Suppression of Oxygen Vacancies in Rutile Ruo2 via In Situ Exsolution for Enhanced Water Electrocatalysis

Elemental vacancies are proposed as an effective approach to tuning the electronic structure of catalysts that are critical for energy conversion. However, for reactions such as the sluggish oxygen evolution reaction, the excess of oxygen vacancies (VO) is inevitable and detrimental to catalysts’ electrochemical stability and activities, e.g., in the most active RuO2. While significant work is carried out to hinder the formation of VO, the development of a fast and efficient strategy is limited. Herein, a protection SrO layer produced successfully at the surface of RuO2 with the in situ exsolution method with perovskite SrRuO3 as the precatalyst, which could significantly hinder the generation of VO. Benefited from the suppression of VO, the surface‐modified RuO2 requires a low overpotential of 290 mV at 100 mA cm−2, accompanied by remarkably high electrochemical stability (100 h) and Faraday efficiency (≈100%). Theoretical investigation reveals that the formation energy of VO in RuO2 is almost doubled in the exsolved RuO2 phase as a result of the weakened RuO bond covalency. This work not only provides insight into the structural evolution of perovskite oxide catalysts but also demonstrates the feasibility of controlling vacancy formation via in situ exsolution.


Introduction
Oxygen vacancies, although generally invisible with ultralow concentrations, play crucial roles in determining the physical and chemical properties of transition metal oxides. [1][2][3] They efficient OER catalysts. [6,14] Doping is a straightforward and efficient method by incorporating guest atoms into the host lattice. Alkali element doping can tune the electronic structure (e.g., Ni, Li, Mn, and Na) and coordination environment (e.g., Pt) of Ru sites, [12,[15][16][17][18] and thus enhance the lattice stability. Transition metal doping was demonstrated to be effective in enlarging the gap between the Fermi level and the O 2p-band center, which could enhance the energy barrier of lattice oxygen overoxidation of RuO 2 and consequently suppress the formation of V O . [19] The vacancy self-healing strategy is also proven to be effective in maintaining both surface and bulk crystal structures. [20,21] Nevertheless, a more simple and effective method is still highly desired for the designing of OER catalysts with both activity and durability.
In situ exsolution has been widely observed during the electrocatalysis process, [22,23] especially for perovskite oxides (ABO 3 ) with highly tunable compositions and electronic structures. [24] By carefully controlling the ABO 3 catalyst stoichiometry, operating potentials, and concentration of electrolytes, one can get either the metallic, the oxides, or their mixtures at the surface. [25][26][27] With this, the catalytic properties (such as activity, selectivity, reaction sites, types of products, and amounts of catalytic sites) can be enhanced significantly because of the optimized catalytic kinetics and structural stabilities. [28][29][30] Because the exsolution process is combined with the concomitant local nonstoichiometry, a question arises as to whether we can suppress V O upon surface reconstruction and protection.
In this work, by using SrRuO 3 perovskite oxide as a precatalyst, we observed the in situ exsolution of RuO 2 nanoparticles, which are covered by the SrO layer. The surface-modified RuO 2 exhibited remarkable catalytic activity with small overpotentials of 209 and 290 mV to reach 10 and 100 mA cm −2 in 1 m KOH, respectively. The catalyst can also survive for a rather long time of 100 h without no significant performance degradation under 100 mA cm −2 . X-ray photoelectron spectroscopy (XPS) patterns of Ru element demonstrated that RuO 2 maintains the electronic structure without overoxidation during long-term testing. Density functional theory calculation suggested that the coating of SrO can increase the formation energy of V O , which is beneficial to the activity and stability of the RuO 2 catalyst. Our work provides a simple and flexible strategy for the design and synthesis of high-performance OER electrocatalysts by in situ surface self-reconstruction.

