Effect of the Morphology of the High-Surface-Area Support on the Performance of the Oxygen-Evolution Reaction for Iridium Nanoparticles.

The development of affordable, low-iridium-loading, scalable, active, and stable catalysts for the oxygen-evolution reaction (OER) is a requirement for the commercialization of proton-exchange membrane water electrolyzers (PEMWEs). However, the synthesis of high-performance OER catalysts with minimal use of the rare and expensive element Ir is very challenging and requires the identification of electrically conductive and stable high-surface-area support materials. We developed a synthesis procedure for the production of large quantities of a nanocomposite powder containing titanium oxynitride (TiON x ) and Ir. The catalysts were synthesized with an anodic oxidation process followed by detachment, milling, thermal treatment, and the deposition of Ir nanoparticles. The anodization time was varied to grow three different types of nanotubular structures exhibiting different lengths and wall thicknesses and thus a variety of properties. A comparison of milled samples with different degrees of nanotubular clustering and morphology retention, but with identical chemical compositions and Ir nanoparticle size distributions and dispersions, revealed that the nanotubular support morphology is the determining factor governing the catalyst's OER activity and stability. Our study is supported by various state-of-the-art materials' characterization techniques, like X-ray photoelectron spectroscopy, scanning and transmission electron microscopies, X-ray powder diffraction and absorption spectroscopy, and electrochemical cyclic voltammetry. Anodic oxidation proved to be a very suitable way to produce high-surface-area powder-type catalysts as the produced material greatly outperformed the IrO2 benchmarks as well as the Ir-supported samples on morphologically different TiON x from previous studies. The highest activity was achieved for the sample prepared with 3 h of anodization, which had the most appropriate morphology for the effective removal of oxygen bubbles.

Introduction

The hydrogen cycle is a promising way to store energy from renewable sources, like solar and wind, through the splitting of water in an electrolyzer. This energy can subsequently be made available on-demand using a fuel cell that combines the hydrogen and the oxygen.[1] However, the efficiency of the water-splitting reaction is limited by the slow kinetics of the oxygen-evolution reaction (OER).[2] Indeed, even with a state-of-the-art catalyst in acidic media, i.e., IrO2, an overpotential of around 0.3 V is usually required to achieve proper current densities.[3] Moreover, iridium is a precious and scarce metal, which hinders the commercialization of this technology. One way to decrease the cost of the device is to improve the utilization of the iridium. An example of this is the deposition of iridium in the form of nanoparticles on a suitable support.[4] This makes it possible to lower the cost as well as to potentially improve the catalyst’s performance, which, together with the properties of the catalytically active sites, strongly depends on the interactions with the support. The so-called strong metal–support interaction (SMSI), mostly known from heterogeneous catalysis, was shown to also affect the structural and chemical changes of the electrocatalyst during the OER.[5,6] The most commonly used support in electrocatalysis is carbon;[7−9] however, several new materials have been proposed[10−12] to address its shortcoming, i.e., poor stability under OER electrochemical conditions.[13] One of the challenges with OER electrocatalysts is to find an appropriate substitute for the traditional carbon support. It must be replaced by a large-surface-area support with a high-enough electrical conductivity, so it does not negatively affect the catalyst’s performance. Additionally, the stability and electronic catalyst–support interactions, as well as the economic criteria, need to be taken into account.[11,14−16] Given the stated requirements, a promising substitute for carbon is titanium oxynitride (TiON), which can be prepared from titanium dioxide (TiO2)[17] or titanium nitride (TiN)[18] in the form of various nanostructures.[19−21] There are different synthesis methods to choose from. Radio frequency (RF) plasma can be used for the oxynitridation of titanium;[22] the mechanical milling of titanium powder in air results in the formation of titanium oxynitride;[23] and low-pressure chemical vapor deposition (LPCVD) can be used to deposit titanium oxynitride on silicon substrates.[24] Of the possible synthesis processes, one of the most direct and controllable is to synthesize TiO2 nanostructures, which are then thermally treated in a reductive ammonia atmosphere (above 600 °C) to transform the TiO2 to TiON.[25] The anodic oxidation of titanium is commonly used for the synthesis of immobilized nanotubular films.[26] It results in TiO2 nanotube arrays that are firmly adhered to the titanium substrate[27] and has been used to fabricate free-standing TiO2 membranes[28] and TiO2 nanotube powders.[29] Before the prepared TiO2 nanotube powders can be used as a support for Ir nanoparticles, their electrical conductivity has to be substantially increased. This can be achieved with a thermal treatment in an ammonia atmosphere. The final step of the OER catalyst’s preparation, i.e., the deposition of Ir in the form of finely dispersed, ultrasmall nanoparticles, was already demonstrated by our group.[30−32] In this investigation, we developed a new, anodic, oxidation-based synthesis process for the cost-effective fabrication of high-performance OER TiON-Ir nanopowder electrocatalysts. The influence of the anodization time on the catalyst’s support morphology, structure, and composition was studied in detail using various state-of-the-art characterization methods, such as X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), scanning transmission electron microscopy (STEM), X-ray diffraction (XRD), and extended X-ray absorption fine structure (EXAFS). We show that the electrochemical activity and stability can be substantially enhanced, compared to the commercial benchmark IrO2 material and to the other best OER catalysts in the literature, exclusively by modifying the TiON support morphology.

