IrNi/C morphology and crystallinity: synthesis temperature
IrNi nanoparticles supported on carbon (IrNi/C) were synthesized using a modified impregnation method that instantly changed to reductive gas conditions at target heat-treatment temperatures, in order to control the crystallinity while maintaining a small particle size. Synthesized IrNi/C materials treated at 400 ℃ and 1000 ℃ were denoted as IrNi/C-LT and IrNi/C-HT, respectively. For comparison, Ir/C-LT and Ir/C-HT were prepared by the same method without the nickel precursor. The high-resolution transmission electron microscopy (HR-TEM) images of IrNi/C-LT and -HT (Figs. 1a and 1d) show a similar particle size (1–1.5 nm) and distribution. Energy-dispersive X-ray (EDX) mapping images of IrNi/C catalysts (Supplementary Fig. 1) reveal that Ir and Ni are homogeneously distributed on the carbon support in atomic proximity suggesting the presence of an Ir and Ni alloy. Based on the TEM EDX images (Supplementary Fig. 2), the atomic ratios of Ir:Ni were 78:21 and 66:34 for IrNi/C-LT and IrNi/C-HT, respectively. However, as shown in the high-angle annular dark field (HADDF) image (Figs. 1b, 1c, 1e, and 1f), IrNi/C-LT showed a polycrystalline and amorphous structure with low crystallinity, whereas IrNi/C-HT displayed an almost single crystal character with high crystallinity. IrNi/C-HT nanoparticles exhibited (111) and (200) facets, suggesting [011] FCC single-crystal structure. The above crystallinity differences were confirmed by X-ray diffraction (XRD) (Supplementary Fig. 3). The (111) reflection of IrNi/C-HT was located at 41.1º, upshifted relative to that of metallic Ir at 40.6º (see JCPDS no. 87–0715), suggesting a disordered Ir-Ni alloy. In contrast, an Ir alloy peak was not observed for IrNi/C-LT due to its small particle size and low crystallinity. Sharp XRD reflections as well as HR-TEM images confirmed the presence of few large Ni particles (Supplementary Fig. 4). A similar relation between temperature and crystallinity at comparable particle size was observed for the pur Ir/C catalysts (Supplementary Figs. 5 and 6). Based on these results, we concluded that a control the crystallinity of IrNi/C catalysts via annealing temperature at comparable 1 ~ 2 nm particle size and alloy atomic ratio was achieved. The rational reason of a small particle size with high crystallinity as described above is as follows. First, the dry product of carbon support with Ir and Ni precursor containing solution leads that Ir and Ni ions attach to the carbon support in atomic-scale fine dispersion. These are hardly reduced in N2 without H2 gas even at high temperatures. This prevents Ostwald ripening/agglomeration during thermal treatments of up to 1000 ℃. Second, when the N2 gas was immediately changed to 10% H2 (99.999%) and 90% N2 at 1000 ℃, reduction occurred quickly, leading higher nucleation density and a finer nuclei size. This effect can be inferred from conventional nucleation theory 24. When the temperature is raised with an H2 atmosphere, the reduction proceeds slowly due to a low temperature. This leads to Ostwald ripening acceleration, resulting that the particle size is not finer, as shown in Supplementary Fig. 7. In that state, only the IrNi/C-HT sample undergoes a sufficient crystallization step to form single-crystal nanoparticles with high crystallinity.
Electrochemical properties and catalytic activity
To investigate the effect of crystallinity on the electrochemical properties, the cyclic voltammetry (CV) of the prepared IrNi/C-LT and IrNi/C-HT was measured under changing upper potential limits (UPL), (Figures 2a and 2b). As the UPL increased during the CV of IrNi/C-LT, the hydrogen adsorption-desorption (Hupd) peak associated with a metallic Ir surface disappeared and two redox peaks associated with Ir(III)/Ir(IV) and Ir(IV)/Ir(>IV) emerged 25. When the UPL was decreased, the CV shape of IrNi/C-LT was maintained. This indicated that the surface of the IrNi/C-LT irreversibly changed from a metallic-IrNi alloy to IrNiOx. In contrast, the Hupd peak area of IrNi/C-HT decreased with increasing UPL but restored again as the UPL decreased. These findings suggested that the surface of IrNi/C-HT reversibly converted between a metallic character and an oxidic IrNiOx character. Again, a similar behavior was obvious for pure Ir/C (Supplementary Figure 8).