Theoretical Investigation of V O Suppression
At the beginning of this work, we investigate the feasibility of in situ exsolution on suppressing V O . It has been widely accepted that the structure collapse of bulk antiferromagnetic (AFM) RuO 2 is a result of the participation of lattice oxygen during the harsh OER process. [6,17] The increased covalency of the metaloxygen bonds as a result of the strong Ru 4d-O 2p hybridization leads to the high activity of the lattice oxygen. [31] This is the root cause for the fast production of V O and structure collapse through the lattice oxygen oxidation mechanism (Figure 1a). [32,33] Since surface reconstruction is inevitable for perovskite oxide catalysts (ABO 3 ), we plan to suppress the formation of Vo through the in situ exsolved protecting layer (Figure 1b,c). [26] Metal oxide substrate (RuO 2 ) with a thin layer of leaching oxidized A position metal (SrO) can be obtained under certain conditions. [26,34] Two models were constructed to explore the difference in V O formation, namely, pure AFM RuO 2 (110), and SrO/RuO 2 (Figures S1 and S2, Supporting Information). Both the (100) and (111) surfaces of SrO are investigated because they are the most energy-favorable surfaces for the cubic system. Iso-surface plotting of the charge-density difference ( Figure 1d and Figure S3a, Supporting Information) and plane-averaged charge-density difference along the z-direction (Figure 1e and Figure S3b, Supporting Information) suggest strong charge depletion around the Sr atoms and the charge accumulation at the Ru and O atoms in RuO 2 . [27] The charge redistribution and the increased filling of O 2p orbital could upshift the Fermi level and weaken the p-d hybridization. This will decrease the covalency of the metal-oxygen bonds (RuO), which is beneficial for the suppression of lattice oxygen activities. As strong evidence, the bonding distance of RuO is increased significantly from 1.93 Å for RuO 2 to 2.02 and 2.07 Å for the SrO (100)/ RuO 2 and SrO (111)/RuO 2 structure (Figure 1f). It is reasonable to conjecture that the overoxidation and V O formation in RuO 2 could be thermodynamically hindered during OER. [19] Given that, we calculated the formation energy of V O (E VO ) in AFM RuO 2 (110) and AFM SrO/RuO 2 slab. [35] This value is increased from 2.18 eV (RuO 2 ) to 3.08 and 4.09 eV for SrO (100) and SrO (111) covered RuO 2 , respectively ( Figure 1f and Figure S4 and Table S1, Supporting Information), which undoubtedly confirmed the successful suppression V O formation.

Synthesis and Characterization of the Precatalyst
SrRuO 3 is chosen to construct the predicted catalyst structure. To highlight the difference between the fresh RuO 2 and in situ exsolved RuO 2 , a small portion of RuO 2 in the precatalyst is necessary. We began with the synthesis of the precatalyst via hydrothermal reaction, followed by sintering at high temperatures (more details are seen in the experimental section). X-ray diffraction (XRD) confirms the existence of two phases, which can be indexed to SrRuO 3 with orthorhombic structure (space group Pnma) and RuO 2 with tetragonal structure (space group P4 2 /mnm) (Figure 2a). Rietveld XRD data using Full-Prof software indicates that the mass fraction of SrRuO 3 and RuO 2 are 65% and 35%, respectively. The scanning electron  The information on the Ru site coordination is the key to understanding the chemical properties. XPS was then conducted on the precatalyst (Figure 2f and Figure S9, Supporting Information). The Ru 3p 3/2 spectra of SrRuO 3 /RuO 2 can be deconvoluted into two kinds of single peaks. The peak with a lower binding energy of 462.4 eV is attributed to the Ru 4+ specie, while the higher binding energy situated at 465.4 eV corresponds to the satellite (Figure 2f). [19] It is worth mentioning that RuO 2 and SrRuO 3 have the similar structure with Ru inside the oxygen octahedral cage, thus we will not distinguish them because of the small difference. [36,37]

