Accelerated Characterization of Electrode‐Electrolyte Equilibration

Operational durability is poorly characterized by traditional (photo)electrocatalyst discovery workflows, creating a barrier to scale‐up and deployment. Corrosion is a prominent degradation mechanism whose thermodynamics depend on the concentration of corrosion products in electrolyte. We present an automated system for characterizing the equilibration of (photo)electrodes with dissolved metals in electrolyte for a given electrode, pH, and electrochemical potential. Automation of electrode selection, electrolyte preparation, and electrolyte aliquoting enables rapid identification of self‐passivating electrodes and estimation of the equilibrium dissolved metals concentrations. The technique is demonstrated for metal oxide photoanodes in alkaline electrolyte, where BiVO4 is found to continually corrode, in agreement the literature. An amorphous Ni−Sb−O photoanode is found to passivate with a Ni‐rich coating on the order of 1 monolayer with less than 1 μM total dissolved metals in electrolyte, demonstrating its suitability for durable photoelectrochemical operation. The automation and throughput of the instrument are designed for incorporation in accelerated electrocatalyst discovery workflows so that durability can be considered on equal footing with activity.


Introduction
Solar photoelectrochemical coupling of the anodic oxygen evolution reaction (OER) with cathodic fuel synthesis is a nascent electrochemical technology limited by both the durability and activity of (photo)electrocatalysts. [1] Traditional approaches to electrocatalyst research relegate durability to the systems-engineering phase of technology development.Corrosion is central to electrode durability and is a molecular-level process that is as fundamental to electrocatalyst performance as the targeted electrochemical molecular transformation, motivating a new regime in electrocatalysis science where corrosion processes are studied on equal footing with catalysis. [2,3]Beyond the underreporting of corrosion properties in the electrocatalysis literature, the field is also hampered by the challenges in experimental characterization of corrosion kinetics and thermodynamics.From the experiment-driven corrosion cartography of Pourbaix, [4] to modern computational assessment of Pourbaix diagrams, [5][6][7][8][9] Pourbaix analysis of a solid-state electrode considers the bulk thermodynamic driving force towards corrosion (or dissolution) at a given pH, applied bias, and electrolyte composition.The concentration of corrosion species, typically represented as the dissolved metals concentration, is central to this thermodynamic assessment.Accelerating experimental characterization of Pourbaix diagrams requires a platform with direct observation of the equilibration of operating electrodes with the dissolved metals in electrolyte.Herein, we report the accelerated durability screening system (ADSS), which addresses this need by measuring the time evolution of dissolved metals concentration in a miniature electrochemical reactor.ADSS is demonstrated herein for characterization of metal oxide photoanodes for solar fuels applications.
While the differences in corrosion behavior in polymer and liquid electrolytes remains a critical consideration in translating catalyst discoveries to commercializable systems, [10,11] discovery research for heterogeneous electrocatalysis is typically conducted in liquid electrolyte.Automated techniques for measuring and identifying electrochemical signatures of corrosion excel at identifying materials with small corrosion current, [12] but in situ characterization of electrocatalysts requires disambiguation of the desired electrochemical current and the undesired electrode dissolution.Online analytical characterization of electrolytes has been well demonstrated with electrochemical flow cells by injecting the electrolyte into an inductively coupled plasma with mass spectroscopy (ICP-MS) or optical emission spectroscopy (ICP-OES) detection. [13,14]Each technique measures corrosion products as well as dissolved or suspended particles, for example from nanoparticle detachment. [15][18][19] These techniques rapidly characterize corrosion under continuous supply of fresh electrolyte.Coupled with measurement of catalytic rates, metrics such as the stability-number [20] can quantify catalyst activity vs. corrosion, which has aided comparison of literature reports with various electrochemical cells and procedures.
To meet the decades-long operational time scales targeted by industrial electrolytic processes, the stability-number needs to above a million. [20]Such a condition can be reached if the electrode can equilibrate with dissolved metals species, wherein the corrosion rate equals the precipitation rate for each element in the electrocatalyst.This opportunity has been recognized in so-called self-healing catalysts, but these typically require more than 10 À 5 M of dissolved metals species, [21,22] which is a high concentration with respect to both the typical solubility limit of corrosion products and the levels of dissolved metals concentration that may compromise device durability. [23]Catalysts should therefore be co-designed for not only activity but also operational durability, which is most rapidly assessed by observing equilibration with a dissolved metals concentration. [2]he ADSS platform uses electrolyte recirculation combined with robotic aliquoting of the working electrolyte to rapidly charac-terize electrode corrosion and its propensity to equilibrate with dissolved metals species at sub-μM concentrations.

