Synthesis and crystal structure characterization of BaRuO 3 . Four BaRuO3 polymorphs with distinct crystal structures are synthesized by elevated synthetic pressure as elaborated in the experimental section and schematically shown in Fig. 1a. The quality and purity of the four different polymorphs (9R-, 4H-, 6H- and 3C-BRO) are assessed by SPXD measurements (Fig. 1b-e). The diffraction peaks are well matched with those of the pure phases, confirming high fidelity synthesis and phase transitions of BaRuO318,19. Compared to conventional laboratory PXD, SPXD provides higher sensitivity and better resolution, thus improving the identification of crystalline structures. Detailed crystallographic data with selected atomic distances, bond angles, and calculated bond valence sums (BVS) are listed in Supplementary Tables S1 and S2. The obtained data further confirm that the as-prepared BaRuO3 is without trace impurities. At ambient pressure, 9R-BRO is the thermodynamically stable phase. It adopts a rhombohedral (R-3mH) structure with Ru3O12 trimers formed by three face-shared RuO6 octahedra. The Ru3O12 trimers are connected to each other by corner-sharing oxygen atoms (Fig. 1a, left). At an elevated synthesis pressure of 3 GPa, the number of corner-shared octahedra increases and results in a reconstructive phase transition to the hexagonal 4H-BRO (P63/mmc) structure (Fig. 1a, middle left). The basic building blocks are Ru2O9 dimers corner-sharing with neighbor Ru2O9 dimers. Further increasing the synthesis pressure to 5 GPa yields a second hexagonal structure 6H-BRO (P63/mmc) with the appearance of new corner-connected structures comprising Ru2O9 dimers with RuO6 octahedral layers (Fig. 1a, middle right). Finally, the cubic perovskite structure 3C-BRO (Pm-3m, Fig. 1a, right) containing only corner-shared RuO6 octahedra is synthesized at 18 GPa. Here we measured only key information for 3C-BRO because of its extremely harsh synthesis condition (18 GPa) and poor yield (~ 5 mg each batch). The as-prepared four samples have diverse crystal structures and comparable micron-size, providing a rich platform to study the correlation between structural feature and intrinsic catalytic activity.
The crystal structures of the polymorphs are further corroborated by high-resolution transition electron microscope (HRTEM, Figure S1), where lattice fringes match well with lattice spacings of the different polymorphs. Energy-dispersive x-ray (EDX) elemental-mapping data confirm the elemental composition and homogeneous distribution of the as-synthesized oxides (Figure S2). From scanning electron microscopy (SEM) images, comparable micro-meter-sized particles are observed in the as-synthesized oxides (Figure S3). Their bulk nature is further confirmed by a rather small surface area calculated from Brunauer-Emmett-Teller (BET) measurements (Figure S4). X-ray photoelectron spectroscopy (XPS) is carried out to evaluate the chemical composition and oxidation state of the oxides. As displayed in Figure S5, the Ba 3d and Ru 3p spectra in all samples are the same, indicating that only Ba2+ and Ru4+ ions exist in the as-synthesized oxides and the oxygen stoichiometry of the oxides is uniform. The similar valence states of 4H-, 6H-and 3C-BRO can also be seen in the results of BVS calculations (Supplementary Table S2). X-ray absorption near-edge spectroscopy (XANES, Figure S6) is employed to finally determine the valence state of Ru. The standard double-perovskite Ca2YRuO6, which has a 4d4-Ru5+ configuration20–22, possesses a double peak feature: (t2g-hole related) and (eg-hole related) (Figure S6). The XANES data for 9R-, 4H-, 6H- and 3C-BRO show a single peak, typical of the 4d4-Ru4+ configuration. To sum up, the pure phase and bulk nature of as-made BaRuO3 polymorphs provide an ideal platform to investigate the structure features of ruthenates and their correlated intrinsic activity on HER.
Assessment of alkaline HER performance. Alkaline HER activities of different BRO polymorphs are collected and compared in this study. In addition, micro-meter RuO2, nano-nickel (~ 20–100 nm) and nano-Pt/C catalysts used in commercial applications are chosen as benchmarks to assess the performance of the as-synthesized micro-sized BROs. Figure 2a shows the linear sweep voltammetry (LSV) polarization curves of the catalysts in the potential range of -0.3 to 0.1 V versus the reversible hydrogen electrode (RHE). Except for Pt/C, 9R-BRO outperforms all control samples in terms of overpotential η at 10 mA cm− 2. It only requires 51 mV for 9R-BRO to reach 10 mA cm− 2, which is substantially lower than that of 4H-BRO (89 mV), 6H-BRO (118 mV), 3C-BRO (143 mV), nano-nickel (362 mV), and RuO2 (69 mV), and strikingly close to that of Pt/C (42 mV). An important electrochemical kinetic parameter is the Tafel slope, with a smaller Tafel slope signaling faster kinetics23–25. As shown in Fig. 2b, the Tafel slopes are 30, 69, 79, 97 mV dec− 1 for 9R-, 4H-, 6H- and 3C-BRO, respectively, indicating that 9R-BRO possesses the best electron transport rate among the four polymorphs. More significant is the fact that the Tafel slope of 9R-BRO is even smaller than that of Pt/C (34 mV dec− 1). Overall, the overpotential @ 10 mA cm− 2 of 9R-BRO is among the most efficient of all ruthenium-based oxides for alkaline HER activities reported to date, even better than that of most nano-structured benchmark catalysts that are more expensive to prepare (Fig. 2c). Considering the hand-ground nature of the 9R-BRO sample and relatively low surface area deduced from BET calculations, the high HER activities must be ascribed to its extremely powerful intrinsic activity. In Figure S7 the activity normalized to BET surface area, it also can be seen that the activity per surface area of 9R-BRO is superior among the polymorphs, highlighting its superior intrinsic activity again.
