Operando electrochemical spectroscopic ellipsometry: Insights into electrochemical behavior of catalyst materials under realistic working conditions

Large-scale hydrogen production using proton exchange membrane water electrolysis (PEM-EL) requires a drastic reduction in the current noble metal content of the electrodes. To achieve this, new analytical techniques for characterizing solid-liquid catalytic interfaces, applicable in operando under a variety of realistic operating conditions, are critical because they provide a deeper understanding of the relationships between catalyst state, structure, and catalytic performance. However, many analytical techniques cannot be applied in liquid environments at realistic potentials and current densities. We propose electrochemical spectroscopic ellipsometric analysis (ECSE), performed in realistic electrolysis cells, as a new analytical operando method for characterizing electrochemical reactions at solid-liquid interfaces under widely varying working conditions. The method is first validated on a platinum surface using ex-situ and operando ellipsometric analyses. Subsequently, the physico-chemistry of a mesoporous IrO x catalyst film under oxygen evolution reaction (OER) potentials was investigated, showing precise reversible and irreversible potential-dependent variations of a number of physical material properties relevant to catalysis.


Main article Introduction
Due to the demand for energy with a low carbon dioxide footprint, efforts to develop sustainable and fossil-free energy sources are of great importance. 1 Hydrogen can act as a renewable energy carrier and as a resource for chemical processes. 2,3 Green hydrogen is produced by electrolysis using renewable energy. Catalysts play an important role by increasing the rate, efficiency, and selectivity of chemical reactions. In water electrolysis, limiting factors are the not well understood reaction mechanism in the oxygen evolution reaction (OER) and the use of expensive materials, e.g. iridium for acidic water splitting. [4][5][6] Significant progress has been made recently in the understanding of electrocatalysts, particularly in water splitting, i.e., both the OER and the hydrogen evolution reaction (HER). The combination of theoretical and experimental studies has led to a new understanding of catalytic trends enabling the development of more efficient electrocatalysts. 4 In situ and/or operando experiments allow the catalytic activity to be analyzed under specific/realistic working conditions during the reaction. Often, these methods are based on X-rays, performed at synchrotron facilities in ultra-high vacuum with specially developed measurement setups and cells. 7 A strong correlation of iridium hydroxo (Ir-OH) surface species with the OER activity was demonstrated using operando X-ray absorption spectroscopy (XAS) and X-ray photoelectron spectroscopy (XPS) on iridium oxide powder materials, observing a mixture of Ir III /Ir IV oxidation states transitioning to higher oxidation states above 1.00 V vs. RHE. 8 A decrease in intrinsic catalyst activity, was attributed to the presence of different IrO2 facets. The authors also suggested that gas bubbles are trapped in the mesopores of the sample blocking parts of the active sites. Direct analytical evidence of these assumptions was not given.
The development of vacuum-free experiments promises easier access to key parameters related to electrochemical activity of electrocatalysts. Spectroscopic ellipsometry (SE) enables a vacuum-free and non-destructive analysis, as it can be used under different environmental conditions. A classical application of ellipsometry is the study of layers on top of semiconductor materials. 9 Further development of spectroscopic ellipsometric measurements allows the in-situ study of interfaces during layer growth, adsorption/desorption of solvents, or biomolecular processes. [10][11][12][13] Ellipsometry is based on determining the change in the polarization state of an electromagnetic wave upon reflection, often quantified in the ellipsometric quantities Ψ and Δ, which are related to an amplitude ratio and a phase shift, respectively.
In-situ ellipsometry is often used to study film growth in coating technology. 10 It can be used to control film thickness during deposition and to study film growth rate, e.g., in atomic layer deposition (ALD). SE can also be used for electrochemical processes. First ex-situ investigations and modeling of mesoporous iridium oxide films used in acidic OER showed accurate results of material specific parameters such as film thickness, porosity and resistivity compared to other methods. Electronic properties could be derived and correlated with activities in the OER. 14 Here, we report on the analysis of electrocatalytic surfaces by operando electrochemical spectroscopic ellipsometry (ECSE). The setup consists of a spectroscopic ellipsometer and an electrochemical flow cell providing a realistic sample environment suitable for spectroscopy (Scheme 1). The setup was validated by a quantitative time-resolved analysis of reaction-induced surface oxidation and reduction of a bulk polycrystalline platinum surface. With this setup, we studied highly active IrOx-based porous model catalysts to track the catalyst surface and the evolved gas during dynamic OER conditions. The ellipsometric model provides time-and potential-dependent information of properties such as resistivity, interband transition energies, and the degree of pore filling with gas. Scheme 1: Illustration of the developed operando electrochemical spectroscopic ellipsometry (ECSE) approach for the analysis of material properties under realistic working conditions. The approach includes the validation of a quantitative time-resolved analysis of reaction-induced oxidation and reduction of a Pt surface and the application for a time-and potential-dependent analysis of material properties of a mesoporous IrOx film during dynamic OER measurements.