Results and Discussion
The ADSS platform (Figure 1a) enables testing of flat, compact electrodes via an o-ring seal to a 3-electrode electrochemical cell equipped with a bipolar membrane to mitigate crossover of dissolved metals species between the working and counter electrodes (Figure 1b).While the working electrode chamber volume is 0.3 mL with no headspace, an additional 13 mL recirculation reservoir with open headspace provides flexibility in the choice of electrolyte volume and enables exhaust of gasses such as the O 2 from the oxygen evolution reaction (OER) studied in the present work.The electrolyte volume and flow rate can be varied with typical values being 4 mL of working electrolyte recirculated at 0.2 mL/s, providing an electrolyte cycle time of 20 s.Workflow automation is implemented via HELAO-async [24] and includes automated exchange of working electrode via translation stages and automated liquid handling such that a series of experiments may be automatically The electrochemical cell is sealed to a selected location on the electrode library, with electrolyte handling and aliquoting via pumps and the robotic liquid handler.Aliquots are stored in vials within the sample tray, where they undergo nitric acid digestion followed by manual transfer to the ICP-MS autosampler (not shown).b.The flow scheme for preparing, operating, and aliquoting electrolyte (not to-scale).The electrolyte syringe pump is automatically filled from a reservoir and is used to infuse the desired volume into the working electrolyte recirculation cell, both at the beginning of each experiment and at programmed infusion times during electrochemical operation.The syringe insertion point for the robotic solution handler is at the center of the recirculation cell.c.The schematic for the electrode-electrolyte interface (not to-scale).The synthesized AÀ BÀ O electrode is operated in contact with a working electrode, leading to dissolved concentrations of A and B. The alterations to the electrode that are commensurate with the amount of dissolved A and B are calculated as the thickness of corroded electrode as well as the thickness of a BÀ O coating that results from superstoichiometric dissolution of A (compared to the as-synthesized composition).
executed, as envisioned by Materials Acceleration Platforms. [25]urthermore, experiments can be performed with scheduled injection of electrolyte via a syringe pump and scheduled aliquoting of the working electrolyte via a syringe.
The robotic liquid handling system can also process the aliquot, for example the nitric acid digestion used in the present work to dissolve any metal-containing precipitates so that ICP-MS quantifies the total dissolved metals concentrations.While mechanical coupling of the ADSS system to the ICP-MS would be required for autonomous decision making based on measured dissolved metals concentrations, for the present effort of characterizing the equilibration of photoelectrodes with the electrolyte, the tray of liquid samples is manually transferred to the ICP-MS autosampler where each aliquot is measured in triplicate.