To assess the commercial potential of 9R-BRO, specific mass activity (A/g) is a vital criterion. As shown in Fig. 2d, despite its low surface area, the specific mass activity of 9R-BRO is 180.3 A/g, which is 2 times higher than that of (commercial) micro-RuO2 (97.4 A/g) and 1.3 times higher than that of (commercial) nano-Pt/C (137.8 A/g). For the all-important specific price activity (A/$) indicator, 9R-BRO is 10 and 8 times better than that of micro-RuO2 and nano-Pt/C, respectively. Nickel-based electrodes are widely used in commercial AEL. We compare the performance of 9R-BRO with that of nano-nickel under identical conditions. The overpotential η at 10 mA cm− 2 is 51 mV (9R-BRO) versus 362 mV (nano-nickel) as shown in Fig. 2a. The Tafel slope is 30 mV dec− 1 (9R-BRO) versus 164 mV dec− 1 (nano-nickel) in Fig. 2b. The specific mass activity (A/g) is 180.3 (9R-BRO) versus 2.9 (nano-nickel) and the specific price activity (A/$) is 33.4 (9R-BRO) versus 0.7 (nano-nickel), a factor of 48 times better (Fig. 2d and Table S3). These figures indicate that 9R-BRO could replace Ni for commercial use in alkaline HER electrolysers. In terms of scalability, 9R-BRO is synthesized by solid-state reaction, a technique commonly used in the industrial production of LiFePO4 and LiCoO2 for cathode materials in lithium batteries. The synthesis of micron-sized oxides by this technique has a proven record of repeatability. As a demonstration, we have prepared a massive 500 g sample (hereafter denoted as 9R-BRO-500) by scaled-up synthesis. The massive sample displays identical alkaline HER activities as laboratory-sized samples (Fig. 3a). Another test necessary for the development of a commercial catalyst is its performance under large current densities typically reached in commercial systems. We plated the 9R-BRO powder on a nickel net (d ~ 5.5 cm; left inset in Fig. 3b) used as the cathode of a commercial AEL of 0.5 Nm3/h capacity. As displayed in Fig. 3b, we obtained an operating cell voltage of ~ 1.74 V at a current density of 4,000 A/m2, which outperforms figures published in the most advanced commercial AEL systems (the normally quoted cell voltages in most commercial systems are over 1.9 V). We thus expect that 9R-BRO can easily undergo further optimizations as a cathode catalyst in commercial AEL systems.
The long-term operational stability of catalysts is another critical aspect in practical applications. As shown in Fig. 3c, a slight fluctuation of current density is observed over 40 hours of operation at an overpotential of 50 mV. The durability of the catalyst is further confirmed by 1,000 continuous cyclic voltammetry (CV) analyses. The 9R-BRO sample, being the thermodynamically stable phase at ambient pressure, exhibits superior structural stability over the other polymorphs (Figure S8). Noteworthy, the AEL system with 9R-BRO also demonstrates considerable stability under industrial conditions in ~ 30 hours (Fig. 3d), emphasizing the excellent application potential of 9R-BRO in achieving cheap hydrogen. One can ascribe this to the more robust structure feature of face-shared RuO6 octahedral proportion for which 9R-BRO has the highest percentage. It is not easy to break apart two face-shared octahedra unless one breaks three nearest neighbor Ru-O bonds. To further assess the structural stability of 9R-BRO, we compare the PXD and XPS data before and after the 1,000 cycles of voltammetry operation. We also place the oxide in a harsh media (1 M KOH and 1 M HCl) for two weeks (Figure S9). We observe no apparent change in the XRD and XPS data pre- and post-treatments, indicating that the superior stability of 9R-BRO.
