IrO₂-decorated Titania Nanotubes as Oxygen Evolution Anodes

1. Introduction

Ti-supported electrodes find applications in many industrial processes, most notably as dimensionally stable anodes (DSAs) in the chlor-alkali, electroplating and electrowinning industries [1,2,3] and as gas diffusion layer electrodes (GDLEs) in polymer electrolyte fuel cells (PEMFCs) and electrolyzers (PEMECs) [4,5]. When the anode reaction of interest is that of oxygen evolution (OER), IrO₂ is the preferred catalyst (especially under acidic conditions) due to the typically slow kinetics and high overpotential associated with the OER [6,7,8] resulting in the need for efficient and stable electrode materials. Minimizing the Ir catalyst loading is crucial given its scarcity and high price; this can be achieved by preparing very thin continuous films or highly dispersed IrO₂ particles over the Ti-based substrate [9,10,11,12,13]. In the latter case, to ensure electrode stability at high anodic potentials and/or aggressive conditions, there is a need for the presence of protective Ti oxides at uncovered locations (e.g. by thermal annealing) or the use of a TiO₂ substrate itself [14,15,16,17,18]; these strategies however, may induce particle agglomeration and/or increased ohmic losses due to thick TiO₂ layer formation.

Depending on the fabrication route, titanium dioxide can exist in various morphologies, such as nanoparticles [19,20,21], nanowires [22,23], nanofibers [24,25], nanosheets [26,27], each exhibiting distinct properties and offering a variety of potential applications. Among them, titania nanotubes (TNTs) have attracted considerable attention over the years owning to their high surface area and easily tunable geometrical parameters (nanotube length, pore size, tube orientation) [28,29,30]. In addition, electrochemical anodization of a Ti foil (the fabrication method most commonly used to produce nanotube structures), offers simplicity and low cost regarding the materials and instrumentation it requires. By adjusting the anodization parameters, such as duration, applied voltage, temperature and electrolyte composition, different nanotube morphologies can be obtained according to the preferred application [29,31,32]. Therefore, TNTs are ideal candidates to be used as OER catalyst supports, since they combine the inherent stability of titania oxides at high anodic potential values with a high surface area for the dispersion of IrO₂ and an open structure for the evolved oxygen gas to escape.

One major drawback of titania nanotubes (and TiO₂ in general) is the fact that they do not possess sufficient electronic conductivity and for that reason they should either be densely decorated by the conducting electrocatalyst or rendered themselves conducting. The conductivity of TiO₂ can be greatly impacted by the presence of defect states (Ti³⁺ interstitials and ${\mathrm{V}}_{\mathrm{O}}^{••}$ oxygen vacancies) in the crystal lattice or by the incorporation of different dopants to modify its band structure [33,34]. The former can be achieved by employing methods such as hydrogenation (annealing in a hydrogen rich atmosphere [35,36,37,38] or via hydrogen plasma treatment [39,40]), chemical reduction (via the reaction with strongly reducing reagents such as NaBH₄ [41,42,43] and CaH₂ [44,45]) or electrochemical reduction (by applying a sufficiently negative potential in buffered solutions) [46,47,48,49].

Most of the research studies reported in the literature focus on the photoelectrochemical applications of titania nanotubes, while few have explored their application as a support material for platinum group metal (PGM) catalysts to promote the oxygen evolution reaction (OER). Dehkordi et al. fabricated IrRuO_x/TNT electrodes through a modified Adams’ fusion method with 80 % w/w loading (of an Ir:Ru ratio equal to 60:40 at%), which showed enhanced activity and stability towards the OER compared to IrRuO_x supported on TiO₂ powder, due to the increased number of active sites and better catalyst distribution [50]. To lower the catalyst loading it is essential to improve the electronic conductivity of the TNTs, which will facilitate the electron transfer from the catalyst surface to the substrate during the OER. With this in mind, Genova-Koleva et al. synthesized IrO₂ and IrRuO_x catalysts supported on plain and Nb-doped TNTs with a 50 wt% catalyst loading, where the IrO₂/Nb-TNTs stood out owning to the high surface area and enhanced conductivity of the support, due to the presence of Nb(IV) species, which allowed also for a better catalyst dispersion [51]. In another study, Lu et al. used hydrogenated TNTs as support and IrO₂ was electrodeposited via potential cycling of the electrode substrate. The catalyst supported on the hydrogenated TNTs showed enhanced catalytic activity towards the OER compared to its plain semiconducting analogues, due to the higher deposition rate of IrO₂ that resulted in higher surface-active area electrodes [52]. Similar catalyst structures have been also investigated for the hydrogen evolution reaction (HER). More specifically, Lačnjevac et al. fabricated hydrogenated TNTs via cathodic reduction followed by galvanic deposition of metallic Ir from its IrCl₃ precursor solution to obtain Ir/H-TNTs. The authors emphasize that the strong interaction between Ir and the conductive TNT support is pivotal in enhancing the catalytic efficiency of these Ir/H-TNTs for the HER [53].