OER Catalytic Performance
Now we begin to investigate the OER behaviors of the catalyst. As revealed in the linear sweep voltammetry (LSV) curves (Figure 3a), the overpotential required to reach the current density of 10 and 100 mA cm −2 are 207 and 290 mV, respectively, for SrO/ RuO 2 catalyst. The values are much smaller than that of the same geometry electrode made of commercial RuO 2 (480 mV to reach 10 mA cm −2 and 650 mV to reach 100 mA cm −2 ) and RuO 2 tested by previous work (360 mV to reach 10 mA cm −2 and 500 mV to reach 100 mA cm −2 ), [38] indicating the enhanced OER catalytic activity for SrO/RuO 2 . The values are also smaller than the SrRuO 3 /RuO 2 precatalyst (220 mV to reach 10 mA cm −2 and 350 mV to reach 100 mA cm −2 ), suggesting an enhancement in activity during in situ exsolution process. These values are also competitive to those reported state-of-the-art Ru-based OER catalysts either at 10 or a higher current density of 100 mA cm −2 (Figure 3b and Table S2, Supporting Information). [10,11,15,[39][40][41][42][43][44] The Tafel slope of SrO/RuO 2 was determined to be 68 mV dec −1 and is lower than pure RuO 2 (133 mV dec −1 ), suggesting the faster reaction kinetics of SrO/RuO 2 (Figure 3c). [45] More importantly, the significant decrease in the tafel slope also implies a change in the rate-determining step of OER reaction. For commercial RuO 2 catalyst, a Tafel slope of over 120 mV dec −1 suggests that the OER may be controlled by the first electron transfer step. While for SrO/RuO 2 catalyst, the Tafel slope of ≈60 mV dec −1 suggests that the OER reaction may be controlled by the chemical desorption subsequent to the first electron transfer step. A multistep chronoamperometry measured under various overpotential without iR correction is shown in Figure 3d. Upon increasing overpotential, the current density responds simultaneously and approaches a relatively stable state rapidly even when the current density reaches ≈1400 mA cm −2 , suggesting the distinguished mass transfer properties and Adv. Mater. Interfaces 2023, 10, 2300279   Figure 3. a) LSV curves of SrRuO 3 /RuO 2 precatalyst, commercial RuO 2 , Cu wire with silver paint, and the LSV curve of RuO 2 from ref. [38]. b) Comparison of the overpotentials to reach the current density of 10 and 100 mA cm −2 of our catalyst with recently reported state-of-the-art Ru-based catalysts. c) Tafel slopes of SrRuO 3 /RuO 2 precatalyst, commercial RuO 2 and RuO 2 from ref. [38]. d) Multicurrent process without iR correction. e) Stability test and f) Faradaic efficiency performance at 100 mA cm −2 .
mechanical stability under a broad current density region. [46] The stability test was conducted by chronopotentiometry under 100 mA cm −2 (Figure 3e). As seen in the inset of Figure 3e, the exsolution is a slow process which sustained for ≈7 h. SrO/RuO 2 catalyst maintains the overpotential with a negligible increase after 100 h continuous operation, suggesting outstanding robustness and electrochemical stability. [47] The high stability can be further confirmed by cyclic voltammetry analysis which displays a near rectangular-shaped window devoid of any redox peaks within a potential range of +1.06 to 1.36 V (vs reversible hydrogen electrode (RHE)), showing exquisite reversibility of Ru active sites ( Figure S10, Supporting Information). [48] The Faraday efficiency (FE) measured at 100 mA cm −2 is determined to be 98% by collecting produced gases (Figure 3f). [49] Adv. Mater. Interfaces 2023, 10, 2300279   Figure 4. a) TEM bright field image of precatalyst after OER stability test. b) Recorded SAED patterns from area 1 (inset Figure 4a) and c) the adjacent particles. d) TEM image of a typical rod-like structure, which can be indexed to RuO 2 phase with (110) surface exposed. e) EDS elemental maps for Sr, Ru overlay. f) Elemental mapping of Ru element. g) XPS spectra of Ru 3p for SrO/RuO 2 and commercial RuO 2 catalysts. h,i) HRTEM images show the in situ growth of a RuO 2 particle at the intersection of two big particles under electron beam illumination for 5 s.