Experimental Section

Synthesis of TiON-Ir Catalysts

The TiON-Ir catalyst was prepared by following the procedure schematically shown in Figure . In the first step, titanium foil (200 μm thick, 99.8%, Baoji Lyne Metals Co., Ltd.) was cleaned with acetone and ethanol in an ultrasonic bath and then anodized in an electrolyte consisting of 0.3 wt % NH4F (99.99%, Sigma-Aldrich) and 2 vol % deionized water in ethylene glycol (99.5%, Carlo Erba Reagents). The foil was anodized at a constant potential of 60 V in a two-electrode electrochemical cell using a stainless steel counter electrode. The anodization times were 1, 3, and 6 h. The amorphous TiO2 nanotube film grown with anodic oxidation was washed with deionized water and ethanol. In the next step, TiO2 nanotubes were simply detached from the metal titanium substrate by bending the anodized foil and thermally treating in ammonia at 700 °C for 10 h to convert the amorphous TiO2 nanotubes into TiON. The flow of ammonia gas was kept constant at 50 cm3 min–1. After that, iridium nanoparticles on TiON were prepared by dissolving 70 mg of iridium(III) bromide hydrate (Sigma-Aldrich, St. Louis, MO) in 1 mL of water at 50 °C. The solution was then mixed with 100 mg of TiON powder (lightly milled in a mortar) at 50 °C until evaporation. Afterward, the mixture was thermally treated in a 5% H2/Ar mixture. The temperature was increased at a rate of 2 °C min–1 to 450 °C for 1 h and then cooled to room temperature with a rate of 3 °C min–1. After the thermal treatment, the final composite material contained 11 wt % iridium nanoparticles, as determined with inductively coupled plasma-optical emission spectroscopy (ICP-OES). The samples were denoted TiON-nh-Ir, where n can be 1, 3, or 6 and represents the anodization time.

Figure 1

Procedure for TiON-nh-Ir catalyst preparation. (a) Anodic oxidation of titanium foil, (b) detachment of TiO2 nanotubes, (c) annealing of detached TiO2 nanotubes in ammonia, (d) deposition of Ir onto TiON nanotubes, and (e) the final TiON-nh-Ir powder catalyst.

Material Characterization

Scanning electron microscopy (SEM) was carried out using a Zeiss Supra 35 VP (Carl Zeiss, Oberkochen, Germany) microscope equipped with an energy-dispersive X-ray (EDX) spectrometer (Oxford Instruments, model Inca 400). The operating voltage was set to 7 kV. Detailed STEM analyses of the structural and elemental features of the samples were performed in a Cs-corrected transmission electron microscope (TEM, CF-ARM JEOL 200) equipped with an SSD JEOL EDX spectrometer and a GATAN Quantum ER dual-electron energy loss spectroscopy (EELS) spectrometer. An operating voltage of 80 kV was employed. XRD spectra of the samples were recorded on a flat, disclike, Si sample holder using an X-ray powder diffractometer PANalytical X’Pert PRO MPD (PANalytical B.V., Almelo, The Netherlands) with Cu Kα1 radiation having a wavelength of 1.5406 Å in the α1 configuration with a Johansson monochromator on the primary side. The diffractograms were recorded with 0.034° resolution and 100 s signal integration time in the 2θ range from 20 to 100°. Phases were identified with the X’Pert HighScore Plus program using the International Centre for Diffraction Data (ICDD) PDF-4+ 2019 database.[33]

An inductively coupled plasma-optical emission spectrometry (ICP-OES) instrument (Varian 715-ES) was used to determine the iridium content in the samples.

The standards were prepared in-house by diluting traceable, certified, ICP-grade, single-element standards (Merck CertiPUR). The dilution of the samples and standards involved ultrapure water (Milli-Q, Millipore), HNO3, and HCl (Merck, Suprapur). The samples were weighed and subsequently digested using a microwave-assisted digestion system (CEM MDS-2000) in 3/1 v/v HCl/HNO3. The digested samples were then cooled to room temperature and diluted with 2% v/v HNO3 until their concentration was in the desired range for ICP-OES measurements.

XPS was used to characterize the surface of the TiON-Ir catalysts (the upper 3–5 nm) using a PHI-TFA XPS spectrometer produced by Physical Electronics Inc. and equipped with an Al-monochromatic source. The analysis area was 0.4 mm in diameter, and two measurements were performed on every sample. The surface composition was calculated without taking into account any carbon, supposing that it originates from surface contamination.