Our research group has focused on the production of PGM catalysts supported on various conducting substrates through the galvanic replacement/deposition method, which is driven by the difference in the standard potentials between a noble (e.g. Ir, Pt, Ru) and a less noble metal (e.g. Ni, Ti). Through galvanic replacement, a (mixed) core-shell or (mixed) core-skin structure is formed (depending on whether the substrate is in particulate or layer form), whereby the metal shell/skin is composed of the catalytic noble metal, and the core consists of a mixture of the noble metal with abundant, less expensive non-noble metals [54]. This process increases the catalytic surface area and enhances precious metal mass activity, allowing for a reduction in PGM loading in electrocatalyst applications. Furthermore, the interaction between the shell /skin and the core components is expected to modify the catalytic properties, potentially improving overall performance [13,54,55,56,57,58,59].

In this work, we investigate the feasibility of the galvanic replacement method on semiconducting and metallic-like black titania nanotubes, with either a close-packed or open structure, using Ni as a sacrificial layer for the galvanic deposition of Ir. The goal is to produce novel, efficient and stable IrO_x TNT-supported electrodes for use as OER anodes. SEM/EDS along with ICP-MS analysis was performed to assess the morphology and composition of the electrocatalysts. Electrochemical characterization was conducted by means of cyclic voltammetry (CV), near-steady state linear sweep voltammetry (LSV) and electrochemical impedance spectroscopy (EIS) measurements. Finally, the stability of the most active electrode towards the OER was investigated by means of constant current chronopotentiometry, while XPS analysis was conducted before and after the stability test to observe changes in its chemical composition.

2. Results

2.1. Microscopic (SEM) and Spectroscopic (EDS, ICP-MS, XPS) Analysis

In the case of two-step bTNTs (Figure 1a), the nanotubes are close-packed, vertically oriented with relatively smooth nanotube surface and distinct honeycomb-like morphology [44,60]. In contrast, the one-step bTNTs (Figure 1b) display a sparse arrangement of the nanotubes with an open structure. The difference in morphology may be attributed not only to the one- and two-step anodization processes but also to the use of a more water-rich anodization electrolyte in the case of open-structure nanotubes. In the two-step anodization process, the pre-formed pattern from the first step is utilized to grow highly ordered nanotubes, whereas in the one-step process, strict control over orientation is more challenging [61,62]. In addition, variations in the electrolyte composition further influence the structure. The water-rich electrolyte used to obtain open-structure TNTs results in increased solubility of the fluoride-rich layer of the nanotube walls, leading to a more open-structure with thinner walls [63,64,65]. Hence, the average nanotube diameter was 193 nm for the one-step/open structure (Figure 1b) and and 92 nm for the two-step/close-packed (Figure 1a) TNTs.

Following electrodeposition of Ni and galvanic replacement by Ir, it appears that the surface of the close-packed bTNTs (Figure 1c) and TNTs (Figure 2) is covered by large (>100 nm) Ir-Ni particles/aggregates that are blocking most/all the pores of the nanotubes. EDS elemental analysis of these close-packed electrodes gave an Ir:Ni atomic ratio of ca. 6 for the bTNTs and 0.13 for the TNTs substrates, indicating a larger remaining Ni content in the latter case, in line with the larger quantity of initially electrodeposited Ni that was needed to grow Ni from the more conducting bottom of the TNTs to their top surface (see discussion in Section 4.1.2 below and Figure S1). Also, in the case of the semiconducting-low electronic conductivity TNTs, Ir electron uptake/deposition is expected to occur only at conducting Ni sites (“protecting” them against further dissolution) and not on titania sites (as should be the case for electron-conducting bTNT sites, thereby leading to excessive Ni dissolution from neighboring uncovered Ni locations).