Corrosion of Bismuth Vanadate
Bismuth vanadate (BiVO 4 ) is a prominent photoelectrode material due to its efficient conversion of photons (with energy higher than its 2.4 eV band gap) to OER photocurrent. [22,26]The photocorrosion of BiVO 4 in alkaline electrolyte has been well documented, [27] motivating our selection of this photoanode material in 0.3 M NaOH electrolyte (pH 13.6) for initial demonstration of ADSS operation.ICP-MS characterization of the asprepared electrolyte, using the same digestion method as for the aliquots during electrochemical operation, resulted in the following values (mean and standard deviation from 3 ICP-MS injections) for the elements characterized in the present work: 0.006 � 0.001 μM Bi, 0.033 � 0.001 μM V, 0.005 � 0.001 μM Ni, 0.001 � 0.0001 μM Sb.These values are all well below those observed during electrochemical operation.
Figure 2 shows the evaluation of a sputtered BiVO 4 working electrode with a SnO 2 : F back contact.The electrochemical measurements include 1 min open circuit potentiometry (CP) measurement before and after a 37 min chronoamperometry (CA) measurement.The series of 9 aliquots characterize the time evolution of the dissolved Bi and V concentrations.Figure 1c illustrates a physical model of the alterations to the working electrode, which are parameterized as 2 thickness values that we calculate using the equations derived in the Supporting Information.For each aliquot, the model provides the equivalent thickness of the BiVO 4 electrode that has dissolved in electrolyte.Furthermore, since the data indicates a preferential dissolution of V at the beginning of CA, the operational electrode surface is coated with a Bi-rich film.We model this film thickness using the molar density of Bi 4 O 7 (0.039 mol cm À 3 ), which is the stable species indicated by the computational Pourbaix diagram. [9]The data in Figure 2 indicate that in the first few minutes of operation, preferential dissolution of V leads to a 1-2 nm coating of a Bi 4 O 7 -like material (Figure 2).Subsequently, BiVO 4 corrodes nearly stoichiometrically at a rate of ca. 1 nm min À 1 , similar to the previous report for photocorrosion (1.2 to 1.3 nm min À 1 ) for the average corrosion over 20 min at 1.23 V vs RHE in pH 12.3 electrolyte. [27] note that the dissolved and coating thicknesses calculated in the present work use the geometric electrode area and bulk molar densities.For a specific surface area ratio (roughness factor) of X, the thicknesses would be lower by a factor of X, which means the reported thicknesses can be considered as upper limits.While the dissolved metals concentrations do not depend on surface roughness or the uniformity of the corrosion processes, we note that all corrosion and passivation thickness in the present work characterize the equivalent changes to the electrode under the approximation of uniform corrosion across the electrode surface.
The red dashed line in Figure 2 indicates that at 24 min into the CA measurement, 3.5 mL of fresh electrolyte was injected to double the working electrolyte volume.This perturbation is designed to evaluate whether and how the electrode is equilibrating with the electrolyte.The steady corrosion rate of BiVO 4 indicates that the corrosion kinetics are not strongly affected by the dissolved metals concentrations in the 1-3 μM range observed in Figure 2.This observation is consistent with the recent report that BiVO 4 corrosion in borate-buffered electrolyte was suppressed via addition of 0.1 M of V 5 + in electrolyte, [22] i. e. the equilibrium dissolved metals concentration of V is much larger than the μM-level concentrations observed in the ADSS experiment.While electrode-electrolyte equilibration at such high dissolved concentrations can in principle be characterized by the ADSS technique, the intention of the technique is to demonstrate equilibration at low dissolved metals concentration.