The Ti K-edge absorption spectra of the TiON-3h-Ir and TiON-6h-Ir samples and the reference Ti compounds (TiO2 anatase and rutile, and TiN) were measured at room temperature in the transmission detection mode using the X-ray absorption fine structure (XAFS) beamline of the ELETTRA synchrotron radiation facility in Trieste, Italy. The samples of powder mixed with carbon black were pressed into homogenous pellets having a total absorption thickness (μd) of about 2 above the Ti K-edge. A Si(111) double-crystal monochromator was used with an energy resolution of 0.5 eV at 5 keV. Higher-order harmonics were effectively eliminated by detuning the monochromator crystals to 60% of the rocking-curve maximum. The intensity of the monochromatic X-ray beam was measured using three consecutive 30-cm-long ionization detectors, filled with appropriate gas mixtures: 200 mbar N2 and 1800 mbar He in the first, 1100 mbar N2 and 900 mbar He in the second, and 140 mbar Ar, 1000 mbar N2, and 600 mbar He in the third detector. The absorption spectra were measured in the interval from −250 to +1000 eV relative to the Ti K-edge. In the X-ray absorption near-edge structure (XANES) region, equidistant energy steps of 0.25 eV were used with an integration time of 1 s per step. Three repetitions of the scans were superimposed to improve the signal-to-noise ratio. The exact energy calibration was established with a simultaneous absorption measurement on a 5-μm-thick Ti metal foil placed between the second and third ionization chambers. The Ti K-edge in the Ti metal is at 4966 eV. The absolute energy reproducibility of the measured spectra was 0.03 eV. Data analyses of the X-ray absorption spectroscopy (XAS) spectra were performed with the Demeter (IFEFFIT) program package.[34]

The Brunauer–Emmett–Teller (BET) surface-area and nitrogen-sorption measurements for the samples were recorded using a Tristar 3000 automated gas-adsorption analyzer (Micromeritics Instrument Corp.) recording at −196 °C. Before the adsorption analysis, the samples were outgassed under vacuum for 12 h at 120 °C.

The Raman spectra were obtained on a confocal WITec Alpha 300 Raman spectrometer. The powdered samples were positioned on a silicon-wafer substrate under a green excitation-laser wavelength of 532 nm. The spectra were measured at four different positions using a 20× objective, an integration time of 1 s, and 100 scans. Four individual Raman spectra were recoded for each sample. The most representative spectrum for presentation was chosen by, first, omitting the most- and the least-intense spectra. From the remaining two, the spectrum approaching more closely the average values of all of the spectra was plotted. Due to the largest spreading of the intensity values, both middle spectra are shown as representative examples in the case of the TiON-6h-Ir sample.

Electrochemical Measurement

For the electrochemical measurements, each sample was dispersed in Milli-Q water (18.2 MΩ cm) to obtain a suspension with a final concentration of 1 mgcat mL–1. After sonication of the suspension, which ensured good homogeneity, a 20 μL droplet was cast-dropped on a 0.196 cm2 glassy carbon working electrode embedded in a Teflon tip. We used a two-compartment, three-electrode setup. Pt wire and Ag/AgCl were used as the counter and reference electrodes, respectively. During the stability test, the Pt wire was replaced by a graphite rod to avoid any dissolution/redeposition of the Pt on the catalyst. The reference electrode was separated from the rest of the cell to avoid Cl– contamination. An Ar-saturated HClO4 0.1 M (Merck, Suprapur, 70%) was used as an electrolyte. The potential was cycled between 0.05 and 1.6 V vs a reversible hydrogen electrode (RHE) at 1600 rpm and 20 mV s–1 until a stable cyclic voltammogram (CV) was obtained (less than 10 cycles). The last stable CV was used to estimate the activity. The potential was controlled with a potentiostat (potentiostat ECI-200, Nordic Electrochemistry). Prior to the experiment, the resistance of the solution was recorded by electrochemical impedance spectroscopy (EIS) and the iR was compensated accordingly during the activity measurement. The potentials are referred to the reversible hydrogen electrode (RHE), which was measured by saturating the electrolyte with hydrogen and measuring the open-circuit potential with a Pt black commercial catalyst on the working electrode. How to rigorously test the stability of OER catalysts is still the subject of discussion, as different protocols are employed.[15,35−38] Here, a chronopotentiometry test was performed as a preliminary stability test to simulate the operating potentials. A current of 1 mA mgIr–1 was applied to the catalyst for 1 and 2 h as suggested by Spöri et al.[39] Then, the electrolyte was replaced with a fresh solution to re-estimate the activity via a stable cyclic voltammogram (CV) at 1600 rpm and 20 mV s–1.

