Dissolution Stability: The Major Challenge in the Regenerative Fuel Cells Bifunctional Catalysis

The hydrous Ir oxide (IrO_x) was prepared using a hydrothermal synthesis method.³⁸ Initially, a solution containing 5 g of H₂IrCl₆.xH₂O (99 wt%, Alfa Aesar) in 100 mL of ultrapure water (18.2 MΩ cm, Milli-Q) was prepared. Then, a 1 M LiOH solution (99 wt%, Sigma-Aldrich) was added to 10 mL of the Ir precursor solution, under stirring, at a rate of 0.2 mL h⁻¹, until a final ratio of [OH⁻]/[Cl⁻] = 6 was achieved. In the next step, the obtained Ir blue suspension was transferred to a polytetrafluorethylene (PTFE) chamber in a stainless steel autoclave reactor, and heated at 180°C for 24 h. After the 24 h heating, the IrO_x black precipitate was filtered, washed with boiling water, and dried at 50°C for 1 h.

The rutile Ir oxide (IrO₂) was obtained after heat treating the IrO_x catalyst at 400°C for 1 h, under air atmosphere.

The Pt/IrO_x and Pt/IrO₂ bifunctional catalysts with Pt:Ir atomic ratios of 1:1 were synthesized by depositing Pt nanoparticles on the Ir oxide supports using a formic acid method.³⁹ In the first step, the oxide support was ultrasonically dispersed in 30 mL of a 2 M formic acid solution (98 wt%, Sigma-Aldrich) and heated to 80°C. After that, an aqueous 5g/100 mL H₂PtCl₆.xH₂O (99 wt%, Alfa Aesar) solution was added dropwise, under stirring. Five minutes after adding the Pt precursor, the solution was allowed to cool down, the catalyst was filtered off, washed with boiling water, and dried at 50°C for 1 h.

Energy dispersive X-ray spectroscopy (EDS) was performed with a Leica-Zeiss LEO 440 scanning electron microscope (SEM) operating at 40 kV. Powder X-ray diffraction was conducted with a Bruker D8 Advance X-ray diffractometer (XRD) with 1.5406 Å Cu Kα radiation; data was collected over a 2θ range of 10–100°, at a scanning rate of 0.075° s⁻¹. The Rietveld refinement method was used to obtain the structural parameters of the synthesized catalysts.⁴⁰

Transmission electron microscopy (TEM) images were acquired with a JEOL JEM 2100 microscope, fitted with a LaB₆ filament, operating at 200 kV accelerating voltage. Scanning transmission microscopy (STEM) and EDS mapping images were obtained with a field emission gun FEI Tecnai G² F20 microscope operated at 200 kV acceleration voltage. In both TEM and STEM analysis, the catalyst powders were dispersed in isopropanol and deposited on a Formvar carbon-coated copper grids (EMS, 400 mesh).

X-ray photoelectron spectroscopy (XPS) was carried out with a ScientaOmicron ESCA+ spectrometer equipped with a high-performance hemispheric analyzer (EA 125), and a monochromatic radiation of 1487 eV; the biding energies of all XPS spectra were calibrated according to the C 1s peak energy at 284.8 eV.

Electrochemical characteri-zation

OER and ORR catalytic activity

Catalysts' electrochemical dissolution stability

Firstly, physicochemical characterization of the studied bifunctional catalysts was undertaken. Atomic and mass compositions of the studied Pt/IrO_x and Pt/IrO₂ bifunctional materials were estimated using energy-dispersive X-ray spectroscopy (EDS) with results presented in Table S1. The EDS results show that Pt deposition on the IrO_x and IrO₂ supports was successful, resulting in Pt/IrO_x and Pt/IrO₂ catalysts with Pt:Ir atomic ratios close to the nominal value of 1:1. Considering the oxygen content in the catalysts, an empirical formula of PtIrO₂ and PtIrO_1.7 can be obtained for the Pt/IrO_x and Pt/IrO₂ materials, respectively. It has to be noted, however, that IrO_x support has relatively complex structure, not reflected by this formula. As an example, using the thermogravimetric analysis to elucidate the composition of hydrothermally prepared IrO_x, Bestaoui and Prouzet suggested that the compound consisted of a hydrated Ir oxihydroxide, with formula IrO_1.45(OH)_1.10·1.5H₂O, while rutile IrO₂ phase can be obtained upon calcination.³⁸