Electrochemical, ex-situ SE and ECSE analysis of platinum
Cyclic voltammetry on a platinum surface is used to validate the ECSE method. Cyclic voltammograms were recorded in a standard three-electrode setup as well as in the flow cell setup, (see the supporting information SI1). The electrolyte was 0.1M HClO4 and a scan rate of 100 mV s -1 was applied. The resulting Pt CV in the standard three-electrode configuration illustrates the in literature reported processes of the Pt-H formation and oxidation, the double-layer region as well as the Pt-O formation and reduction contributing to the characteristic shape. 15 The measurement in the flow cell setup exhibits the same characteristic processes and shape, with slightly lower signal intensity (current density response).
For the study of materials in ECSE, it is useful to study the materials in advance in an ex-situ analysis and develop a parameterization of the dielectric function. As a result, the resistivity at the surface (Pt/PtOx/air interface) with a value of 1.20 • 10 -5 Ohm cm and a thickness of about 0.33 ± 0.26 nm for the PtOx layer were obtained and are shown in the supporting information SI2.
The operando investigations of Pt using the ECSE flow cell setup are described in supporting information SI3 in detail. The ellipsometric measurement was corrected for the window and electrolyte effect using a Si wafer with a known oxide layer thickness (see supporting information SI4). ECSE was measured between -0.28 V and 1.35 V against an Ag/AgCl reference electrode (+0.256 V vs. RHE at pH 1), a Pt wire as counter electrode and a cycle speed of 2 mV s -1 . The model derived from the ex-situ Pt analysis (see supporting information SI2) was used together with the dielectric function of the electrolyte which served as the ambient above the Pt surface. Figure 1 presents the results of the electrochemical and spectroscopic ellipsometric measurements of the polycrystalline platinum surface. The shape of the cyclic Pt voltammogram at a reduced scan rate of 2 mV s -1 (solid line) differs from the CV at a scan rate of 100 mV s -1 (dashed line), which is due to the dependence of the current response on the scan rate ( Figure 1a). 16 However, the characteristic HER and OER regimes as well as a change of the current density in the range of 0.6 VRHE are clearly visible, which probably indicates the Pt-O formation and/or reduction (oxygen chemisorption).
The corresponding Ψ and Δ spectra (Figure 1b   This hysteresic behavior agree well with the previously described oxidation/reduction mechanism of Pt during cyclovoltammetry. Oxide layers on platinum of up to 1.0 nm were previously determined ellipsometrically. 17 Investigations of the formation and growth of oxide films on platinum electrodes showed that the growth rate can be up to 1.2 nm min -1 at an oxidative potential of 1.20 VAg/AgCl (~ 1.46VRHE). 18