Results and Discussion

Characterization of TiON-nh-Ir Electrocatalysts

SEM, STEM, BET, XRD, XPS, and XAS at Ti K-edge and ICP-OES analyses were performed to fully characterize the synthesized electrocatalysts. As the morphology of the TiON support has an important influence on the gas-evolving catalytic performance of the catalyst,[40] the influence of the anodization time during the first step of the synthesis process as well as the milling procedure in the last step of the synthesis was studied. A detailed description of the morphological changes during the starting phases of the synthesis process can be found in the Supporting Information (Section S1). As demonstrated by the thorough analysis of our samples, morphology is the only parameter we varied in our study.

The morphology of the prepared TiON-nh-Ir electrocatalysts is shown in Figure . The reason for the crushed nanotube arrays in all of the images is the wet impregnation technique, which included slight milling to obtain a homogeneous distribution of the iridium. Figure a–c shows as-prepared electrocatalysts, while Figure d–f shows the same samples after an additional, more rigorous, milling procedure. The final electrocatalysts presented in Figure d–f are schematically shown in Figure g. The milling procedure had the largest influence on the morphology of the TiON-Ir powders that were anodized for 3 and 6 h. The reason for this is the higher mechanical stability of the nanotube arrays prepared during 1 h of anodization, which is a consequence of the shorter length and thicker wall of these nanotubes. With a longer anodization time, the wall of the nanotubes at the top of the immobilized film becomes thinner (Figure S1), resulting in greater brittleness when applying physical force during milling. It can be observed that the sample anodized for 1 h (Figure a) consists of relatively large nanotube arrays, which become smaller after milling (Figure d). The morphology of the sample anodized for 3 h (Figure b) shows clusters with a nanotubular morphology, which can still be found after reducing the cluster size with milling (Figure e). In the sample anodized for 6 h, no nanotubular morphology features can be recognized anymore, even before the rigorous milling (Figure c). This is due to the formation of less mechanically stable nanotubes during the excessive etching in the longest anodization process. However, the cluster size of the 6 h sample is the largest before and after the rigorous milling (Figure f). The average cluster size of the final electrocatalyst anodized for 3 h shown in Figure e is approximately 50 ± 3 μm, compared to approximately 121 ± 7 μm in the case of TiON-6h-Ir shown in Figure f. The main difference between the electrocatalysts prepared with anodic oxidation and other synthesis procedures is the defined orientation resembling the nanotubes. It should be noted that the time for the anodic oxidation should not be too long, as, in this case, the top surface starts losing the morphology of the nanotubes during anodization. If the anodization process is extended for even longer times, the nanotubes would start turning into nanoparticles with a morphology similar to that of an electrocatalyst prepared from commercial nanopowders (e.g., TiO2 P25). The described loss of the nanotubular structure and the formation of nanoparticles occurred to some extent in the case of the TiON-6h sample and are shown in Figure S2c,d.

Figure 2

SEM micrographs of the synthesized catalysts: TiON-1h-Ir (a), TiON-3h-Ir (b), TiON-6h-Ir (c) and milled TiON-1h-Ir (d), TiON-3h-Ir (e), and TiON-6h-Ir (f). Inset figures (b) and (e) show the nanotubular morphology of the TiON-3h-Ir sample. A schematic presentation of all three milled catalysts is shown in (g).

Additional morphological characterization of the samples was performed with STEM, and the average particle size of the iridium was determined. The average size of the iridium nanoparticles was, for all of the samples, between 3.1 and 3.4 nm, as seen in Table and Figure S4. The iridium nanoparticles are evenly dispersed on the TiON support, as seen for TiON-3h-Ir in Figure and TiON-1h-Ir and TiON-6h-Ir in Figure S5.

Table 1

Average Particle Size as Determined by the STEM Image Analyses and the Amount of Nitrogen and Oxygen in the TiON Support for the TiON-nh-Ir Samples, as Determined with EELS and XPS

	TiONx-1h-Ir	TiONx-3h-Ir	TiONx-6h-Ir
average particle size (nm)	3.1 ± 1.0	3.3 ± 1.1	3.4 ± 1.1
number of particles	500	430	370
TiONx Support Composition (atom %) (Determined with EELS)
N	20 ± 3	19 ± 5	22 ± 3
O	34 ± 4	34 ± 4	32 ± 3
O/N	1.71 ± 0.46	1.98 ± 0.77	1.5 ± 0.36
TiONx Support Composition (atom %) (Determined with XPS)
N	13 ± 1.3	15.6 ± 1.6	13.2 ± 1.3
O	53.4 ± 5	51.6 ± 5	54.3 ± 5
O/N	4.11 ± 0.41	3.31 ± 0.33	4.11 ± 0.41

Figure 3

STEM images of TiON-3h-Ir showing well-dispersed Ir nanoparticles (a) and atomic-resolution image of an individual Ir nanoparticle (b).