The X-ray diffraction (XRD) profiles of the IrO_x and IrO₂ materials before Pt loading are shown in Figure S1. The XRD diffractogram of the IrO_x catalyst exhibits the existence of two broad peaks with low intensity, characteristic of an amorphous material. These peaks are attributed to the most intense reflections of the rutile⁴² and hollandite^43,44 phases of IrO₂ and iridate solid solutions, respectively. This result is in line with literature reports.⁴⁵ Regarding the IrO₂ support, the diffraction peaks corresponding to the tetragonal IrO₂-rutile structure (PDF 15-870) are clearly seen. The crystalline structure of the support materials is also reflected in the XRD data of the bifunctional catalysts presented in Fig. 1a. Due to the amorphous nature of the IrO_x, the Pt/IrO_x catalyst only exhibits peaks corresponding to the reflection of the Pt face-centered cubic (fcc) (PDF 4-802) phase, while for the Pt/IrO₂ material, peaks corresponding to the reflection of tetragonal IrO₂-rutile (PDF 15-870) and (fcc) Pt are observed.

Zoom In Zoom Out Reset image size

Fig. 1b shows the magnified region of the Pt(200) diffraction peaks, which was chosen as it does not overlap with any diffraction peaks of the IrO₂ structure. Here, a shift to a higher angle can be observed for the Pt/IrO_x catalyst, while for the Pt/IrO₂ material the Pt(200) peak position coincides with that of the Pt black, which is in line with literature.¹⁷ It is suggested that there is a lattice contraction in the Pt crystalline structure supported by IrO_x. The mean crystallite sizes and lattice parameters of the studied catalyst were calculated using the Rietveld refinement technique⁴⁰ and the results are compiled in Table S2. The calculated mean crystallite sizes (D) of the Pt phase in the Pt/IrO₂ and Pt/IrO_x catalysts are 3.8 and 9.9 nm, respectively. Regarding the lattice parameters (a_exp), the calculated value for the Pt/IrO_x material was 3.908 Å, smaller than those of Pt/IrO₂ (3.923 Å) and Pt black catalysts (3.918 Å), confirming a contraction in the Pt lattice for Pt/IrO_x. For the IrO₂ phase in the studied catalysts, an increase in the IrO₂ crystallite size is observed after the Pt deposition procedure, ranging from 3.4 for the IrO₂ material to 5.9 for the Pt/IrO₂ catalyst. The origin of this particles increase is still to be clarified. The structural parameters of the IrO_x phase could not be calculated due to its amorphous nature.

The catalysts' surface composition was evaluated using the X-ray photoelectron spectroscopy (XPS) technique. The high resolution spectra of the Pt 4f and Ir 4f orbitals of the Pt/IrO_x and Pt/IrO₂ catalysts are shown in Figure 1c. The Ir XPS spectra can be deconvoluted into two doublets, which correspond to the Ir 4f_7/2 and Ir 4f_5/2 spin-orbit components with energy separation of 3.2 eV and fixed area ratio (Ir 4f_5/2: Ir 4f_7/2) of 3:4. Table S3 summarizes the Ir 4f_7/2 binding energies and the percentage contribution of the different Ir species in the bifunctional catalysts. The first peak centered at the lower binding energy corresponds to the Ir³⁺ oxidation state, while the second peak is attributed to the higher oxidation of Ir⁴⁺.⁴⁶ The XPS data shows that for Pt/IrO_x catalyst 53% of the Ir species are in the Ir³⁺ oxidation state, while for the Pt/IrO₂ material most of the Ir atoms are in the Ir⁴⁺ form (66%). The presence of Ir³⁺ species in the IrO₂ support implies that the calcination at 400°C for 1 hour is not enough to fully convert the amorphous hydrous oxide to the rutile. Higher crystallization can be obtained using a higher calcination temperature and/or with a longer calcination time.^22,38,47 An additional peak centered at ∼67.5 eV can be observed in both Ir 4f spectra, which corresponds to a satellite due to the conduction-band interactions during the electrons photoemission.⁴⁸ Figure S2 shows the XPS spectra of the O 1s orbital of the bifunctional catalysts. A shift to lower binding energies can be seen for the O 1s spectrum of the Pt/IrO₂ catalyst compared to the one of the Pt/IrO_x material, which corresponds to a conversion of the oxihydroxide in the amorphous support to the rutile Ir oxide in the Pt/IrO₂ catalyst.