Figure 2c demonstrate the electrocatalytic performance of the catalysts during a set of subsequent linear voltammetric scans probing first the HER, then the OER, and then again the HER reactivity, all in 0.05 M H2SO4. Prior to the OER test, the catalysts were activated by 50 CV cycles to generate an IrOx layer. Compared with Ir/C catalysts, IrNi/C catalysts demonstrated higher HER catalytic activity, comparable with commercial Pt/C. Ir/C-LT demonstrated a more enhanced OER performance compared to Ir/C-HT, due to the large amount of redox active Ir centers represented by the redox charge (Figures 2a and 2b). The Ir-Ni alloys generally shows a superior catalytic activity for the OER, regardless of their crystallinity. To evaluate metal-oxide reversibility, HER performances were re-measured after the OER test. While the IrNi/C-HT catalyst maintained high HER reactivity, that of the IrNi/C-LT catalyst decreased. These distinctly different levels of reversibility are reflected by the HER overpotentials (Figure 2d) and the Tafel plots (Supplementary Figure 9), as well. Interestingly, the analogous behavioral trends were observed in the HOR/OER reversibility (Figures 2e and 2f): Initial HOR performance of IrNi/C catalysts is comparable with the commercial Pt/C catalyst. After 10 CV cycles, the HOR catalytic activity of IrNi/C-LT dropped markedly, but that of IrNi/C-HT decreased only slightly. These results evidence that Ir-based nano-catalysts with high crystallinity show a remarkable reversibility in their surface electrochemistry and their associated electrocatalytic reactivity for oxidation and reduction reactions, such as HER/OER and HOR/OER.
To explain the reversible character of the IrNi/C-HT, we tracked the electronic structure and oxide thickness of IrNi@IrNiOx core-shell nanoparticles under intermittent OER and HER operating conditions using in situ/operando X-ray absorption near-edge structure (XANES), which was measured at the Ir L3-edge to probe electron transitions from 2p to 5d. Since XANES is a bulk sensitivity technique, it is difficult to show the oxidation state of local structures. The white line of core-shell Ir nanoparticles represents the overall d-hole character, which is the redox state of both the oxidized Ir in the shell and that of the metallic Ir in the core. If the core of the catalyst state is fixed to metallic-Ir, the white-line position of the core-shell catalyst roughly represents the d-band hole count of Ir in the shell (Figure 3a) 26. Moreover, if the particle size is the same, the oxide thickness of the core-shell can be roughly estimated by the white-line area (Figure 3b).
The white-line peak positions in the experimental sequence are illustrated in Figure 3a. Under the OER condition, the white-line position of both IrNi/C catalysts is positively shifted with a d-band hole count of 5.4 compared with the Ir/C catalyst (d-band hole count of 5.2), as shown in Figure 3c. The d-band hole count was calculated by white-line position and formal d-band hole count function (Supplementary Figure 10). The large number of vacancies in IrNiOx/C enhances the electrophilic character of oxygen, resulting in smaller kinetic barriers for OER 26-28. These XANES results can explain the high activity of IrNi/C catalysts for OER. After OER and OER-HER operations, IrNi/C-LT has a higher white-line position than IrNi/C-HT, indicating a higher oxidation state of IrNi/C-LT after reaction. However, under HER condition, the white-line position of IrNi/C-HT is higher than that of IrNi/C-LT (Figure 3d). To explain this phenomenon, we estimated the chemical state of the catalyst under HER conditions. The HER Tafel slope of IrNi/C-HT (43 mV dec–1) indicates that the Heyrovsky step in the Volmer-Heyrovsky mechanism is the rate determining step for IrNi/C-HT 29. Adsorbed hydrogen (Hads) coverage is expected to be 0.25–0.50 for the Heyrovsky step 30. Based on previous research, Hads increases the d-band hole count 31,32. Thus, Hads coverage of IrNi/C-HT would slightly increases the white-line position during the HER. For IrNi/C-LT, it is assumed that the IrOx surface adsorbs protons at a cathodic potential and converts to an Ir(OxHy) species which possesses a lower d-band hole count than IrOx 33,34.