The EELS and XPS results show that the compositions are very similar (Tables and S1) for all three samples. As can be seen from the XPS results in Figure a, the surface concentrations of the titanium, oxygen, nitrogen, and iridium elements are independent of the anodization time. However, a difference in the O/N ratio determined with EELS (within the uncertainty of the measurement, the values are between 1.5 and 2) and XPS (within the uncertainty of the measurement, the values are between 3.3 and 4.1) can be observed (Table ). The reason for this is the oxidation of the TiON surface (upper 3–5 nm) that most likely occurred during milling. This is beneficial as it should protect the bulk material from further oxidation during the electrocatalytic experiments. The peaks observed in the Ti 2p XPS spectra (Figure b) are characteristic of TiO2 (Ti4+ at 458.6 eV), Ti–ON (457.2 eV), and TiN (455.7 eV).[41] Some 65% of the titanium is in the TiO2 state, while 20% is Ti–ON and 15% is TiN. The XPS spectra of Ir 4f (Figure c) are also similar for all of the samples. They mainly consist of peaks characteristic for iridium in the oxidation state zero (Ir0 at 60.9 eV) with a possible minor portion of iridium in oxidation state 4 (Ir4+) or 3 (Ir3+).[42] The metallic state of iridium is expected from the synthesis procedure as no oxidation step was introduced; however, it changes with the electrochemical oxidation.

Figure 4

Surface composition in atom % determined with XPS on TiON-nh-Ir samples (a). Normalized XPS spectra of Ti 2p (b) and Ir 4f (c) of milled TiON-1h-Ir, TiON-3h-Ir, and TiON-6h-Ir.

Independently, the presence of TiO2 in the samples was also confirmed using Raman spectroscopy (Figure ). The spectra reveal the presence of anatase[43] through the appearance of bands at 144 (Eg), 197 (Eg), 399 (B1g), 515 (A1g, B1g), and 625 (Eg) cm–1. However, the bands’ intensities differ for the different samples (Figure ) but also at various sites on the same sample (all four spectra recorded at various sites of each sample are shown in Figure S6 in the Supporting Information). The Raman spectra in Figure were chosen as the most representative, with the arrows marking the measured range of the 144 cm–1 band intensities. The intensities are the highest and showed the smallest spread for the TiON-3h-Ir sample. This indicates the largest ordering with respect to the anatase phase. Three out of four spectra exhibited considerably lower intensities of anatase bands in TiON-1h-Ir (Figure S6), which reveals that the TiO2 is closer to the amorphous state. The spectra of TiON-6h-Ir reveal the largest spread of the spectral intensities, indicating that the long-range ordering in anatase could be in any state between the 1 and 3 h samples.

Figure 5

Raman spectra of TiON-nh-Ir powders. The representative spectra are shown, while all of the measurements are gathered in Figure S6 in the Supporting Information. Two representative spectra are presented for the TiON-6h-Ir sample due to the largest spread of the intensity values. The arrows denote the intensity ranges of the 144 cm–1 anatase bands detected for each type of sample.

The basis of the anatase bands increases with an increase in the band intensity. Such effects have already been ascribed to the nanometric size.[44] However, a debate has been going on about the possible presence of TiN or TiON bands in this region. Namely, the Raman spectrum of TiN is characterized by a broad double band at 190–350 cm–1 and an additional one between 550 and 565 cm–1.[45,46] The relative ratios among those bands change with the deficiencies in the nitrogen TiN.[47] The described TiN, TiN, and TiON bands could be superimposed in the considered spectral range. For example, Prokes et al.[48] studied nitrided TiO2 nanopowder. The authors suggested the presence of TiO2–N due to the appearance of the shoulder band near 550 cm–1 in the Raman spectrum, dominated by the anatase bands. A similar assumption can be made for our TiON-nh-Ir samples since the band at 552 cm–1 can be observed in the spectra (Figures and S6). XPS (Figure b) reveals the largest amount of Ti–(O,N) and TiN in the TiON-3h-Ir sample, which matches our results from Raman spectroscopy (Figures and S6). The exact determination of the TiON, TiN bands and the solution to this dilemma demand a careful preparation of various standard compounds and a determination of their spectra, which will be addressed in our future work.

An interesting feature that also appears in the Raman spectra is a shoulder band at approximately 715 cm–1 (Figures and S6). Such a band can confirm the presence of the Ir4+ state observed by XPS (Figure c). Specifically, we detected 541 and 712–717 cm–1 Raman bands for various commercial IrO2 samples. The literature reports the appearance of IrO2 at 561 and ∼720 cm–1 in chemical-vapor-deposited and sputtered samples.[49] The deviation from the bands of the IrO2 single crystal (561 (Eg), 728 (B2g), and 752 (A1g)), i.e., the red-shift of the bands’ locations and asymmetric broadening, was ascribed to the nanometric size effects and residual stress effects. Consequently, 715 cm–1 could reflect the presence of oxidized Ir in the 4+ oxidation state. The second relevant band, i.e., 541 cm–1, could be blurred by the presence of the 515 cm–1 band of anatase and the vibration at 552 cm–1.