Regarding the Pt 4f XPS data, the spectra can be deconvoluted into three doublets with peaks separation of 3.35 eV and area ratio (Pt 4f_5/2: Pt 4f_7/2) of 3:4, which corresponds to Pt in the Pt⁰, Pt²⁺, and Pt⁴⁺ oxidation states.⁴⁹ The binding energies of the Pt 4f_7/2 components and the percentage contribution of the different species are shown in Table S3. The Pt surface atoms in both catalysts are mostly found in the metallic form (Pt⁰); for the Pt/IrO_x catalyst, the Pt⁰ percentage is close to 70%, while for the Pt/IrO₂ material a lower percentage was found (56%). The XPS data was also used to estimate the Pt/Ir ratios in the catalysts' surface layer. The calculated Pt/Ir ratios for the Pt/IrO_x and Pt/IrO₂ catalysts are close to 3/1 and 2/1, respectively. Here, since Pt nanoparticles are deposited on the Ir oxide supports, and XPS is a surface-sensitive technique, the obtained Pt content is higher than that from EDS data (Table S1), which corresponds to a bulk analysis.

The TEM micrographs of the bifunctional Pt/IrO_x and Pt/IrO₂ materials are displayed in Figures 1d and 1e, respectively. The images show that for both materials the Pt and IrO_x/IrO₂ nanoparticles are highly agglomerated, making the distinction between the Pt phase and the oxide support difficult. Because of the agglomeration of the catalyst, the mean particle sizes cannot be precisely estimated. The TEM images of the IrO_x and IrO₂ catalysts without Pt are shown in Figure S3. The IrO_x catalyst is composed by spherical nanoparticles with particle size in the 1–2 nm range (Fig. S3a). The smaller IrO_x particles favors the deposition of Pt over itself, resulting in bigger crystallite size, and, consequently, in a higher Pt concentration in the catalyst surface, as shown in the analysis of XPS data. For the IrO₂ material, the agglomeration and particle size increase as consequence of the calcination procedure (Fig. S3b).

To better understand how Pt nanoparticles are dispersed on the oxides supports, EDS chemical mapping was performed, and the results for the Pt/IrO_x and Pt/IrO₂ materials are exhibited in Figures S4 and S5, respectively. One can note that for both catalysts the Pt nanoparticles are well distributed over the entire surface of IrO_x and IrO₂ materials. For the Pt/IrO₂ catalyst, some Pt richer regions can be distinguished, which is the result of some particles agglomeration.

The cyclic voltammetry (CV) profiles of the investigated bifunctional Pt/IrO_x and Pt/IrO₂ catalysts are exhibited in Fig. 2. For comparison purposes the CV profiles of the Pt black, IrO_x and IrO₂ materials are also shown in Fig. S6. Both CVs in Fig. 2a have higher double-layer currents compared to that of pure Pt (Fig. S6a). The main difference between the voltammetric profiles of the Pt/IrO_x and Pt/IrO₂ is observed in the so-called "hydrogen region" in the 0.05–0.4 V range. While the Pt/IrO₂ catalyst exhibits well-defined peaks corresponding to the adsorption/desorption of hydrogen on Pt, in a similar way to Pt black (Fig. S6a) and others Pt/IrO₂ catalysts,^20,21 for the Pt/IrO_x material no peaks can be seen, resulting in a profile comparable to that for the IrO_x (Fig. S6b) and others hydrous Ir oxides.^50,51

Zoom In Zoom Out Reset image size

The CO stripping voltammetry profiles of the bifunctional catalysts are displayed in Fig. 2b, which also includes the data obtained for the Pt black catalyst. The Pt/IrO_x stripping profile exhibits a single peak centered at ∼0.69 V with a shoulder at ∼0.73 V, such profile is similar to that of Pt black material, and had also been reported for others Pt catalysts.^52,53 Regarding the Pt/IrO₂ catalyst, the CO stripping profile exhibits two peaks: a main peak at ∼0.67 V and a higher potential peak at ∼0.75 V. The Pt electroactive surface areas (A_ECSA) were calculated by integration of stripping peaks of Fig. 2b, and the values are shown in Table I. The A_ECSA values for the Pt/IrO_x and Pt/IrO₂ materials are 10.5 and 14.0 m² g_Pt⁻¹, respectively, which are bigger than that for the Pt black catalyst (7.7 m² g_Pt⁻¹).