To observe the oxide thickness under OER/HER operating conditions, white-line areas were identified by in situ/operando XANES. In all conditions, IrNi/C-HT catalysts showed smaller white-line areas than IrNi/C-LT catalysts, implying that IrNi/C-HT possesses a thinner IrOx layer than IrNi/C-LT during OER and HER (Figure 3b). In particular, the IrNi/C-HT catalyst showed a highly reduced white-line area under the HER condition (Figure 3d), which is close to the metallic-Ir foil. Thus, IrNi/C-HT nanoparticles possess an almost metallic surface under the HER condition, leading to high HER activity. The white-line area clearly exhibits the difference between IrNi/C-HT and IrNi/C-LT under OER and HER condition. It suggests that oxide thickness is a key controlling factor of the reversibility phenomenon of IrNi@IrNiOx core-shell nanoparticles with high crystallinity.
To study the effect of the crystallinity on the surface electronic structure in more detail, we analyzed a depth-resolved Ir 4f XPS spectra using two different levels of kinetic energy (KE). The associated inelastic mean free path associated with KEs of 210 eV and 550 eV are almost 0.5 and 0.9 nm, respectively, which represent the atomic shell layers of 2 and 3–4 Ir atoms, respectively 35. The Ir 4f spectrum can be deconvoluted into the three types of Ir species: Ir0 (metallic-Ir), IrⅣ (rutile-type IrⅣOⅡ–2), and IrⅢ (amorphous IrⅢOⅠ3) 19. The details of the deconvoluted XPS results are shown in Supplementary Table 1. The Ir depth profile of pristine IrNi/C electrodes indicates that the majority of nanoparticles were in the metallic state with soft oxidation of the surface (Supplementary Figure 13). After OER, the spectrum of the IrNi/C-LT catalyst at 210 eV KE is fully shifted to the position of the 100% IrⅢ species (Figure 4a). The Ir 4f line shape of the IrNi/C-LT catalyst at 550 eV KE is a mixture 53.2% IrⅢ and 46.8% IrⅣ with no metallic species, suggesting that the surface of IrNi/C-LT is completely changed to a thick amorphous IrNiOx layer. This thick amorphous IrNiOx shell of IrNi/C-LT is conserved after the HER condition, revealing an irreversible Ir oxide layer on IrNi/C-LT. Meanwhile, the composition of the IrNi/C-HT catalyst after OER is 69.6% IrⅢ and 30.4% IrⅣ at 210 eV KE and 39.4% IrⅢ, 26.8% IrⅣ, and 39.4% Ir0 at 550 eV KE (Figure 4b), indicating that an extremely thin IrNiOx layer has been synthesized on the metallic subsurface. After HER, the ratio of IrⅢ species at 210 eV KE is significantly decreased and 38.2% metallic-Ir is observed, demonstrating that some of the thin IrNiOx layer was turned into the metallic surface. The Ir depth profile of IrNi/C-LT and IrNi/C-HT summarized in Figure 4c, clearly demonstrates that the thick IrNiOx layer of IrNi/C-LT synthesized by OER is maintained after HER and that the thin IrNiOx layer of IrNi/C-HT converts to metallic-Ir after HER.