Ti K-edge X-ray absorption spectroscopy analysis (XANES and EXAFS) of the as-prepared and milled TiON-3h-Ir and TiON-6h-Ir was performed to compare the chemical state and local structure of the Ti cations in the samples (Figure S7a–c). The XANES results clearly show that the local structure, symmetry, and valence state of the Ti cations are the same in all of the samples. The anodic oxidation of the samples has no effect on the Ti chemical state in the sample, independent of the anodization time. We noted that the X-ray absorption spectroscopy analysis cannot distinguish the long-range ordering of the catalyst’s surface, which Raman spectroscopy does. Therefore, these two characterization techniques are very complementary.

The total amount of Ir in the samples was measured by ICP-OES. This is crucial for determining the appropriate electrocatalytic activity, which was normalized to the mass of iridium in each sample. The amount of iridium in all of the samples was approximately 11 wt %. Again, the observed morphological differences due to the different anodization time did not influence the amount and size of the deposited iridium nanoparticles on the TiON supports.

The chemical structures of all of the samples were determined with XRD analyses (Figures and S7). Figure shows the XRD spectra of the sample anodized for 3 h for all of the stages of its preparation. The TiO2 nanotubes grown by anodic oxidation are amorphous and show no diffraction peaks (Figure , top). Annealing in ammonia at 700 °C transforms the amorphous nanotubes into TiON (Figure , middle). The diffraction peaks of TiON are related to (⧫) cubic titanium oxide nitride at 37.1° (111), 43.1° (200), 62.5° (220), 75.0° (311), and 94.4° (400) angles (PDF 01-084-4872).[33] The XRD spectrum of TiON-3h-Ir is shown at the bottom of Figure . The additional diffraction peaks are observed at 40.7° (111), 47.3° (200), 69.1° (220), 83.4° (311), and 88.1° (222) angles, corresponding to cubic Ir (○) (PDF 04-007-8342).[33] The comparison of the XRD spectra of all three TiON-nh and TiON-nh-Ir samples is shown in Figure S8 and reveals that the chemical structures of all of the TiON-nh and all of the TiON-nh-Ir are the same. The diffraction domains of the surface TiO2 (for instance, anatase seen on TiON-3h) seen by Raman spectroscopy are obviously too small to sufficiently scatter the X-rays and be measured by our machine.

Figure 6

XRD spectra for amorphous TiO2 (a), TiON-3h (b), and TiON-3h-Ir (c) samples.

Electrochemical Activity of TiON-Ir

The electrochemical activity for the OER was determined for the case of several Ir-based analogues (Figure ). The commercial catalyst, IrO2 (Alfa Aesar), was chosen to represent the state of the art, while other homemade “benchmark catalysts” were synthesized to obtain Ir on TiN and Ir on TiON supports. The latter two were prepared either from the commercial TiO2 powder or from TiO2 nanoribbons (see Section ). The “benchmarks” were compared to the anodically grown analogues. This comparison is justified by the fact that such a selection makes possible an appropriate comparison of (i) the support type effect in terms of chemical composition as well as (ii) the potential effect of the morphology on OER’s performance. In addition, a thorough comparison with Ir nanoparticles on Ti-based support from the literature can be difficult due to the differences in ink preparation, activation protocol, ionomer–catalyst ratio, or electrochemical conditions. However, the comparison of our samples with regard to the current literature can be found in Table S2.

Figure 7

OER polarization curves. The curves were obtained at 20 mV s–1, in 0.1 M HClO4, 1600 rpm.

The influence of the electrode morphology is clearly seen when the anodically grown catalysts are compared to the nonanodized analogues. Evidently, the anodically grown samples are superior for the entire OER region (Figure ; see also Table ). The comparison demonstrates an up to twofold-higher OER activity in the case of the anodically grown analogues (measured at 1.55 V, Table ), which highlights the decisive role of the support morphology for the OER. More specifically, according to the structural analysis, the morphology of the TiON support does not affect either the iridium deposition or the extent of the dispersion of the Ir nanoparticles, i.e., a similar Ir particle size distribution is found for all anodically grown analogues (Figure S4). Additionally, even when the nanotube-array structure was reduced during mechanical grinding, the dispersion of iridium remained the same for all samples (Figures a and S5).