For the open-structure IrO_x(Ni)/bTNTs (Figure 1d) Ir nanoparticles with an average diameter of 74 nm (56–93 nm) can be seen, some residing at the rim or the interior of the pores (where sacrificial Ni apparently could be pre-deposited) and no Ni was detected by EDS, indicating its complete dissolution/replacement. To confirm that all Ni has been dissolved even from the surface of the open-structure bTNTs (and therefore, that Ir has been deposited on conducting open-structure bTNT sites), XPS analysis was also carried out for the as-prepared sample and no Ni was detected whereas Ir was detected at 4-7% atomic concentration depending on location.

XPS measurements were also performed after the stability test of the open-structure IrO_x/bTNT sample and no significant changes were observed in the surface chemical composition of the electrode. The corresponding XPS spectra for the as prepared and used IrO_x/bTNTs electrodes are shown in Figure 3. Figure 3a shows the Ir 4f spectrum, which consists of 4 main components. The 4f_7/2 and 4f_5/2 doublet corresponding to the main XPS 4f peaks for Ir⁴⁺ (IrO₂) and their two associated satellite peaks (at a binding energy ̴1 eV above the main peaks), which have their origin in the excitation of Fermi energy (E_F) electrons to unoccupied state at ̴1 eV above the E_F of IrO₂. The main difference between the two samples is the satellite-to-main peak intensity ratio, which is higher for the “used” sample. This can indicate that IrO₂ in the “used” sample has more available unoccupied states just above the Fermi level. Other possible interpretations are the presence of Ir³⁺, as its binding energy has been reported to be at a higher value than for Ir⁴⁺ [66] (there does, however, not seem to be a consensus on the peak position and shape of the 4f peaks for Ir³⁺) or the presence of hydroxides of Ir⁴⁺ [67]. Figure 3b shows the Ti 2p spectrum, which corresponds to Ti⁴⁺ and Ti³⁺. The 4+/3+ peak intensity (area) ratio seems almost unchanged between the two samples. Figure 3c shows the O 1s spectrum, showing components attributed to oxides, hydroxides and compounds containing adsorbed oxygen species. No significant differences between the two samples are observed. Finally, Figure 3d gives the C1s spectrum, as this peak was used to slightly correct the energy scale after charge compensation by setting the C-C component to 284.5 eV. The “used” sample has less oxidised C, suggesting its removal during OER operation in the acidic environment.

Finally, Ir loading was determined as 0.3 mg cm^-2 for the open-structure IrO_x(Ni)/bTNTs by the ICP-MS technique (following electrode etching as described in Section 4.3). Since both XPS, EDS and ICP-MS analysis revealed that Ni is not present in the open-structure bTNTs catalyst, it will henceforth be referred to as IrO_x/bTNTs instead of IrO_x(Ni)/bTNTs.

2.2. Electrochemical Characterization

2.2.1. Surface Electrochemistry of the Catalysts

The as-prepared Ir/Ni-, Ir-loaded electrodes were scanned multiple times in a 0.1 M HClO₄ deaerated solution with a scan rate of 50 mV s^-1, until stable voltammetry was recorded, in the potential range of:

• 0.0 V_RHE to +0.6 V_RHE, corresponding to the potential range between the onset of hydrogen evolution and the end of Ir double layer potential region, for the electrochemical dissolution of any surface uncovered/unreacted Ni [68,69] and the formation of an Ir(Ni) mixed core-skin structure [57] in those samples that contain Ni, until only the peaks attributed to the adsorption and desorption of an under-potentially deposited hydrogen (UPD-H) layer on metallic Ir could be clearly recorded, as depicted indicatively in Figure 4 for the case of open-structure Ir/bTNTs. The electrochemical surface area (ECSA) of metallic Ir is related to the charge associated with the adsorption of a H monolayer on Ir; the former could be calculated from the voltammograms of Figure 4 as 21.8, 49.4 and 239 cm²_Ir cm^-2 for the close-packed Ir(Ni)/TNTs, the close-packed Ir(Ni)/bTNTs and the open-structure Ir/bTNTs, respectively [13,56,70], in line with the increase in IrO₂ coating roughness/particle dispersion depicted in Figure 2, Figure 1c and Figure 1d respectively that correspond to these three different electrode types.