The description of phenomena related to crystallinity and HER/HOR/OER reversibility are illustrated in Figure 4d. IrNi/C-LT having low crystallinity exhibits an abundant grain boundary, which accelerates the penetration of ions into nanoparticles and results in the formation of a thick IrNiOx layer 36. Ex-situ HAADF and EDS mapping images shows that IrNi/C-LT (Supplementary Figure 14) consists of a thick amorphous IrNiOx. This thick irreversible IrNiOx demonstrates excellent performance for OER but poor performance for HER and HOR. On the other hand, IrNi/C-HT with high crystallinity has no grain boundary defects, enabling formation of the very thin IrNiOx layer. Ni of IrNi/C-HT still located with the Ir, confirming the core of the metallic structure (Supplementary Figure 14). This thin IrNiOx of IrNi/C-HT is reversibly converted to a metallic structure under HER/HOR conditions, showing high reversibility for OER/HER/HOR.
Mechanism study of reversible catalytic activity
To clarify the reversible reactivity character of Ir-Ni/HT further, we correlated it to surface dissolution processes at the atomic scale and were able to uncover the underlying mechanism that leads to reversibility. To achieve this, we conducted in situ/operando ICP technique using an electrochemical flow cell 37. It is well known that electrochemically prepared IrOx is irreversibly oxidized 38-41. Hence, there is a need for a mechanism study on the cause of reversible conversion of the IrNiOx layer on IrNi/C-HT to a metallic surface. When cycling between 0.05 and 1.5 VRHE to produce IrNiOx species on nanoparticles, IrNi/C-LT has a large peak compared to IrNi/C-HT, indicating that IrNi/C-LT possesses a large amount of amorphous IrNiOx. Subsequently, three types of peaks were observed.
First, the peaks of OER intermediate dissolution located at the highest potential, are associated with an intermediate Ir species during OER and lattice oxygen participation (denoted as O1 ~ O3). For the peak area of dissolution during OER, IrNi/C-HT is lower than that of IrNi/C-LT owing to the thin IrNiOx layer of IrNi/C-HT. To confirm stability of IrNi/C-HT during reaction, the dissolution profile under OER and HER conditions was conducted and is shown in Supplementary Figure 16. IrNi/C-HT under OER and HER conditions showed a lower Ir dissolution as compared to IrNi/C-LT, despite similar OER catalytic activity (Figure 2c), demonstrating high stability of IrNi/C-HT. Second, the peaks of reversible reduction dissolution (denoted as R1 and R2) which is located at close to 1.0 VRHE, close to the standard potential of Ir oxidation reaction (0.926 V + 0.0591 pH), exhibited electrochemical dissolution of reversible Ir oxides. Third, the peaks of irreversible reduction dissolution (denoted as I1, I2 and I3) located at the lowest potential, representing cathodic dissolution due to irreversible Ir oxide. Both peaks indicate the cathodic dissolution of Ir oxide. Interestingly, there is no peak of reversible reduction dissolution for IrNi/C-LT, and IrNi/C-HT after 1.3 V. When the upper potential was increased from 1.3 to 1.5 VRHE, a peak of reversible reduction dissolution was observed, indicating the production of reversible Ir oxides during the OER 40. For the IrNi/C-HT, cathodic dissolution is significantly lower than that of IrNi/C-LT, indicating that metallic surface of IrNi/C-HT is not derived by removal of IrNiOx layer through dissolution.