Passivation of Nickel Antimonate
We recently reported an amorphous Ni 0.5 Sb 0.5 O y electrode with excellent operational durability in pH 10 electrolyte. [28]Starting with an electrolyte volume of 3 mL, Figure 3 shows the application of the Figure 2 ADSS protocol to evaluate the durability of NiSbO y in pH 13.6 electrolyte.The data indicates dissolution of ca.0.1 nm of film within the first few minutes of operation, after which no further stoichiometric corrosion of the film is observed during the subsequent 20 min of operation.Upon the doubling of the electrolyte, the amount of dissolved NiSbO y also doubles over the next 10 min.These data strongly suggest that the electrode is approximately at equilibrium with electrolyte containing ca. 15 nM of both Ni and Sb.
The super-stoichiometric corrosion of Sb in this first experiment on the NiSbO y electrode indicates a Ni-rich coating, which we model as a NiOOH (0.044 mol cm À 3 ) film based on the established identification of this phase of Ni under alkaline OER conditions. [29]While the coating thickness is less than 1 nm for the initial experiment, the concomitant formation of this  2 with potential and current measured by the potentiostat, dissolved metals concentrations measured in triplicate for each aliquot by ICP-MS (there are 3 data points for each concentration but they are often strongly overlapped), and the resulting thickness under a flat-electrode assumption for the dissolved NiSbO y film as well as a NiOOH-like coating (see legends and labels in Figure 2).The red dashed line corresponds to the electrolyte-doubling injection, which resulted in an approximate doubling of the dissolved film thickness.Combined with the subsequent stabilization of the dissolved metals signals, these data indicate a quasi-equilibrium of the electrolyte with the NiSbO y electrode coated with less than 1 nm of an NiOOH-like material.To evaluate whether this type of coating is required for equilibration with the electrolyte, the light orange arrows in between each column of panels indicate the transfer of working electrolyte from the end of the ADSS experiment on one electrode to the subsequent experiment on a fresh, as-synthesized NiSbO y electrode.This process is repeated twice, resulting in the second and third columns.The sub-0.1 nm dissolution thickness and ca. 1 nm NiOOH-like coating in these latter 2 experiments reveal the importance of the NiOOH-like coating in establishing a quasi-equilibrium with the electrolyte under these photoelectrochemical conditions.passivation layer and the near-equilibrium dissolved metals concentrations in electrolyte precludes us from inferring whether this coating would form even if the electrolyte contained the dissolved metals prior to first contact with the electrode.
To probe the behavior of the electrode with pre-dissolved metal in electrolyte, we performed additional experiments wherein the electrolyte is reused with a fresh, nominally duplicate electrode.For each of the 2 additional experiments shown in Figure 3, an as-synthesized Ni 0.5 Sb 0.5 O y electrode was used, while the electrolyte from the end of the previous experiment was used as the initial electrolyte.In each of these measurements, since the starting electrolyte contains dissolved metals near the equilibrium value, less than 0.1 nm of film is dissolved, while ca. 1 nm of NiOOH-like coating is consistently formed, indicating that this coating acts as a passivation layer to suppress further Sb corrosion.Regarding the persistent corrosion of a small amount of NiSbO y from each as-synthesized electrode, we offer two possible explanations.The surface layer of the as-synthesized electrode may be more susceptible to corrosion than the underlying film.Alternatively, since a thin NiOOH-like passivation layer is required to passivate Sb corrosion, some Sb must dissolve from the surface, which creates undercoordinated Ni atoms, some of which are prone to dissolution before they can re-organize into a passivation layer.Disambiguating these transient and dynamic processes are beyond the scope of the initial ADSS-based assessment, whose focus is on accelerated characterization of electrode-electrolyte equilibration.
In our previous electron microscopy characterization of NiSbO y films prepared by the same method as those used herein, we observed an apparent roughness factor of more than 10, [28] suggesting that the passivation layer thickness for the electrodes in Figure 3 is on the order of 0.1 nm, which approximately corresponds to a monolayer of NiOOH coating the NiSbO y semiconductor.Given the intrinsic catalytic activity of NiOOH, this in situ formation of a thin protective and catalytic coating on a visible-band gap light absorber is an ideal architecture for a photoelectrode.These results strengthen the promise of NiSbO y as a photoelectrode that can achieve longterm operational durability, motivating continued work to improve the carrier transport, absorptivity, and radiative efficiency of this amorphous semiconductor.

Conclusions
We report the accelerated durability screening systems (ADSS) for characterization of corrosion and self-passivation of (photo)electrodes.The automation of the ADSS enables durability-focused down-selection in accelerated electrocatalyst discovery workflows.We demonstrate that within an hour of experimentation we can differentiate between non-passivating (BiVO 4 ) and passivating (NiSbO y ) electrodes while also determining the approximate equilibrium dissolved metals concentrations for passivating electrodes, which informs their suitability for device implementation.Electrolyte recirculation is central to this accelerated screening technique and motivates electrolyte characterization via a handful of scheduled aliquots, as opposed to quasi-real-time monitoring.The additional capability of scheduling electrolyte infusions is critical to validating whether the system is approaching thermodynamic equilibrium, and we envision that this technique can be expanded to observe the response of an electrode to a wide variety of electrolyte or electrode perturbations.Finally, we highlight the value of ADSS in answering a pervasive question in electrocatalysis research: "Do we measure on what we synthesized?" [30]By using the dissolved metals concentrations to determine the effective thickness of the dissolved material and the passivation layer, the ADSS experiment provides a model for the operational surface of the electrode without the need for operando surface characterization.The measurement of equilibrium dissolved metals concentrations at a given electrochemical condition also enables unprecedented experimental mapping of Pourbaix diagrams for multi-element electrodes.