• 0.0 V_RHE to +1.2 V_RHE, where the characteristic reversible Ir surface oxide/hydroxide peaks of the Ir (III) ⇌ Ir (IV) redox transformations are recorded at E_a=0.85 V_RHE and E_c=0.75 V_RHE [59,69], as can be seen in Figure 5 for all electrode types,
• 0.0 _VRHE to +1.5 _VRHE, for the electrochemical anodization of metallic Ir to different oxidation states (IV, V) in order to form stable, porous 3D-IrO_x which also extends to the interior of the material [71,72,73]. In Figure 5 the electrochemistry due to anodically generated IrO_x appears above +1.0 _VRHE [59,74,75]. From the CVs, the charge corresponding to Ir oxides, which is representative of the electroactive surface area available for OER [76,77,78], was calculated as 16.6, 26 and 80 mC cm^-2 for the close-packed IrO_x(Ni)/TNT and IrO_x(Ni)/bTNTs, as well as the open-structure Ir_Ox/bTNTs, respectively (again, in line with coating roughness/particle dispersion shown in the SEM micrographs of Figure 1 and Figure 2).

2.2.2. Oxygen Evolution Reaction

The electrocatalytic electrodes were tested towards the OER by means of near-steady state experiments comprising of LSV curves recorded at 5 mV s^-1, from +1.2 to +1.7 V_RHE. The current interrupt method was applied at +1.2, +1.3, +1.4, +1.5 and +1.6 V_RHE, to estimate the uncompensated resistance and correct the applied potential for ohmic losses. The overall activity of IrO_x supported on untreated-semiconducting-low conductivity TNTs towards OER (Figure 6a) is inferior to that of IrO_x supported on reduced-conducting bTNTs, in line with its lower oxide electroactive surface area depicted in Figure 5 above and the continuous/non-particulate morphology of the corresponding deposit depicted in the SEM micrograph of Figure 2 above. Τhe IrO₂ catalyst supported on the open-structure bTNTs exhibited the highest currents-activity towards OER. This can be attributed again to the higher electroactive surface area of IrO₂ on these substrates (again, in accordance with its calculated IrO_x charge and its higher catalyst particle dispersion). Also, its open substrate morphology even after the introduction of the IrO₂

nanoparticle catalyst, should prevent clogging of the nanotubes with generated O₂ during the OER (which may be the case for the close-packed substrate electrodes). The Tafel slope for OER (Inset of Figure 6a) was estimated as 58, 75 and 70 mV dec^-1 for the three electrode types, which is common for IrO_x supported electrodes and a mechanism based on a rds that entails the chemical activation of adsorbed OH species (produced after the first, one-electron transfer step from H₂O to the anode electrode) [57,79,80].

Figure 6b presents data normalized for IrO₂ quantity and dispersion as described by the charge related to its surface electrochemistry (by integration of the anodic part of Figure 5). Hence, current values reported in this way take into account differences in electroactive area between samples and are therefore representative of samples’ intrinsic activity (in the absence of any other parameters such as mass transfer limitations). Although the onset potential for OER appears to be similar for all electrodes studied, the superiority of the open-structure IrO_x/bTNT electrode at higher overpotentials becomes apparent, pointing to the beneficial effect of the Ir-bTNT interactions (higher in the absence of remaining Ni on these samples-see Microscopic and Spectroscopic Characterization above) and/or of its open structure (see Figure 1d above) that limits extensive surface blockage by large oxygen bubbles.

2.2.3. Electrochemical Impedance Spectroscopy (EIS)

The response of the open-structure IrO_x/bTNTs electrode is presented in Figure 7 (as Nyquist diagrams) at potentials of +1.50 and +1.55 V_RHE within the potential range of the OER. In these diagrams, two distinct semicircles can be observed, with the second (larger) semicircle showing slight deformation at higher frequencies, a phenomenon more pronounced in other similar systems [57,77]. This suggests the co-existence of three semicircles that can be modeled using the equivalent electrical circuit R_sol(R_tQ_t)(R_pQ_p)(R_ctQ_dl). At very high oscillation frequencies, the solution resistance (R_sol) was determined where the first semicircle intersects the real axis (Z_re). Electrode porosity was also analyzed, with the first semicircle (modeled by R_tQ_t) attributed to the porosity of the IrO_x/bTNTs system (as depicted by SEM), and the second semicircle (R_pQ_p) reflecting the nano-porosity of the 3D IrO₂ oxides themselves. The pore resistance of iridium oxide is expected to be low since the dispersed fine Ir particles cannot accommodate dense and thick IrO_x layers. At lower frequencies, the third circuit (R_ctQ_dl) corresponds to the OER, representing the electron transfer resistance to the IrO₂ catalyst during OER and charging of the electroactive material double layer (i.e. associated with phenomena at the electroactive surface area/electrolyte solution interface where electron exchange during the OER occurs).