ECSE investigation of a mesoporous IrOx film calcined at 375 °C
For modeling of mesoporous iridium oxide films during electrochemical investigations, a modified model as described in ref. 14 was used. The model consists of a Ti substrate layer (multi-peak model), a TiOx layer (single Tauc-Lorentz oscillator), and a mesoporous IrOx layer. The latter was modeled by an a-BEMA parameterization with an additional isotropic BEMA for the volume fractions of gas (air/void) and electrolyte within the pores instead of the air/void as described in ref.
14. In addition, the electrolyte was used as the ambient layer above the mesoporous IrOx layer and its parameters obtained from the calibration procedure (see supporting information SI4).
The polished Ti substrate was analyzed before dip-coating and after film deposition and calcination. The thickness of the TiOx layer was determined before (2.6 nm ± 0.5 nm) and after the calcination procedure    The dielectric function obtained by fitting can be converted into a valence electron energy loss spectrum (see supporting information SI7) as described in ref. 14. This allows a more detailed analysis of the electronic structure of the IrOx material (matrix) under electrochemical working conditions. Figure 2d shows the surface (−Im((1 + ) −1 )) VEELS spectra which have two features, here referred to as A and B peaks. [21][22][23] These features can be assigned to transition energies from the O 2p orbital to the partial occupied (A peak) and unoccupied (B peak) Ir 5d t2g orbital. A slight shift of the center energy to higher energies is observed especially for the A peaks as well as a change in intensity for both A and B peaks. Based on crystal field theory, the Ir 5d orbital splits into t2g and eg sub-levels. 24  Additionally, electrical parameters such as resistivity, free carrier concentration as well as the volume fraction of the produced gas (filling factor) were obtained. Figure 3 displays a) the filling factor, b) the resistivity, and c) the free carrier concentration as a function of the potential. The filling factor indicates a constant value close to zero for the forward-scanning cycle (black circles) until a potential of about 1.45 VRHE and a current density of 2 mA cm -2 are reached. As the potential continues to increase, the filling factor increases rapidly to a value of about 46%. In the back-scanning cycle (white circles), the filling factor decreases more slowly than in the forwards cycle and reaches a value close to zero at a potential of about 1.30 VRHE. The model analysis thus indicates that the gas adheres to the pores surface and is slowly replaced by the electrolyte.
The resistivity values are completely reversible and decrease with increasing potential. This trend matches with the a constant to slightly increasing conductivity at low-potentials and a stronger increase at a potential of about 1.20 VRHE observed earlier by the group of Haverkamp. 25 The assumption that the changes in the ellipsometric spectra are caused by changes in resistivity has to be validated. Therefore, operando ECSE measurements were performed on a non-conductive mesoporous titanium oxide (TiOx) film in the OER regime and in the potential range between 0.40 VRHE and 1.40 VRHE. In both potential regions, no changes in the Ψ and Δ spectra with changes in potential were observed (see

Discussion of the potential dependent changes of the IrOx material properties during OER
The changes in the electrical and electronic properties of the IrOx film during the OER as described above can probably be explained by changes in the band structure. In their work, Mortimer and coworkers describe electron transfer processes that can occur at the electrode-electrolyte interface. 26 In this process, the probability of electron transfer increases when the Fermi level increases to a level that facilitates charge transfer to an acceptor or oxidized state, or decreases to a level that facilitates charge transfer from a donor or reduced state. 26 When an anodic potential is applied, the Fermi level decreases and reaches an energy level where electrons can be transferred from the donors in the electrolyte to the metal, increasing the density of states of the acceptors. 26 In the case of the IrOx film, the anodic potential could result in the Ir 5d t2g level being lowered. It has been reported that a shift of the Ir 5d t2g level to lower energies causes an increase in the p-d intermediate band transition energies, which can be explained by a higher electron density as well as a higher occupation of the d-band. 27 Finally, the adhesion of the produced gas should be discussed. The fraction of gas-filled versus electrolyte-filled pore volume (filling factor) shows a hysteresis loop (Figure 3a). This indicates that oxygen within the catalyst layer is only removed slowly. Stoerzinger et al. suggest this behavior in their studies on the contribution of lattice oxygen species from RuO2 films and nanoparticles to the oxygen gas using online electrochemical mass spectrometry (OLEMS). 31 Therefore, it may be possible that diffusion limits gas dissolution and transport. However, since there is a complex interplay between the surface and pore properties, no definitive conclusion can be drawn with the present data and further detailed analysis has to be performed.

Conclusion
With the ECSE approach, small changes on a platinum surface are detectable in cyclic voltammetry studies, such as oxidation and reduction of the time-and potential-dependent PtOx layer growth and the associated changes in electrical properties. This fundamental study of operando ECSE measurement on a platinum surface demonstrates the accuracy of potential-dependent surface analysis and opens a wide field for electrochemical and electrocatalytic studies under operando conditions.
Operando ECSE analysis in the OER regime of a mesoporous IrOx film, calcined at 375 °C, proves that the dielectric function changes due to the shift in the Ir oxidation states and the change of the p-d interaction. A shift of the A peak in the VEEL spectra is visible, which was previously associated with a linear correlation of the intrinsic OER activity. The changes in the Ir 5d t2g band and the p-d interband transitions are also directly reflected in the electrical properties of the material. According to the current state of knowledge, the model analysis provides for the first time an analysis of the volume fraction of the produced gas and gives an indication of an accumulation of gas within the pores.
The analytical ECSE approach also provides the opportunity to study different materials under a variety of electrochemical processes to gain a deeper understanding of catalytic systems, batteries, and material corrosion. The ability to perform fast, vacuum-free, and non-destructive characterization under realistic potentials and current densities could also be used to study other electrochemical reactions such as the HER, HOR, or ORR in the future.