Table 2

Electrochemical Performances (Averaged over Three Measurements)

sample	mass activity at 1.55 V vs RHE (A gIr–1)	Tafel slope (mV dec–1)
TiONx-1h-Ir	360.9 ± 18.7	±60
after degradation (2 h)	245.4 ± 12.8	±72
TiONx-3h-Ir	520.3 ± 50.6
after degradation (1 h)	405.8 ± 39.5	±63
after degradation (2 h)	385.0 ± 37.5	±77
TiONx-6h-Ir	369.5 ± 35.1	±65
after degradation (2 h)	223.5 ± 21.2	±77
TiO2(P25)Nx-Ir	143.9 ± 12.1	±71
TiNIr	147.5 ± 11.4	±75
IrO2-AA	99.8 ± 14.3	±70
TiONx(NR)-Ir	213.2 ± 26.9	±77

Within the family of anodically grown analogues, the following trend in the OER’s performance is obtained. At low overpotentials (<1.52 V), the activity trend is in the following order: TiON-1h-Ir > TiON-3h-Ir > TiON-6h-Ir. However, at higher potentials (>1.52 V), this trend is altered: TiON-3h-Ir > TiON-6h-Ir > TiON-1h-Ir. To understand this better, it is important to note that the OER polarization curve consists of two regimes: (i) at lower potentials, the OER is governed by kinetics, which is influenced by the chemical structure of active sites,[50−52] their surface structure,[9,52,53] electronic interactions,[52,54] and interactions with the support;[15,55−58] while (ii) at higher potentials, the OER regime is controlled by electron transport and the transport/removal of the oxygen bubbles. The latter is caused due to confinement and results in the alternation of the local concentration of reactants or products. More specifically, it results in a decrease of local pH[59] or accumulation of evolved oxygen that can further lead to bubble formation. The gas-bubble effect not only limits the mass transport but also lowers the active surface area due to the generated O2 gas. The latter blocks the active sites and thus hinders the access of water as a reactant. Hence, the effective detachment of gas bubbles is essential for an efficient reaction to proceed at high potentials. Notably, the frequency and critical bubble diameter of the detachment decisively depend on the electrode morphology.[60−66] The results presented here indicate that (i) in the kinetic regime (low potentials) the nanotube-array morphology (characteristic feature in TiON-1h-Ir) decisively promotes the OER, whereas during the transition to higher potentials a combination of cluster size and nanotubular morphology (characteristic feature in TiON-3h-Ir) decisively governs the reaction proceeding. The cluster size in particular should decisively impact the OER proceeding via bubble management (at higher potentials, i.e., higher reaction rates).[60−66] However, also TiON particle-to-particle electron resistance through the oxidized surface contact points could lower the OER.[31,67] As TiON-1h-Ir has the lowest number of them statistically, this could also explain its higher OER activity at the onset. This reasoning fits well when considering the case of TiON-6h-Ir. Its OER performance in the low-potential region is the least among the three analogues, which can be ascribed to the absence of a nanotubular morphology and the highest amount of contact points, whereas at the same time, a more promising OER performance is shown at higher potentials. It seems that TiON-3h-Ir has the best tradeoff between the electron and mass transport issues.

Note that based on the XPS, EELS, and Raman analyses, the surface of the support for all three analogues is rich in the TiO2 phase (see discussion in Section ). As shown before, TiO2 can also induce passivation (besides TiON particle-to-particle contact points), which influences the electron access for Ir nanoparticles, and hence the OER’s kinetics.[68,69] To provide circumstantial evidence, we again rely on XPS and Raman analyses. Both techniques reveal the highest amount of the Ti–(O,N) phase to be present in the TiON-3h-Ir sample (Figures b, 5, and S6). This phase is important so as to retain sufficient electron conductivity for the OER (note that TiON has significantly larger electronic conductivity than TiO2 as directly shown in our recent study),[55] which is in agreement with the high catalytic activity of TiON-3h-Ir. Additionally, the Raman analysis reveals the different degrees of TiO2 crystallinity, where the TiON-3h-Ir analogue has the most crystalline and TiON-1h-Ir has the most amorphous TiO2 (Figure ). We hypothesize that the differences in the crystallinity should be ascribed to the milling pretreatment. A correlation of the degree of TiO2 crystallinity and the OER performance indicates that TiO2 crystallinity could play an important role at higher OER potentials/current densities (Figure ), which would be in line with the better electrochemical stability of the crystalline TiO2 in comparison to the amorphous variant.[36,70−72] If not stabilized properly, the TiO2 layer can become too thick for efficient electron conductance/tunneling.