Oxygen reduction reaction polarization curves in sulfuric acid electrolyte are presented in Fig. 2c. A positive current at E > 0.9 V can be observed for the bifunctional catalysts as a result of the influence of IrO_x/IrO₂ capacitive currents. Argon-corrected polarization curves can be seen in Fig. S8. The current densities (j) at 0.85 V are summarized in Table I. The Pt/IrO₂ (2.3 mA cm⁻²) catalyst exhibits a similar current density to that of the Pt black (2.5 mA cm⁻²), even with just half the Pt loading, and higher than that of the Pt/IrO_x catalyst (1.3 mA cm⁻²). A similar trend is obtained considering the specific activity values (j_SA). In this case, the Pt/IrO₂ (87.1 μA cm_Pt⁻²) and Pt black (85.3 μA cm_Pt⁻²) catalysts exhibit similar j_SA values, and those are ∼30% higher than that for the Pt/IrO_x material (65.3 μA cm_Pt⁻²). The catalysts' mass activities of the studied materials are shown in the mass transport corrected Tafel diagrams in the Fig. S9a. All curves exhibit two linear regions with slopes close to 60 and 120 mV dec⁻¹ (for the sake of simplicity just shown for Pt black). At potential values lower than 0.9 V (E < 0.9 V), the Pt/IrO₂ has an ORR mass activity higher than that of Pt black, while the Pt/IrO_x catalyst presents the poorer catalytic performance. Mass normalized currents (j_m) calculated using the Koutecky-Levich diagrams (not shown) are presented in Table I. The Pt/IrO_x and Pt/IrO₂ catalysts exhibit j_m values of 9.8 and 35.1 mA mg_Pt⁻¹, respectively. The j_m value obtained for the Pt/IrO₂ catalyst is ∼2.3 times higher than that of Pt black (15.5 mA mg_Pt⁻¹), while that for the Pt/IrO_x is ∼40% lower than that of the pure material.

The catalysts' activities for the OER are shown in the polarization curves in Fig. 2d. The current density values at 1.55 V are presented in Table I; the Pt/IrO_x catalyst exhibits a current density of 11.2 mA cm⁻², a value almost 2.5 times higher than that for the Pt/IrO₂ material (4.7 mA cm⁻²), and similar to that of pure IrO₂ (12.2 mA cm⁻²), even with half the Ir load. Mass normalized OER polarization curves are displayed in Fig. S9b, while the mass activities (j_m) currents at 1.55 V are shown in Table I. The current density of the Pt/IrO₂ and IrO₂ catalysts at 1.55 V are 24.5 and 32.0 mA mg_Ir⁻¹, respectively. Regarding the Pt/IrO_x catalyst, the j_m at 1.55 V is 60.4 mA mg_Ir⁻¹, also lower than that of pure IrO_x (79.6 mA mg_Ir⁻¹). However, the polarization curves in Fig. S9b show that for potentials values higher than 1.58 V the Pt/IrO_x catalyst becomes more active than the IrO_x material, which may be related to better electronic conductivity or improved gas release.

The catalysts' stability in terms of electrochemical dissolution was evaluated using an SFC-ICP-MS setup under three protocols, comprising cycling in the ORR (0.6–1.1 V), OER (1.1–1.6 V), and ORR–OER (0.6–1.6 V) potential windows. Potential resolved dissolution profiles for Pt and Ir are displayed in Fig. 3.