Chemicals
For the synthesis of the mesoporous iridium oxide film iridium acetate (Ir(CH3COO)n, 99.95% metals basis, ca. 48% Ir) was used from chemPUR and a triblock-copolymer (PEO-PB-PEO, containing 20400 g mol -1 polyethylene oxide (PEO) and 10000 g mol -1 polybutadiene (PB)) was purchased from Polymer Service Merseburg GmbH 32 . Ethanol (EtOH, >99%) was used as solvent from Sigma-Aldrich. All chemicals were used as received. A conductive titanium sheet (25 x 40 mm) was used as substrate, which was polished using a suspension consisting of non-crystallizing colloidal silica suspension (200 ml; 0.25 µm; Struers, OP-S), hydrogen peroxide (40 ml) and nitric acid (1 ml), followed by a cleaning procedure with EtOH. The platinum foil for operando ECSE studies was polished using a 3 µm diamond suspension (Struers, DiaDuo-2) and cleaned with ethanol.

Synthesis of mesoporous metal oxide films
The mesoporous iridium oxide film was synthesized according to the synthesis which is described by Ortel et al. 33 In a slightly modified synthesis, 169 mg of the PEO-PB-PEO polymer template were dissolved in 7.5 ml ethanol at 40 °C. After complete dissolution, 844 mg iridium(III) acetate were added and the solution was stirred for 1h at 40 °C. The dip-coating solution was transferred into a preheated Teflon cuvette (2h at 50 °C) and dip-coating was immediately performed under controlled dip-coater conditions (25 °C, 40% relative humidity) with a cuvette heater (45 °C) and a withdrawal rate of 200 mm min -1 . As-synthesized films were dried for 5 minutes under the same conditions and subsequently calcined in a preheated muffle furnace in flowing air for 5 minutes at temperature of 375 °C.
The mesoporous titanium oxide film was synthesized according to the synthesis which is described by Ortel et al. 32 In a slight modified synthesis, 181 mg of the PEO-PB-PEO polymer template were dissolved in 5 ml ethanol at 40 °C. In a second solution 1 g TiCl4 was mixed with 5 ml ethanol under an Ar atmosphere. After complete dissolution both solutions were mixed for 1h at 40 °C. Dip-coating was subsequently performed under the same conditions as described above.

Physicochemical characterization
SE measurements were performed with a variable angle spectroscopic ellipsometer M2000 DI (J.A. Woollam) in a spectral range between 0.7 eV and 6.5 eV (192 nm and 1697 nm). Samples were measured ex-situ at angles of incidence of 65°, 70° and 75° and acquisition times between 5 s and 10 s.
Ψ and Δ spectra were analyzed with the software CompleteEASE (v. 6.42) using the models described above by using the model functions (see supporting information SI2).
Operando electrochemical spectroscopic ellipsometric analysis of all samples was performed with a flow cell setup in the ellipsometer described before (see supporting information SI3).
X-ray reflectometry (XRR) were performed at a Bragg-Soller X-ray diffractometer system with a flat secondary monochromator and fixed Cu Kα-tube (Seifert XRD 3000TT).

Electrochemical testing
Electrocatalytic measurements were performed using a Gamry interface 1000 (IFC1000) with the Gamry Framework Data Acquisition software (v. 7.07). The platinum sheet or coated titanium substrates served as working electrode, a platinum wire as counter electrode and an Ag/AgCl (3M NaCl; flow cell type) as reference electrode. All potentials are recorded under room temperature and referred to the RHE. N2purged perchloric acid was used as electrolyte solution (0.1M HClO4).
Platinum CVs were recorded in a standard three-electrode setup and in the flow cell setup in a potential range between -0.28 VAg/AgCl and 1.35 VAg/AgCl (+0.256 V vs. RHE at pH 1) with a scan rate of 100 mV s -1 and 2 mV s -1 under operando conditions. Operando ECSE investigations of the metal oxide films were performed either in the OER regime (1.20 VRHE -1.48 VRHE) or in a lower potential range (0.40 VRHE and 1.40 VRHE) with scan rates of 1 mV s -1 .