Structure Analysis of SrO/RuO 2
TEM was conducted to shed light on the morphology and composition changes of the SrRuO 3 /RuO 2 precatalyst after electrochemical testing. Rod-shaped particles can be observed in the selected area as shown in Figure 4a and Figures S11-S14 in the Supporting Information. The SAED pattern for particle 1 in Figure 4a can be assigned to (011), (110), and (101) planes of tetragonal RuO 2 (the Joint Committee on Powder Diffraction Standards (JCPDS) # 00-040-1290) (Figure 4b), [50] while the electron diffraction rings recorded from the adjacent particles can be assigned to cubic SrO phase (JCPDS # 00-006-0520) (Figure 4c). [51] TEM image on a selected rod clearly indicates the existence of a thin SrO layer on the {110} planes of RuO 2 (Figure 4d). The configuration of RuO 2 coated by SrO is also verified by the EDS elemental maps (Figure 4e,f and Figure S12, Supporting Information). Sr distributes uniformly in the whole catalyst, while Ru only distributes in the rod-shaped area.
Similar distributes of Sr and Ru were also detected in other regions (Figures S13 and S14, Supporting Information). We thus proposed that RuO 2 nanorod was obtained after the exsolution of SrRuO 3 during the OER process, leaving a very thin layer of SrO on the surface (The main phase of SrO can be dissolved in the electrolyte). Now, the catalyst can be definitely characterized as SrO/RuO 2 , rather than the SrRuO 3 /RuO 2 precatalysts.
After the determination of catalyst structures, we need to answer the question of whether V O is suppressed for the exsolved RuO 2 catalyst. For this purpose, XPS patterns of both SrO/RuO 2 catalyst and conventional RuO 2 catalyst were recorded and compared (Figure 4g and Figures S15-S19, Supporting Information). As expected, the Ru 3p 3/2 binding energy of Ru 4+ specie in RuO 2 after OER testing was 0.3 eV higher than that in SrO/RuO 2 , suggesting a higher oxidation state for RuO 2 (Figure 4g). While no significant changes for the Ru coordination environments was observed in SrO/RuO 2 catalyst, judging from the comparison of Ru 3p binding energy with the fresh RuO 2 catalyst. We can conclude that the Ru element maintains its chemical states without overoxidation during OER. The highresolution transmission electron microscopy (HRTEM) analysis provides more information on the high structural stability of the RuO 2 phase. Figure 4h shows the HRTEM images of RuO 2 [50] on the bottom and the cubic SrO phase on the top-right part of the picture. [51] Interestingly, we occasionally observed the in situ exsolution process during the electron beam illumination. A crystallite was newly formed between RuO 2 and SrO phases as exhibited in Figure 4i. The lattice can be ascribed to the (200), (101), and (101) planes of RuO 2 according to the fast Fourier transform results ( Figure S20, Supporting Information). The staggered atomic layer caused by multifold twins are clear in this newly formed tetragonal RuO 2 particle. [18] All these results are consistent with our theoretical and experimental predictions.

Origin of High OER Performance
In the end, we tried to understand the origin of the high OER performance of the in situ exsolved SrO/RuO 2 . The equivalent circuit of the Nyquist plots is shown in the inset of Figure S21 in the Supporting Information. The diameter of the semicircle at high frequency is almost potential-independent, suggesting a fast electron transfer process. In contrast, the significant change in the low-frequency region with increasing overpotential reveals the rate-determining step nature of the ion migration corresponding to potential. [52] Moreover, the electrochemical impedance spectroscopy (EIS) spectra of SrRuO 3 /RuO 2 precatalysts, SrO/RuO 2 after OER and commercial RuO 2 catalyst were compared as presented in Figure 5a. Compared to SrRuO 3 /RuO 2 precatalyst and commercial RuO 2 , the decrease in semicircular diameter for SrO/RuO 2 could be attributed to the fast electron transfer by SrO covering. [18] Bode phase plots also revealed that the time constant for the adsorption process decreased from 0.3 s for the pristine catalyst to only 0.2 s for the phase transitioned one (Figure 5b).
To give further insight into the enhancement of the OER performance, we investigated the chemical states and electronic structure of the adsorbed oxygen species at the surface of SrO/ RuO 2 catalyst by XPS (Figure 5c and Figures S22 and S23, Supporting Information). The O 1s spectra can be divided into three oxygen contributions (Figure 5c), which are the lattice oxygen (O 1 ), surface-absorbed oxygen (O 2 2− /O − ) and/or hydroxyl groups (OH) (O 2 ), [53] and adsorbed molecular water (O 3 ), located at the binding energy around 529.0, 530.6, and 533.0 eV, respectively. [54] It has been proposed that the high coverage of adsorbed hydroxyls species can accelerate the formation of OO bond in OOH, which is the rate-determining step for OER. [55] A high O 2 /O 1 (θ) ratio generally corresponds to a higher OER activity. [56,57] As listed in Table S3 in the Supporting Information, the value of θ in commercial RuO 2 catalyst is reduced from 1.73 to 1.31 after OER testing, suggesting a decrease in OER activity. In contrast, for the SrO/RuO 2 catalyst, the O 2 species is still dominant after long-term testing with the θ ratio as high as 3.65. We thus proposed that the suppression of V O helps shift down p-band center away Fermi level, which leads to not only an overall stable structure but also elevated intrinsic OER activities (Figure 5d). [58][59][60]

Conclusion
In summary, we successfully suppressed the formation of V O in rutile RuO 2 through an in situ exsolution strategy. With the existence of a thin protecting SrO layer, the V O formation energy is nearly doubled. As a result, SrO/RuO 2 exhibits both high activities and stabilities toward OER in alkaline conditions and is even better than most recently reported Ru-based electrocatalysts. The experimental observation confirms the in situ exsolution process could provide a new route to devise and synthesize the OER catalyst with outstanding performance.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.