Experimental Electrode Synthesis and Characterization
Thin film electrodes were prepared using ratio-frequency (RF) magnetron co-sputtering of metal targets onto 100 mm-diameter soda lime glass (with SnO 2 : F conducting layer) substrate in a custom-designed combinatorial sputtering system (Kurt J. Lesker, CMS24) described in detail previously. [31]The deposition proceeded in a mixed O 2 (0.9 mTorr) and Ar (5.1 mTorr) gas with 10 À 8 Torr base pressure with 2 in.magnetron sources containing metal targets (ACI Alloys) operated at powers of 11, 142, 100, and 35 W for Bi, V, Ni, and Sb, respectively.The thin film libraries were annealed at 610 °C in air.
Identification of the appropriate samples on the sputter composition library thin films proceeded by x-ray fluorescence (XRF, EDAX Orbis) and x-ray diffraction (XRD, Bruker DISCOVER D8 diffractometer with Cu Kα radiation from a IμS source) characterization.The metal characteristic peak intensities were extracted from the Orbis software and converted to metal contents and normalized compositions using the sensitivity factor for each element calibrated by commercial thin film standards with ca. 10 at.% uncertainty.The oxygen stoichiometry was not measured in the sputtered composition libraries.The XRD acquisition was performed with a 0.3 mm collimator, and patterns were indexed using entries in the Powder Diffraction File from the International Crystallography Diffraction Database (ICDD) [32] via the Bruker EVA software.The identifications for the phases of tested materials are shown in Supporting Information.

Electrolyte and Electrochemistry
The custom electrochemical cell was machined from polyether ether ketone with o-ring seals (Viton, McMaster Carr).The o-ring seal to the working electrode provided ca.0.317 cm 2 of electrodeelectrolyte contact.OCP and CA measurements were performed using a 3-electrode Gamry 1010 potentiostat with a miniature Ag/ AgCl reference electrode (LF2, Innovative Instruments).Electrolyte was prepared by dissolving 0.3 M NaOH Macron ACS pellets in deionized water (Millipore Advantage A10) The pH of the electrolyte was measured to be 13.6.The electrolyte reservoir was coupled to a 50 mL syringe pump (KD Scientific) with a second syringe pump containing deionized water for system cleaning between measurements.Upon syringe pump infusion of electrolyte into the working electrode chamber, the working electrolyte was continually bubbled with 1 atm O 2 gas.Recirculation through Puri-flex tubing with an in-line Masterflex CL peristaltic pump was performed on both the working electrolyte and counter electrolyte.The counter electrolyte was also 0.3 M NaOH with a Pt wire counter electrode and was separated from the working electrode by a bipolar membrane (FumaSEP).The Pt concentration in the as-prepared electrolyte was measured by 3 ICP-MS injections to be 0.0022 � 0.0001 μM, and the corresponding value for the 144 aliquots acquired during electrochemical operation is 0.0018 � 0.0001 μM, demonstrating that the Pt concentration is consistently low and that no crossover of Pt from the counter electrode was observed.During OCP and CA measurements, the electrode was illuminated with 385 nm light from a fiber-coupled light emitting diode (Thor Labs).The illumination profile was dictated by the natural profile from the termination of the optical fiber within the working electrode compartment approximately 0.4 cm from the working electrode surface.Without electrolyte present, the total illumination through the working electrode port was measured to be ca.19 mW (Newport Model 1918-R Power meter with a Model 818-UV sensor).Aliquots of 0.1 mL volume were extracted from the recirculation reservoir via a robotic solution handling system (Thermo Scientific) operated via the PAL Sample Control software.Electrode selection and mounting use a 3-axis translation stage (Dover Motion) with a Galil controller.Computer control of all components proceeded via custom software developed in the HELAO-async [24] framework.

Electrolyte Characterization
Each 0.1 mL electrolyte underwent nitric acid digestion by 0.9 mL of 4 % nitric acid.The resulting 1.0 mL vial for each aliquot was loaded into the autosampling ICP-MS (Thermo Fisher Scientific iCAP™ RQ) where 3 measurements were performed per vial to determine the concentration of dissolved metals.The instrument was calibrated using calibration solutions wherein 0.3 M NaOH was diluted 1 : 10 with 4 % nitric acid and injected with elemental standards comprising 0.5, 1, 5, and 10 ppb for each of the following elements: Ag, Cd, Co, Cr, Cu, Fe, Hg, Mn, Ni, Pb, Tl, V, Zn, Sb, Pt.