Table 1 below presents all the parameters of the equivalent electrical circuit R_sol(R_tQ_t)(R_pQ_p)(R_ctQ_dl) fitted to the EIS data. The R_ct is inversely related to the OER reaction rate. When multiplied by the double layer capacitance, C_dl (which is indicative of the actual surface of the electroactive material during the charge transfer of OER), it becomes the inverse of a time constant, representing in an inverse manner the intrinsic electrocatalytic activity of an electrode [77]. Q_dl (the constant phase element, CPE, circuit component) was converted to C_dl using the Mansfeld-Hsu equation [81] (Εq.1):

$${\mathrm{C}}_{\mathrm{d}\mathrm{l}}={\mathrm{Y}}_{\mathrm{d}\mathrm{l}}\mathrm{}{\mathsf{\omega }}^{{{\mathrm{n}}_{\mathrm{d}\mathrm{l}}}^{-1}},\mathrm{where}\mathsf{\omega }={\left({\displaystyle \frac{1}{{\mathrm{R}}_{\mathrm{c}\mathrm{t}}{\mathrm{Y}}_{\mathrm{d}\mathrm{l}}}}\right)}^{{\mathrm{n}}_{\mathrm{d}\mathrm{l}}^{-1}}$$

(1)

2.2.4. Stability Testing

To assess the stability of the most electrochemically active electrode (open-structure IrO_x/bTNTs) towards the OER, chronopotentiometry was employed and a current density equal to 10 mA cm^-2 was applied for 72 h in 0.1 M HClO₄ electrolyte. As can be seen from Figure 8, the electrode showed significant stability to the harsh acidic conditions during the OER and only a small shift of the potential equal to ca 20 mV was recorded.

Furthermore, from the XPS measurements before and after the stability test (Figure 3) no significant changes in the chemical composition of the surface of the as prepared and used electrode (IrO_x/bTNTs) were observed, indicating the stability of both the substrate (bTNT) and the catalyst (IrO_x) that ensures sustained conductivity and catalytic activity respectively.

3. Discussion

3.1. Catalytic Electrode Morphology

As can be seen from the SEM micrograph of Figure 1d, it is the combination of good electronic conductivity (bTNT substrate) and open structure (one-step anodization) that ensures the formation of well-dispersed IrO₂ nanoparticles on top and inside the mouth of the pores of such samples. In all other cases (close-packed bTNTs and TNTs) a densely populated surface or a continuous IrO₂ film is formed (Figure 1c and Figure 2a respectively). In the former case, the open structure ensures solution contact with the entire substrate surface hence facilitating the dissolution of pre-deposited Ni from all locations and the concurrent Ir deposition onto multiple conducting bTNT sites (during the galvanic replacement step). In the latter cases, the closed structure dictates Ni replacement preferentially on the surface of the substrate where it comes in contact with the Ir replacement solution. This interpretation is further supported by the fact that complete Ni dissolution was confirmed both by EDS, XPS and ICP-MS experiments in the case of the open-structure bTNT substrates.

3.2. Catalytic Electrode Performance

As can be seen in Figure 6a OER catalytic activity per substrate (nominal-projected) geometric electrode area follows the trend: IrO₂/bTNT > IrO₂(Ni)/bTNT > IrO₂(Ni)/TNT. This trend is due to a combination of surface area, intrinsic catalytic activity and mass transfer effects. Indeed, the same trend can be confirmed for catalyst dispersion/roughness by the corresponding SEM images of Figure 1 and Figure 2, as well as by the electroactive surface area itself, as estimated by the surface voltammetry shown in Figure 4 and Figure 5.

Normalizing for/removing the simple effect of surface area variation (Figure 6b) the trend remains the same and coincides with a decrease in Ni content/increase in direct IrO₂-bTNT interactions. It seems therefore that these IrO₂-bTNT interactions are more favorable to OER than those already reported for IrO₂-Ni [55,57,59,78]. As stated in [82] and reported in references therein, bTNTs prepared via solid state reduction by CaH₂ are characterized by the presence of neutral oxygen vacancies (with both the positive charge and electron polarons located at nearby Ti sites). These vacancies could act as sinks for O atoms of the IrO_x network destabilizing/activating the Ir-O-Ir bonds [78], a prerequisite of both the cation- and anion-type of mechanisms proposed for OER at IrO₂ [83].