Zoom In Zoom Out Reset image size

Results for the ORR potential range in Fig. 3a reveal that while the Pt dissolution process cannot be clearly resolved during the positive-going sweep (anodic dissolution), the dominating dissolution process is taking place during the negative-going sweep (cathodic dissolution). Previous dissolution data for polycrystalline Pt and Pt nanoparticles shows that at the chosen scan rate, using the current experimental setup, both dissolution processes can only be resolved when cycling to higher potentials.^24,32 Similarly to Pt, cathodic dissolution is the main Ir degradation process, which is in line with results for electrochemically grown IrO_x.⁵⁴ Considering the cathodic dissolution for the first cycle, the potential onset, estimated as the potential at which signal-to-noise S/N = 3, for Pt and Ir dissolution is ∼1.0 V and ∼0.7 V. These values are similar to that of metallic Pt and Ir electrodes.^54,55 In this protocol, the dissolution amount of Pt is almost 3 times higher than that for Ir in the Pt/IrO_x and Pt/IrO₂ materials. For both materials, dissolution of Pt and Ir is rather constant along cycles.

Fig. 3b shows the dissolution profiles of the bifunctional catalysts cycled in the OER potential region. One can see that during the first cycle both materials exhibit a Pt anodic dissolution peak with potential onset at 1.20 V, accompanied by a less intense reduction peak at 1.15 V. In the second and third cycles the anodic dissolution peaks are almost imperceptive, while cathodic Pt dissolution, although small, is still observed. The anodic dissolution of Ir from IrO₂ and especially IrO_x supports is higher in this potential region than in the ORR potential range (Fig. 3a). The potential onsets for Ir dissolution of Pt/IrO_x and Pt/IrO₂ catalysts are 1.45 V and 1.52 V, respectively. These values are in line with the OER onsets shown in Fig. 2d, and are higher than those reported for electrochemically grown IrO_x (0.7–0.8 V)⁵⁴ and for IrO₂ (1.30 V) catalyst obtained at 450°C.²³ Ir is not dissolving cathodically in this region as much lower cathodic potential is required to reach cathodic dissolution on Ir.⁵⁴

Finally, when cycled between 0.6–1.6 V (Fig. 3c), the transient dissolutions of Pt and Ir, composed of an anodic and cathodic counterparts, are clearly seen. Compared to the protocols used in Figs. 3a and 3b, cycling in between 0.6–1.6 V increases the catalysts' instability. In this protocol dissolution of Pt is much higher, which is in line with literature reporting high Pt dissolution when potential protocol includes potentials of Pt oxidation and reduction.²⁴ Also cathodic dissolution of Ir is higher in this protocol.

The Pt and Ir dissolved amounts in each protocol can be calculated by the integration of dissolution profiles in Fig. 3. Summarized data on dissolution of Ir and Pt in the studied protocols are shown in Fig. 4. While cycling in the ORR region (0.6–1.1 V), the calculated (three cycles-averaged) dissolved amounts of Pt and Ir for the Pt/IrO_x catalyst are 7.3 ± 0.7 and 2.5 ± 0.4 pg cycle⁻¹, and for the Pt/IrO₂ material values are 6.6 ± 0.2 and 2.1 ± 0.03 pg cycle⁻¹, respectively. Cycling in the OER region (1.1–1.6 V), one can note an increase in the amount of dissolved Ir; for the Pt/IrO_x catalyst, the measured Ir dissolution is 21.6 ± 1.0 pg cycle⁻¹, a value more than 4 times higher than that for the Pt/IrO₂ (5.1 ± 0.4 pg cycle⁻¹). Regarding the Pt dissolution, cycling in higher potentials do not cause a significant increase in the dissolved amounts for the Pt/IrO_x (7.4 ± 0.3 pg cycle⁻¹) and Pt/IrO₂ (7.9 ± 0.3 pg cycle⁻¹) bifunctional catalysts compared to the protocol involving cycling in the ORR region. Significantly higher dissolution rates are observed when the catalysts are cycled in the 0.6–1.6 V range. The calculated dissolved amount of Ir for the Pt/IrO_x material is 45.1 ± 8.6 pg cycle⁻¹, and 23.6 ± 4.3 pg cycle⁻¹ for the Pt/IrO₂ catalyst. For Pt dissolution, the calculated amounts are even higher, being 115.2 ± 1.5 pg cycle⁻¹ for the Pt/IrO_x, and 143.9 ± 2.8 pg cycle⁻¹ for Pt/IrO₂. The dissolution rate calculated for the Pt black material (Fig. S10) under the same protocol is 130.4 ± 2.1 pg cycle⁻¹, this value is similar to that of the bifunctional catalysts, even with double Pt loading, and shows that supporting Pt in IrO_x and IrO₂ does not increase its stability in terms of dissolution.