Figure 1 .
Figure 1.a.A schematic of the ADSS instrument.The electrochemical cell is sealed to a selected location on the electrode library, with electrolyte handling and aliquoting via pumps and the robotic liquid handler.Aliquots are stored in vials within the sample tray, where they undergo nitric acid digestion followed by manual transfer to the ICP-MS autosampler (not shown).b.The flow scheme for preparing, operating, and aliquoting electrolyte (not to-scale).The electrolyte syringe pump is automatically filled from a reservoir and is used to infuse the desired volume into the working electrolyte recirculation cell, both at the beginning of each experiment and at programmed infusion times during electrochemical operation.The syringe insertion point for the robotic solution handler is at the center of the recirculation cell.c.The schematic for the electrode-electrolyte interface (not to-scale).The synthesized AÀ BÀ O electrode is operated in contact with a working electrode, leading to dissolved concentrations of A and B. The alterations to the electrode that are commensurate with the amount of dissolved A and B are calculated as the thickness of corroded electrode as well as the thickness of a BÀ O coating that results from superstoichiometric dissolution of A (compared to the as-synthesized composition).

Figure 2 .
Figure 2. The results from ADSS characterization of BiVO 4 in pH 13.6 electrolyte at OCP and 1.23 V vs RHE under 385 nm illumination.The electrochemical and aliquot data acquired during OCP at the beginning and end of the experiment are shown in orange, with all CA-associated data in blue.The potential and current are measured by the potentiostat, the Bi and V concentrations are measured by ICP-MS, where for each element and each time point, there are 3 markers corresponding to triplicate measurement of each aliquot.The markers are often overlapping due to consistency in the measurements.The time evolution of the dissolved metals concentrations along with the variation in the triplicate measurements are used to model the value and uncertainty in the dissolved BiVO 4 thickness as well as a Bi 4 O 7like coating, where each thickness value is an upper limit due to the assumption of a flat electrode.As illustrated in the legend, the thickness value corresponding to each aliquot is shown as a vertical bar representing the mean plus and minus the standard deviation of the distribution of values derived from the ICP-MS data.The height of the bar is often smaller than the line width of the light blue trend line.The dashed red line traversing all panels indicates the time of infusion of fresh electrolyte to double the working electrolyte volume, which instantaneously lowers the dissolved metals concentrations and proportionally lowers the increase in concentrations with time for a given corrosion rate.

Figure 3 .
Figure 3.The results from ADSS characterization of 3 NiSbO y electrodes in pH 13.6 electrolyte at OCP and 1.23 V vs RHE under 385 nm illumination.The first column corresponds to the analogous ADSS experiment of Figure2with potential and current measured by the potentiostat, dissolved metals concentrations measured in triplicate for each aliquot by ICP-MS (there are 3 data points for each concentration but they are often strongly overlapped), and the resulting thickness under a flat-electrode assumption for the dissolved NiSbO y film as well as a NiOOH-like coating (see legends and labels in Figure2).The red dashed line corresponds to the electrolyte-doubling injection, which resulted in an approximate doubling of the dissolved film thickness.Combined with the subsequent stabilization of the dissolved metals signals, these data indicate a quasi-equilibrium of the electrolyte with the NiSbO y electrode coated with less than 1 nm of an NiOOH-like material.To evaluate whether this type of coating is required for equilibration with the electrolyte, the light orange arrows in between each column of panels indicate the transfer of working electrolyte from the end of the ADSS experiment on one electrode to the subsequent experiment on a fresh, as-synthesized NiSbO y electrode.This process is repeated twice, resulting in the second and third columns.The sub-0.1 nm dissolution thickness and ca. 1 nm NiOOH-like coating in these latter 2 experiments reveal the importance of the NiOOH-like coating in establishing a quasi-equilibrium with the electrolyte under these photoelectrochemical conditions.