As far as mass transfer effects are concerned (expected to intervene at potentials higher than those of pure kinetic control i.e. the ones within the potential range of the Tafel plots of the Inset of Figure 6a) these are also expected to diminish as one moves from a film to a particulate and finally to a catalyst-decorated electrode, since oxygen bubbles formed during OER at various locations are less likely to coalesce and form a mass transfer barrier to reacting H₂O. Indeed, the LSV curves corresponding to the IrO₂(Ni)/TNT electrode (Figure 6) show clear signs of the onset of a current plateau at the more positive potential values studied, characteristic of mass-transfer limitations.

When it comes to comparison with external standards, the best performing electrode of this work (open-structure IrO₂/bTNT) is estimated to show an overpotential of η=240 mV at 10 mA cm^-2 and current densities of 70 mA cm^-2 and 258 mA mg_Ir^-1 at η=300 mV. These values are comparable or better than similar electrodes reported in the literature (see Table 2).

As far as the EIS results are concerned, the existence of two semi-circles for the open-structure IrO₂/bTNT electrodes (unlike film IrO₂ or polymer-embedded IrO₂ particle electrodes that give rise to a single, deformed semicircle [77]) is indicative for the existence of two types of porosity of different scales in our samples, one of the TNT-supported IrO₂ and that of the porous 3D IrO₂ itself that has not been previously reported. R_ctC_dl values (bearing units of a time constant and being representative-in a reverse manner-of intrinsic catalytic activity [77]) have been estimated to be in the 178-71 Ω mF (ms) range, which translates to TOF values of 5.62-14.08 s^-1 for potentials in the +1.50 – +1.55 V_RHE range. These TOF values are higher than the ones reported for anodically grown IrO₂/Ir (0.4 – 0.286 s^-1) [77] and IrO_x(Ni)/GC electrodes prepared by galvanic replacement (0.496 s^-1) [57], and comparable or better than those for IrO_x(Ni)/Ti (11.48 s^-1 [13]), confirming again the beneficial effect of the bTNT (and, to a smaller extent, of the TNT) substrate in enhancing the catalytic activity of IrO₂ towards OER.

Finally, any reservations regarding the stability of the reduced bTNT support (that had to be used for best catalytic activity instead of TNTs) at the high anodic potentials of OER should be attenuated following the 72 h stable operation of these electrodes under OER in acidic conditions and the unchanged chemical states of both Ti and Ir, as confirmed by the XPS data of Figure 3.

4. Materials and Methods

4.1. Preparation of IrO₂ Catalysts Supported on TNTs and bTNTs

4.1.1. Preparation of Open-Structure and Close-Packed TNTs and bTNTs Substates

A Ti foil (0.25 mm thick, 99.5%, Thermo Fisher Scientific) was cut into 1 cm x 1cm squares and was sonicated for 10 min in acetone, ethanol and doubly-distilled (d.d.) H₂O, to remove surface contaminants. After sonication, the Ti foil was left to dry in air before further use. The anodization took place in a one compartment cell, with Pt foil (2.5 cm x 1.3 cm) as the counter electrode/cathode, which was placed 2 cm away from the working electrode/anode (Ti foil). By applying 60 V through a DC power supply (DSC Electronics, Model DP60-15H) in one anodization step for 120 min in glycerol solution (≥ 99.0%, Sigma Aldrich) containing 0.5%wt NH₄F (analytical grade, Sigma Aldrich), 10%wt d.d. H₂O, we were able to obtain open-structure TNTs (ca 2 μm long) [65].The close-packed TNTs were fabricated in two anodization steps by applying 60 V for 30 min+90 min in ethylene glycol solution (analytical grade, Sigma Aldrich), containing 0.25wt% NH₄F and 2wt% d.d. H₂O [89] (resulting in 10 ± 2 μm long nanotubes). After the first anodization step, scotch tape was used to remove the nanotube film, and the patterned Ti foil was sonicated for 10 min in ethanol and left to dry before the second step. Throughout the experiment, the temperature was kept at 25°C using a thermostat. To enhance the adhesion of the nanotubes to the Ti foil, the electrodes were kept in the anodization solution for 1 h [44]. The as-prepared TNT/Ti foils were annealed at 500°C for 2 h in a muffle furnace with a heating rate of 1.7^oC min^-1 to obtain TNTs with anatase structure. In the case of bTNTs, the substrate was annealed under the same heating conditions but in a sealed quartz ampoule in direct contact with CaH₂ (CaH₂ ≥ 97.0% powder, Sigma Aldrich), which resulted in the preparation of metallic-like conductive bTNTs [45,90].