Zoom In Zoom Out Reset image size

The Pt/IrO_x and Pt/IrO₂ catalysts exhibit higher currents in the so-called double-layer and Ir oxidation/reduction potential regions compared to the Pt black material, which is due to the capacitive behavior of the oxide supports, especially IrO_x. Interestingly, both bifunctional catalysts show comparable currents in these regions. This can be explained by higher pseudocapacitance of IrO_x support (Fig. S6) but smaller Pt particle size in the Pt/IrO₂ material. On the other hand, the two catalysts behave very differently in the hydrogen region. Similar CV profiles were also reported by Yao et al. for both mixed and deposited Pt on IrO_x. The authors suggested that the hydrogen adsorption/desorption on Pt is affected in some way by the amorphous support.¹⁷ We suggest that the observed absence of the typical Pt H_upd characteristics for IrO_x supported Pt can be related to the lower conductivity of the IrO_x support in the 0.05–0.4 V potential range. Indeed, studies on the electrochromic effect of IrO_x, showing that a transition from a semiconducting to metallic behavior at E > 0.5 V vs. RHE,^56,57 confirm this statement.

Unlike hydrogen, it seems like CO can adsorb on the surface of Pt even if it supported by IrO_x, as after the CO adsorption taking place in the discussed above low conductivity region of IrO_x (0.1 V vs. RHE), clear CO stripping signal was recorded. While of a very high importance, the full understanding of the difference between hydrogen and carbon monoxide adsorptions on Pt/IrO_x requires a more detailed study and is beyond the scope of the present work. For both Pt/IrO_x and Pt/IrO₂ catalysts one can note the existence of several CO stripping peaks. The existence of multiple peaks for CO oxidation on Pt nanoparticles is still not fully understood and is typically assigned to several factors such as particle size, particles agglomeration, CO adsorption on different Pt sites (terraces, edges, and corners), and even the nature of the used support.⁵⁸ It's worth noting that CO adsorbs only on Pt, as no CO oxidation peaks were observed for the IrO_x and IrO₂ supports (Fig. S7). The latter implies the absence of metallic Ir on the surface. This is also supported by the XPS data (Fig. 1), while XRD discards any significant amount of metallic Ir in bulk.

Adsorption of CO on Pt for both catalysts allows one to estimate electrochemical surface area, A_ECSA, of Pt. As summarized in Table I, A_ECSA increases in the order Pt/IrO₂ > Pt/IrO_x > Pt black, and reflects a decrease in the Pt size of the bifunctional materials, in agreement with the XRD results in Table S2. It is worth noting, however, that the A_ECSA values of the Pt/IrO₂ (14.0 m² g_Pt⁻¹) catalysts is smaller than those reported for Pt/C catalysts with similar particle sizes.⁵⁹ This difference can be explained by a higher degree of Pt nanoparticles agglomeration on IrO₂ (Figs. 1d and 1e) and even by some blockage of the Pt surface by the oxide support.⁶⁰