To provide more insight, a Tafel analysis was performed. In the case of the anodically grown catalyst, similar Tafel slopes were found (60–65 mV dec–1, Table ). These values are comparable to the literature values for both the bulk, rutile IrO2 and the electrochemically grown oxide.[15,73−75] Similar Tafel slopes indicate that the rate-determining step (RDS) is the same for all analogues within the family of anodically grown catalysts. In comparison, the Tafel slopes of nonanodically grown catalysts are higher (70–77 mV dec–1, Table ). Considering the small difference in the slope, we could reasonably expect that the RDS is the same for the two composite types compared here. Therefore, alternative reasoning is needed to explain the slope difference between the anodically grown and the nonanodized analogues. The literature data for similar systems[15,76,77] suggests that the Tafel slopes are changed due to the interaction between the Ir nanoparticles and the support. It has been shown previously that an unusually large Tafel slope can be associated with these interactions when using semiconducting catalytic materials such as TiO2.[68,78] However, considering that both representatives of the TiON-based analogues (anodically grown and nonanodically grown) have the same chemical composition, we ascribe the steeper Tafel slope (the case of nonanodized samples) to the more extensive passive surface layer of TiO2.[68,78] This reasoning is based on the Tafel slope’s change after the electrochemical degradation (Table ), where the Tafel slopes of anodically grown analogues increased to the values of the nonanodized samples (see discussion below). This is in line with the electrochemical oxidation of TiON to TiO2 during the degradation protocol.[68,69]

Electrochemical Stability of TiON-Ir

We noted that a standard OER benchmark degradation protocol is currently still lacking.[65] Different accelerated degradation tests (ADTs) have been used in the literature.[14,31−34] The most common ones are chronopotentiometry and potential cycling in the region of the OER potential (e.g., between 1 and 1.6 V vs RHE). The constant-current protocol was chosen as it is more destructive than the potentiodynamic treatment.[65] Apart from that, cycling to a potential higher than 1.6 V vs RHE can induce bubbles and block the surface of the catalyst, which can protect the catalyst from degradation, and thus lead to the wrong conclusions.[65,79] Longer stability tests can also be an issue for the backing electrode. Indeed, here, glassy carbon electrodes were used, which can undergo degradation after long exposures to the high potentials. Thus, other backing electrodes, such as Au or Pt, should be employed for longer degradation tests in the future.[30] Here, a preliminary degradation test was performed on the catalyst by applying a constant current of 1 mA mgIr–1 for 2 h. Electrochemical stability was investigated for the case of the three anodically grown catalysts. Our results show that the most stable catalyst is TiON-3h-Ir, which retained 74.3% of the initial activity after 2 h (Figure ). Importantly, the majority of the performance loss occurs after half of galvanostatic perturbation; 78% of the initial activity is reached after 1 h (Figure ). This suggests that catalyst-activity decay is slowing down, which means it could sustain the majority of the performance for an extended operating time. The activity loss could originate from (i) the dissolution of Ir nanoparticles, (ii) the physical detachment of Ir nanoparticles, or (iii) the electrochemical oxidation of TiON, leading to the formation of a passive nonconductive TiO2 layer. However, according to a recent study, the effects of (i) and (ii) are not considered to play a significant role during degradation tests under ambient conditions and the timeframe of the thin-film rotating disk electrode (TF-RDE) setups.[65] Instead, we ascribe the activity loss to the influence of the electrochemical oxidation of the TiON support, leading to the formation of TiO2. This is supported by a slight increase in the Tafel slope after degradation perturbation (Table ). Additionally, according to the OER’s performance after the degradation protocol, we ascribe the performance–retention trend to the different crystallinity degree of the surface TiO2, where the TiON-3h-Ir analogue contains the most crystalline phase.[57] Even though additional dedicated studies are needed to investigate this effect further, it is reasonable to expect that the crystallinity of TiO2 would play an important role, based on studies of oxide analogues.[36,70−72]

Figure 8

Activity remaining for the different samples after 2 h at 0.1 A mgIr–1. TiON-1h-Ir in red, TiON-3h-Ir in blue, and TiON-6h-Ir in green.

Conclusions

We have shown that Ir in the form of nanoparticles on a TiON support made from anodically grown TiO2 nanotubes exhibits a 5-times-better mass activity than commercial IrO2 and at least 2-times-better mass activity than Ir on TiON with different morphologies. Three different anodization times were investigated for the preparation of the TiO2 nanotubes. The results for the OER’s activity were in the order TiON-3h-Ir > TiON-6h-Ir ≥ TiON-1h-Ir. The better activity was assigned to the differences in the morphology of the support, as other properties like Ir particle size and metal loading were kept the same. Furthermore, the ratio between O/N was also similar in all of the samples, as determined by EDX, EELS, and XPS. The amount of iridium was approximately 11%, as measured by ICP-OES, and the average Ir particle size was 3.3 nm, as observed in STEM. The best stability was found for TiON-3h-Ir, which is promising as it is also the most active catalyst. We ascribed its superior performance to the nanotubular morphology and the presence of the anatase TiO2 phase found on its surface by Raman spectroscopy. This anatase TiO2 passivates and thus protects the TiON surface and at the same induces beneficial effects on the Ir OER performance. In this way, we successfully synthesized an active and stable OER catalyst by supporting Ir nanoparticles on a TiON support made from anodically grown TiO2 nanotubes.