4.1.2. Preparation of IrO_x(Ni)/TNTs and bTNTs

Initially, galvanostatic electrodeposition of Ni took place, in a one-unit cell in a Watt’s type of bath (30 g NiSO₄•6H₂O, 2.8 g NiCl₂•6H₂O, 4 g H₃BO₃ in 100 mL d.d. H₂O), with Pt as the auxiliary electrode and saturated calomel electrode (SCE) as the reference electrode. The applied current density was equal to -5 mA cm^-2 for close-packed TNTs, -3.7 mA cm^-2 for close-packed bTNTs and -2 mA cm^-2 for open-structure bTNTs, and it was selected based on the results of preliminary cyclic voltammetry (CV) scans to ensure Ni deposition occurred within the kinetic region/charge transfer control regime. The total charge density of the electrodeposited Ni was equal to 34-36 mC cm^-2 for bTNTs and 29.2 C cm^-2 for TNTs (with an expected 100% current efficiency for this type of Watts bath) and the temperature was held at 25^oC (bTNTs) and at 55^oC (TNTs) with constant magnetic stirring. The near thousand-fold higher charge density of electrodeposited Ni on the semiconducting TNTs was employed to achieve Ni penetration through the porous TNT substrate and down to its more conducting thin oxide/metal Ti base. In contrast to bTNTs, whereby nanotubes show good electronic conductivity along the nanotube walls, filling/partial filling of the pores of TNTs is essential to ensure electronic conductivity between the IrO₂-decorated electrode surface and the Ti base current collector (with Ti at the bottom of the TNTs expected to possess higher conductivity due to a thinner surface oxide layer). Figure S1 of the supplementary material presents a SEM top-view micrograph of the bottom part of a Ni/TNT film (originally in contact with its Ti base), detached from a Ni/TNT/Ti electrode. It can be seen that deposited Ni at the bottom of TNTs follows the size and well-ordered pattern of the TNT matrix grown on the Ti basis, thus confirming that the filling of those TNTs filled starts from the TNT/Ti interface. This is in line with what has been reported for other annealed TNTs [91,92] and electrochemically treated amorphous TNTs [93].

Immediately following Ni deposition, the Ni/TNTs and Ni/bTNTs electrodes were immersed in a freshly prepared deaerated exchange solution (pH≈ 3), consisting of 1 mM HCl + 1 mM K₂IrCl₆ (≥99.9% trace metal basis, Sigma Aldrich) for 15 min at 65^oC, for the galvanic replacement of Ni particles or layers by Ir. During galvanic replacement, metallic Ni is oxidized and dissolves as Ni²⁺ in the solution while Ir⁴⁺ is reduced to metallic Ir on the substrate surface (by electron uptake at Ni or conducting bTNT sites; Eq.2), due to the positive difference in the standard reduction potentials of the Ir^(IV)/Ir^(III) (E⁰=+0.86V vs. SHE) and Ni^(II)/Ni (E⁰=–0.257V vs. SHE) redox couples [54].

xNi + IrCl₆^2– → Ir(x-2)Ni + 2Ni²⁺ + 6Cl^–

(2)

Finally, electrochemical anodization of metallic Ir was conducted by means of potential cycling with CV in 0.1 M HClO₄ (70%, Merck), in the potential range between H₂/O₂ evolution, to form porous IrO_x and dissolve any surface Ni not covered by Ir [13]. A three-electrode cell was used for this purpose, with Pt as the counter and SCE as the reference electrode.

4.2. Electrochemical Setup and Procedures

The electrochemical characterization of the as-prepared electrodes was conducted in a three-electrode cell, with Pt as the counter electrode and SCE as the reference electrode, which were separated from the working electrode by porous glass frits, to facilitate ionic conductivity in the bulk electrolyte solution. Cyclic and linear sweep voltammograms as well as EIS spectra were recorded with the help of an Autolab PGSTAT302N (Eco Chemie, Utrecht, The Netherlands) workstation, controlled via NOVA 1.11.2 software (Eco Chemie, Utrecht, The Netherlands). All electrochemical processes were carried out at room temperature.

CV was employed to investigate the surface electrochemistry of the catalysts, using consecutive scans in three potential ranges at a scan rate of 50 mV s⁻¹ in 0.1 M HClO₄ until stable voltammetry was recorded. The first range (-0.3 V to + 0.3 V_SCE) targeted the dissolution of unreacted Ni and the formation of an Ir(Ni) core-shell structure [57]. The electrode was transferred in a fresh deaerated 0.1 M HClO₄ solution and scanned again. In the second range (-0.3 V to +0.9 V_SCE), the formation of reversible Ir(III)/Ir(IV) oxides/hydroxides occurred, while in the third range (+0.3 V to +1.2 V_SCE), anodization of Ir to higher oxidation states (Ir(IV) and Ir(V)) formed a stable, porous 3D-IrO_x structure [71,72,73].