With respect to the catalysts activities, in line with the literature results for Ir-based OER catalysts, bifunctional IrO_x-supported catalysts were more active toward OER. The higher activity of both chemical and electrochemical grown amorphous Ir oxide catalysts,^22,47,61,62 is typically assigned to its chemical and electronic structures, like the presence of an Ir hydroxo (Ir-OH) layer,⁶³ and the combined existence of Ir³⁺ sites and vacancies in the oxygen electronic structure.⁶⁴ Recently we have shown that the participation of lattice oxygen makes OER faster on such catalysts.³⁶ Comparing to literature, the OER activity of the Pt/IrO_x catalyst studied here is higher than those previously reported for others similar bifunctional materials in sulfuric acid electrolyte.^17,45 In terms of ORR activity, among the materials studied here, Pt/IrO₂ (35.1 mA mg_Pt⁻¹) catalyst exhibits the best gravimetric ORR kinetic currents, which can be attributed to the smaller Pt particle size. With respect to the specific activity values (j_SA), the current density obtained for the Pt/IrO₂ (87.1 μA cm_Pt⁻²) is close to that of Pt black (85.3 μA cm_Pt⁻²), and both are more active than the Pt/IrO_x catalyst (65.3 μA cm_Pt⁻²). Similar activity was also reported by Kong et al. for Pt/IrO₂ catalysts.^20,30 Tafel coefficients of 60 and 120 mV dec⁻¹ have been extensively reported for Pt-based electrodes and are explained in terms of the coverage degree of the electrode by adsorbed oxygen, following the conditions described by a Temkin or Langmuir isotherms at low and high overpotential values, respectively.⁶⁵ Similar Tafel slopes for the bifunctional catalysts and for Pt black implies that the ORR mechanism is not affected by the presence of the IrO_x and IrO₂. Thus, one can summarize that while Pt/IrO_x is more active for OER, Pt/IrO₂ performs better during ORR.

According to literature, potential cycling in the relatively mild ORR potential region during the long-term stability tests results in the degradation in active surface area and catalytic performance.^28–30 As shown in this work, degradation during the cycling in the 0.6–1.1 V potential window can be attributed to dissolution of both Ir and Pt. Indeed, the onsets of measurable anodic dissolution for Pt and electrochemically grown hydrous Ir oxide in H₂SO₄ are at ca. 0.95–1.1 V,⁶⁶ 0.7–0.8 V,⁵⁴ respectively, and thus within the potential window range of the ORR degradation protocol. Despite the lower onset, dissolution of Ir is significantly lower, which is explained by a higher stability of IrO_x and IrO₂ toward the cathodic dissolution. Comparing the two studied materials, Pt and Ir dissolution rates for both Pt/IrO_x and Pt/IrO₂ are similar, although it seems like the latter is slightly more stable. In contrast to these results, Jovanovič et al. showed that dissolution of Ir from IrO₂ is two order of magnitude lower compared to metallic Ir or electrochemically grown hydrous Ir oxide when cycled.³⁷ This discrepancy may be related to the existence of the less stable IrO_x species in our Pt/IrO₂ catalyst. Based on these results, it can be summarized that while staying in the ORR potential region it is the stability of Pt which is of the highest concern. Although significantly lower, dissolution of Ir, however, should not be completely disregarded.

The dissolution profiles are very different when catalysts are investigated in the OER potential region. Thus, concerning the Pt dissolution, there is a substantial difference between the first and the subsequent cycles. Also, while for the first cycle most of Pt dissolves anodically, subsequent dissolution is predominantly cathodic. This behavior of Pt is attributed to Pt passivation by formation of stable oxides.²⁴ Small cathodic dissolution peaks reveal that already at 1.1 V some dissolution takes place, which is in line with our recent analysis undertaken for polycrystalline Pt in which the onset of Pt cathodic dissolution was found at 1.25 V vs. RHE.⁵⁵ In that study, the anodic Pt dissolution on the partially reduced Pt surface was assumed to explain this dissolution process. Regarding the Ir dissolution, the anodic dissolution peaks with dissolution onsets similar to those of OER are observed. These results are in accord with previous Ir dissolution results stating that beyond the OER onset the Ir dissolution increases, and that the oxygen evolution and Ir dissolution share common intermediates.^35,54,67 In terms of dissolved amounts, a reduction in the Ir dissolution peaks of the Pt/IrO_x material in the second and third cycles of Fig. 3b, should also be noted, which can be a result of the partial blockage of Ir active sites by oxygen bubbles, and/or to conversion of IrO_x to IrO₂.⁶⁸ The lower dissolution stability of IrO_x compared to IrO₂ in the OER region is in accordance with literature,^22,37 and is consistent with other works stating that while amorphous Ir oxide is more active for the OER, the crystalline IrO₂ has higher electrochemical stability.^22,23,47,69 Employing the recently introduced stability number concept,³⁶ it is calculated that both material have a similar S-number of ca. 10⁴. The calculated S-number for the IrO₂ support is 2–3 orders of magnitude lower than that reported by Geiger et al., reflecting a relatively high content of Ir(III) species in the IrO₂ prepared in this work. The latter is a consequence of the relatively low calcination temperature.^22,23 Comparing Pt and Ir dissolution, one can see that the Pt and Ir dissolution peaks have similar intensities for the Pt/IrO₂ material in the OER potential range, while for the Pt/IrO_x catalyst the Ir dissolution is more pronounced than that of Pt.