Mayrhofer et al. analyzed the reduction of Ir hydroxide using the in situ/operando ICP technique and suggested that cathodic dissolution can be explained by dissolution and re-deposition phenomenon and/or incomplete oxide reduction 38,39. Based on this hypothesis, we proposed a description of the reversible oxide film of IrNi/C-HT, as shown in Figure 5c. The particularly small amount of cathodic dissolution of IrNi/C-HT suggests that converted IrNiOx almost transformed to the metallic phase, and did not dissolve into the electrolyte. IrNi/C-HT has thin IrNiOx layer, resulting in a metallic surface after dissolution of IrNiOx. The adsorbed hydrogen (Hads) on the metallic surface can serve to promote Ir reduction to metal 42-44. Based on the Tafel slope, the Hads coverage expected to be 0.25 ~ 0.50 during HER is sufficient for metal reductions. This metallic surface of IrNi/C-HT served as a substrate for deposition of dissolved Ir ion to accelerate the reduction of Ir ions to metallic-Ir, which supports the reversible property of IrNi/C-HT. For IrNi/C-LT, the thick oxide layer is retained despite cathodic dissolution. The IrNi/C-LT surface may have an IrOxHy phase under HER condition, leading to re-deposition of dissolved Ir ions and non-complete reduction of the oxide. Markovic et al. reported that dissolution/re-deposition phenomena can be measured by the amount of dissolution with different sweep rate 45. Fast scan rates prevent dissolved ion from diffusion layer to bulk electrolyte, leading to a lower amount of dissolved ion. Thus, a high dissolution amount ratio of slow/fast scan rate exhibits enhanced dissolution/re-deposition property. The dissolution amount ratio of slow/fast scan rate of IrNi/C-HT between 0 and 1.5 VRHE is 0.310, which is higher than 0.233 for IrNi/C-LT. This result indicates that IrNi/C-HT has a dominant dissolution/re-deposition mechanism due to the metallic subsurface acting as a substrate for reduction.
Harnessing catalytic reversibility in a large-scale single PEM fuel cell.
To apply and harness the catalytic reversibility of Ir-Ni alloy nanoparticles with respect to the HOR and OER in a real electrochemical device, fuel starvation experiments were conducted in a single PEM fuel cell that was built using Ir-Ni/C-HT and Ir-Ni/C-LT as the catalytically active component in the anode catalyst layer (Figure 6a). The initial fuel cell performance of IrNi/C-LT and -HT is lower than that of the commercial Pt/C catalyst due to lower HOR activity and metal composition (Supplementary Figure 17). For the fuel starvation experiment, supplied anode gas was converted from H2 to Ar at a current density of 100 mA cm‒2. When H2 was exhausted at the anode, the anode potential of the commercial Pt/C catalyst increased to 1.8 VRHE for OER instead of HOR, leading to a reverse potential phenomenon. After 20 s, the anode potential once more increased to 2.4 VRHE for carbon oxidation reaction (COR), as shown in Figure 6a and Supplementary Figure 18 9. This reaction conversion was confirmed by in-situ exhaust gas analysis (Supplementary Figure 19). In contrast, the cell voltage of both IrNi/C cells increased and maintained at 1.4–1.5 V for OER, and no CO2 gas was observed by in situ exhaust gas analysis, exhibiting that high OER activity of IrNi/C catalysts prohibits corrosion of carbon support. To observe HOR catalytic activity after fuel starvation, the fuel cell performance and impedance before and after the fuel starvation test were measured and is shown in Figure 6b, Supplementary Figure 17, Supplementary Figure 20, and Supplementary Table 2. The performance degradation at 0.6 V of commercial TKK Pt/C catalysts was 37.2% due to the carbon support corrosion. IrNi/C-LT reduced fuel cell performance by 26.9% despite no carbon corrosion, indicating low reversibility for HOR and OER. Conversely, the performance of IrNi/C-HT decreased slightly to 9.2%, suggesting that the reversible catalytic activity of IrNi/C-HT works well under real fuel starvation condition.
To confirm the reversibility of IrNi/C-LT and -HT for HER and OER under real device conditions, reverse voltage experiments were performed in a water electrolyzer, which was to change the cell polarity every 20 min with a constant current density of 40 mA cm–2 (Figure 6c). As shown in Figure 6d, both catalysts were applied with the same cell voltage of 1.65 V to obtain 40 mA cm–2 for the first 20 min. The cell voltage of IrNi/C-LT gradually increased whenever the polarity of the electrode was changed, but in the case of IrNi/C-HT, the cell voltage remained stable even after 10 polarity conversions. These results demonstrate that IrNi/C-HT retains bifunctional catalytic activity with reversibility to HER and OER under real device conditions. Our approach promotes the reversibility of nanocatalysts that enable a variety of electrochemical reactions, which can be used as catalysts to resist the reverse voltage in fuel cells and water electrolysis systems.