peroxide area-under-curve over 1200 s. While O2concentration values (a), (b) should be considered an accurate, quantitative result, the peroxide values in (c), (d) should be considered semi-quantitative, due to peroxide sensor cross-sensitivity to H2. Materials like Au do produce some amount of H2during the cathodic phase. the magnitude of the charge density applied in the pulse had little effect on the peak and total amounts of ORR observed. The interpulse spacing has a minimal contribution on the overall change in %O2or H2O2 concentration. Since both pulse amplitude and interpulse have no definitive impact on ORR levels, the data presented in figure 4feature the nine different AC conditions all pooled to give a mean for a given material, with the standard deviation between conditions expressed by the error bar. Oxygen depletion and peroxide generation are expressed by both peak values, as well as the area-under-curve (AUC), the integral of the measured amperometric signal over time (1200 s) signifying the total amount of oxygen reduced/peroxide generated. During biphasic current pulses, voltage transients were collected (figure S12). 4. Discussion 4.1. ORR on electrode materials during CV and DC measurements In this study, we have decided to investigate ORRs on commonly-used neurostimulation electrode materials, aiming to find out to what extent these reactions affect concentrations of dissolved oxygen and hydrogen peroxide near the electrode. We selected eight representative electrode materials which we fabricated in thin-film form. Additionally, as a ninth material we studied a commercial TiN sample with high charge-injection capacity. The logic behind this study was to first measure CV in oxygenated versus deoxygenated electrolytes. This kind of characterization unambiguously reveals the presence of ORR. Next, we performed step-voltage chronoamperometric measurements to establish the magnitude of equilibrium ORR current density that is possible at a given electrode material. During the chronoamperometric measurements, O2or H2O2concentration is registered in real time in the solution above the electrode surface. The magnitude of ORR currents, as well as the onset potentials, varied greatly between the samples, which is why the current density y-axes plotted in figures 2and 3are all different. These differences originate because of the electrocatalytic properties of each material with respect to ORR. Measured peroxide concentrations in this study varied over four orders of magnitude, from 1 µM to 10 mM. The use of a Clark-type chronoamperometric detection system is, to the best of our knowledge, the only method to measure peroxide concentrations over such a large 8 J. Neural Eng. 19 (2022) 036045 J Ehlich et al [12] Shepherd R K, Carter P, Dalrymple A, Enke Y L, Wise A K, Nguyen T, Firth J, Thompson A and Fallon J B 2021 Platinum dissolution and tissue response following long-term electrical stimulation at high charge densities J. Neural Eng. 18 036021 [13] Morton S L, Daroux M and Mortimer J T 1991 The role of oxygen reduction in electrical stimulation of nervous tissue Proc. Annual Conf. Engineering in Medicine and Biology Society vol 13 pp 552–3 [14] Cogan S F, Ehrlich J, Plante T D, Gingerich M D and Shire D B 2010 Contribution of oxygen reduction to charge injection on platinum and sputtered iridium oxide neural stimulation electrodes IEEE Trans. Biomed. Eng. 57 2313–21 [15] Musa S, Rand D R, Bartic C, Eberle W, Nuttin B and Borghs G 2011 Coulometric detection of irreversible electrochemical reactions occurring at Pt microelectrodes used for neural stimulation Anal. Chem. 83 4012–22 [16] Dijk G, Ruigrok H J and Connor R P O 2021 PEDOT:PSS-coated stimulation electrodes attenuate irreversible electrochemical events and reduce cell electropermeabilization Adv. Mater. Interfaces 82100214 [17] Egert U, Schlosshauer B, Fennrich S, Nisch W, Fejtl M, Knott T, Müller T and Hämmerle H 1998 A novel organotypic long-term culture of the rat hippocampus on substrate-integrated multielectrode arrays Brain Res. Protocols 2229–42 [18] van Ooyen A, Topalov G, Ganske G, Mokwa W and Schnakenberg U 2009 Iridium oxide deposited by pulsed dc-sputtering for stimulation electrodes J. Micromech. Microeng. 19 074009 [19] Abdullaeva O S, Sahalianov I, Ejneby M S, Jakešová M, Zozoulenko I, Liin S I and Głowacki E D 2022 Faradaic pixels for precise hydrogen peroxide delivery to control M-type voltage-gated potassium channels Adv. Sci. 92103132 [20] Wise D L and Houghton G 1966 The diffusion coefficients of ten slightly soluble gases in water at 10–60 ◦CChem. Eng. Sci. 21 999–1010 [21] van Stroe-biezen S A M M, Everaerts F M, Janssen L J J J and Tacken R A 1993 Diffusion coefficients of oxygen, hydrogen peroxide and glucose in a hydrogel Anal. Chim. Acta 273 553–60 [22] Weltin A and Kieninger J 2021 Electrochemical methods for neural interface electrodes J. Neural Eng. 18 052001 [23] Song C and Zhang J 2008 Electrocatalytic Oxygen Reduction Reaction PEM Fuel Cell Electrocatalysts and Catalyst Layers: Fundamentals and Applications ed J Zhang (London: Springer) pp 89–134 [24] Donahue M J, Sanchez-Sanchez A, Inal S, Qu J, Owens R M, Mecerreyes D, Malliaras G G and Martin D C 2020 Tailoring PEDOT properties for applications in bioelectronics Mater. Sci. Eng. R140 100546 [25] Mitraka E et al 2019 Electrocatalytic production of hydrogen peroxide with Poly(3,4-ethylenedioxythiophene) electrodes Adv. Sustain. Syst. 31800110 [26] Schaldach M, Hubmann M, Weikl A and Hardt R 1990 Sputter-deposited TiN electrode coatings for superior sensing and pacing performance Pacing Clin. Electrophysiol. 13 1891–5 [27] Katsounaros I, Schneider W B, Meier J C, Benedikt U, Biedermann P U, Auer A A and Mayrhofer K J J 2012 Hydrogen peroxide electrochemistry on platinum: towards understanding the oxygen reduction reaction mechanism Phys. Chem. Chem. Phys. 14 7384 [28] Kirino T 2000 Delayed neuronal death Neuropathology 20 95–7 [29] Lipton P 1999 Ischemic cell death in brain neurons Physiol. Rev. 79 1431–568 [30] Sies H and Jones D P 2020 Reactive oxygen species (ROS) as pleiotropic physiological signalling agents Nat. Rev. Mol. Cell Biol. 21 363–83 [31] Sies H 2017 Hydrogen peroxide as a central redox signaling molecule in physiological oxidative stress: oxidative eustress Redox Biol. 11 613–9 [32] Gamper N, Zaika O, Li Y, Martin P, Hernandez C C, Perez M R, Wang A Y C, Jaffe D B and Shapiro M S 2006 Oxidative modification of M-type K +channels as a mechanism of cytoprotective neuronal silencing EMBO J. 25 4996–5004 [33] Lodola F, Rosti V, Tullii G, Desii A, Tapella L, Catarsi P, Lim D, Moccia F and Antognazza M R 2019 Conjugated polymers optically regulate the fate of endothelial colony-forming cells Sci. Adv. 5eaav4620 15