Significantly higher dissolution of both Pt and Ir is observed when potential protocol involves both ORR and OER. In this protocol, the anodic and cathodic dissolution processes for Pt and Ir are easily resolved, confirming the transient dissolution of both metals.³² A significant increase in Pt dissolution rates during transitions from low to high potentials for commercial Pt/C catalyst was previously reported by Pizzutilo et al.⁷⁰ Besides dissolution, also other degradation processes such as particles coalescence, growth, and detachment, as identified by IL-TEM measurements, accelerate in this potential window.²⁸ As already pointed in the analysis of Figure 3b, Ir dissolution for the Pt/IrO_x catalyst is higher than that for the Pt/IrO₂. Nonetheless, dissolution of Pt is of the highest concern here. The observed higher Pt dissolution for the Pt/IrO₂ catalyst may reflect some effect in the nanoparticles size difference,³⁴ although other effects like catalysts' porosity, and the catalytic layer thickness can have an important role in the dissolution process.^33,71 Regarding the specific dissolution values (normalized by electroactive surface area), the Pt dissolution rate of ∼5.5 ng cm⁻² cycle⁻¹ was found for the Pt/IrO_x and Pt/IrO₂ materials, while the specific Ir dissolution of ∼0.09 ng cm⁻² cycle⁻¹ and ∼0.14 ng cm⁻² cycle⁻¹ were found for Pt/IrO_x and Pt/IrO₂, respectively. These results are in line with those reported for polycrystalline Pt,³² Pt/C,³³ metallic Ir,⁶⁷ and electrochemically grown hydrous Ir oxide films.⁵⁴ Therefore, one can conclude that no specific bifunctional stabilization effect can be derived for the Pt/IrO_x and Pt/IrO₂ catalysts.

The catalytic activity results attest that a superior ORR performance is obtained with the IrO₂ supported Pt material. Regarding the OER, higher currents are achieved while using the Pt/IrO_x material. The adequate catalyst should thus be chosen considering if the fuel cell or electrolyzer mode of the URFC is a priority.

The calculated dissolved Pt amounts for the Pt/IrO_x and Pt/IrO₂ bifunctional catalysts in the 0.6–1.6 V range suggest that this could become a significant problem while using these catalysts in the oxygen electrode of URFC. Moreover, since in such devices the catalysts would be submitted to even higher anodic potentials (1.5–1.7 V and higher for smaller catalyst loadings) during the electrolyzer mode operation, with subsequent potential decrease to (0.5–0.7 V) in the fuel cell operation, even higher dissolution rates are expected. Assuming a constant dissolution rate per cycle, the complete dissolution of Pt would be reached after ∼1,700 cycles and ∼1,500 cycles for the Pt/IrO_x and Pt/IrO₂ catalysts, respectively. Thus, if the operation mode of a URFC is switched twice a day, after 10,000 hours of operation about half of the Pt loading in these catalysts would be dissolved (steady dissolution is not considered here). Regarding the Ir dissolution, the complete degradation of the IrO_x and IrO₂ supports would take place after ∼4,000 and ∼9,000 cycles between 0.6–1.6 V, respectively. It should be noted, however, that unlike dissolution of Pt which seems to be similar in aqueous and fuel cell environments,²⁴ Ir is significantly more stable in proton exchange membrane water electrolyzers.³⁶ Moreover, by employing a higher calcination temperature, even higher stability toward Ir dissolution can be easily achieved.

This new finding put the whole concept of using Pt and Ir as a bifunctional oxygen catalyst under question. Since the main issue lays in the intrinsic instability of Pt toward transient dissolution, new strategies of replacing Pt with a more stable catalyst have to be suggested. While current work is not aimed to present such solutions, it clearly shows that dissolution stability must be always considered during the development of new materials. Moreover, it points on the bottleneck of such catalysts stability, i.e. dissolution of Pt, while stability of Ir may be considered sufficient.