The correlation between crystal structure feature and alkaline HER activity. Bulk conductivity could be a critical criterion to affect catalytic performance. In Figure S10, the partial density of states (PDOS) is calculate, showing that all the as-made oxides are metallic26. Experimentally, the temperature-dependent electrical resistivities of 9R-, 4H-, and 6H-BRO are measured (Figure S11). All three polymorphs exhibit low resistivity at the same order of magnitude at room temperature. In addition, previous work has shown that 3C-BRO displays a similar resistivity behavior27. The metallic behavior of all four polymorphs is beneficial to electron transportation and thus the HER rate.
To explain the substantial activity discrepancy shown by different polymorphs, structural analysis of the four BaRuO3 polymorphs is counted on and shown in Figure S12, where the subscripts “F” and “C” denote face- and corner-shared atoms, respectively. Diverse crystal structures of BROs consist of distinct RuO6 connections. Specifically, the cubic perovskite structure of 3C-BRO contains only corner-shared RuO6 octahedra, and the other three polymorphs consist of different proportions of face/corner-shared RuO6 octahedra. It is worth noting that, as displayed in Figure S13b, the Ru-Ru bond distance in face-shared RuO6 octahedra of 9R (2.53 Å), 4H (2.55 Å) and 6H (2.57 Å) is smaller than the nearest neighbor Ru-Ru distance in Ru metal (2.65 Å). This is a sign of exceptionally strong intermetallic bond formation caused by direct Ru-d/Ru-d interaction in the ruthenates28–30. It is expected that different RuO6 connections can alter the electron exchange interaction mode between metal atoms19. The magnetic properties of oxides are good indicators of the interaction of the Ru 4d orbitals. The nearest-neighbor exchange interaction in face-shared octahedra is direct electron hopping across the 180° Ru-Ru bond, leading to antiferromagnetic (AFM) behavior. In corner-shared octahedra, electron hopping between Ru ions mediated by an O ion should produce ferromagnetic (FM) behavior19. As there is no obvious deviation between zero-field-cooled (ZFC) and field-cooled (FC) data of these oxides, in Fig. 4a, we only show the ZFC mode data. With the decrease in faced-shared and increase in corner-shared octahedral proportion in these polymorphs, the magnetic behaviors of BRO demonstrate systematic alteration from the antiferromagnetic behavior (9R-BRO) to ferromagnetic behavior (9R-BRO). The measured magnetic properties agree well with the expected electronic exchange interaction in the different polymorphs. Taken together, more face-shared oligomer gives shorter Ru-Ru bond with direct d-d interaction.
More DFT calculations are performed to further elucidate the effect of Ru-Ru interaction. First, we use the crystal orbital Hamiltonian population (COHP) results (Fig. 4b) to establish that in bulk crystals, the Ru-Ru bonding orbitals of face-shared octahedra reside in the range between − 5.8 to -7.8 eV (the pink range) below EF. Mapping the PDOS in real space at this energy window (see Fig. 4b) shows that the Ru-Ru bonding orbitals connect face-shared octahedra in typical 9R-BRO. Given the d-d interaction in the face-shared octahedra, the electron could hop back and forth between the face-shared Ru atoms with enhanced itinerant31–33, promoting electron transfer in the crystal. In addition to charge transportation, hydrogen release is another important step in determining HER efficiency after water dissociation. Water dissociation generates *OH that anchors on the surface Ru and releases H* to vicinal O* atoms. Through a Tafel or Heyrovsky step, the adsorbed H* then forms H2 molecules and leaves the surface. In this process, the adsorption-free energy of H* (ΔGH*) is a key descriptor, with a ΔGH* value close to zero indicating a superior HER activity34–36. On the BRO surface, there are two types of oxygen sites: OT or OB (see inset of Fig. 4c), where the OT can be found in both face- and corner-shared RuO6 octahedra, OB are present only in face-shared octahedra. At OT sites, the values of ΔGH* are too large to favor easy hydrogen evolution from OTH* species. On the other hand, ΔGH* at OB sites are small, favoring HER. As shown in Fig. 4c, ΔGH* for OBH* is small, being less than |0.15 eV| for the 9R-, 4H- and 6H- polymorphs. For 9R-BRO, its ΔGH* is close to the top of the volcano plot (optimal value), being − 0.01 eV. It is also noteworthy that 9R-BRO has the highest proportion of OB sites. The values of ΔGH* for OBH* sites suggest that HER activities should be highest for the 9R-polymorph, followed by 4H- and 6H-BRO, in the order found by the experiment. To sum up, the correlation between d-d interaction intensity and HER activity can be seen in Fig. 4d, where the overpotential of BRO polymorphs sharply decreases with increasing proportions of face-shared RuO6 octahedra in a sequence of 3C, 6H, 4H and 9R, as the face-sharing Ru-O linkage increases from 0% (in 3C-BRO) to 33% (in 6H-BRO), 50% (in 4H-BRO), and finally to 66% (in 9R-BRO). These results highlight the merits of 9R-BRO in delivering excellent catalytic reactivity in alkaline HER, consistent with experimental observations.