Electrocatalytic activity towards the OER was evaluated using linear sweep voltammetry (LSV) at a potential sweep rate of 5 mV s⁻¹ between +0.9 V and +1.4 V_SCE. The current interrupt method was applied at various potentials (+0.9 to +1.3 V_SCE) to estimate uncompensated resistance and correct the applied potential for ohmic losses. Chronopotentiometry was used to assess the stability of the electrodes for the OER by applying a constant current of 10 mA cm⁻² for 72 h.

Finally, EIS measurements were carried out to verify the value of the solution resistance (R_s) that was previously determined using the current interrupt technique during the LSV, and to calculate both the charge transfer resistance (R_ct), characteristic of OER activity, and the double-layer capacitance (C_dl), which is representative of the electroactive surface of the catalyst. EIS studies were conducted in the frequency range between 6 kHz and 0.1 Hz at DC potential values of +1.20 and +1.25 V_SCE, recording the current response at an AC voltage amplitude of 10 mV. To find the equivalent circuit and determine values for the associated components, the NOVA 2.1 software was used.

The potential values reported are converted to the reversible hydrogen electrode (RHE) using the equation:

ERHE = ESCE + 0.244 + (0.059 x pH)

(3)

The equilibrium potential for the oxygen evolution reaction in 0.1 M HClO₄ at room temperature is calculated as + 0.927 V_SCE.

4.3. Microscopic and Spectroscopic Characterization

The as-prepared TNTs and their modified bTNTs analogues were observed with Field-Emission Scanning Electron Microscopy (FESEM, JSM-7610F PLUS, JEOL Ltd, Akishima, Japan) supported by an Energy Dispersive X-Ray Spectroscopy system (EDS, Oxford, Instruments Ltd, Oxford, UK), to obtain information about the morphology and the relative composition of the samples. In order to determine the quantity of deposited Ir supported on the best performing open-structure bTNTs, trace metal analysis was carried out with Inductively Coupled Plasma-Mass Spectroscopy (ICP-MS). The electrode was dissolved in boiling aqua regia (37% HCl, ChemLab; 65% HNO₃, Merck) for 20 min and after reaching room temperature, the leachate was diluted in 2% v/v HNO₃. The analysis was conducted by a Thermo Scientific iCAP Q ICP-MS, controlled via Q Tegra software. In addition, X-ray Photoelectron Spectroscopy (XPS) analysis was conducted to assess the surface atomic composition and the analysis was performed in a Thermo Scientific Thetaprobe X-ray Photoelectron Spectrometer, with a monochromated Al-Ka1 X-ray beam (hν = 1486.6 eV) and an analysis area of approximately 400 µm diameter.

5. Conclusions

• Electrodeposition of sacrificial Ni on conducting bTNTs and its subsequent galvanic replacement by Ir resulted (depending on substrate type) into Ir particles (<100 nm) for open-structure bTNTs or into larger aggregates for close-packed, bTNTs; in the case of open-structure bTNTs these particles were highly dispersed (some residing inside the nanotubes), thus increasing the electroactive area while at the same time retaining an open electrode structure.
• For the semiconducting TNTs, an increase in the charge of electrodeposited Ni was necessary to rectify their original low electrical conductivity and resulting in the filling of the nanotubes for Ni deposits to act also as a current collector. The eventual formation of a continuous Ir(Ni) film on the surface following galvanic replacement, resulted in a significant decrease in the electroactive surface area.
• The open-structure IrO_x(Ni)/bTNTs (more precisely, IrO_x/bTNTs since no Ni has been detected after the galvanic replacement/metal exhange process), exhibited an enhanced activity towards the OER. This can be attributed to the higher surface area of the support, higher Ir dispersion and catalytic activity (due to IrO₂-bTNT interactions) as well as less pore clogging during O₂ evolution. An overpotential of η=240 mV at 10 mA cm^-2 and a mass-specific current density of 258 mA mg_Ir^-1 at η=300 mV has been recorded, rendering them comparable or better than similar electrodes reported in the literature. Furthermore, the optimized electrodes, when tested for prolonged periods of time under OER conditions, were characterized by good short-term (72 h) stability.