The fabrication of the single-layer (n = 1) cylindrical SRO (C-SRO) is schematically shown in Fig. 1a. An SRO layer was firstly deposited on (001)-oriented STO substrate as a conductive connecting layer (see Supplementary Fig. S1), followed by an SAO as the sacrifice layer. Then SRO/BTO/SRO was grown on SAO, here upper and lower SRO layers are asymmetric in thickness, 16 and 2 nm respectively, for producing an artificial strain distribution. As the water-dissolvable nature of SAO, SRO/BTO/SRO heterostructure and the rigid SRO will be separated once the sample is submerged into the deionized water. In the dissolution, the freestanding SRO/BTO/SRO gradually crackles and coils into the cylindrical sharp C-SRO due to the strain, as shown in Fig. 1b and c. After a complete solution of SAO, the C-SROs adhere to the rigid SRO and connect eventually with an aid of the exposed surface of rigid SRO.
We performed the X-ray diffraction measurements on rigid SRO/STO, and the SRO/BTO/SRO/SAO/SRO/STO heterostructure as shown in Fig. 1d. The rigid SRO layer is strictly epitaxial along with the (001) orientation. After inserting the SAO layer, the orientation of the epitaxial SRO/BTO/SRO/SAO/SRO/STO remains without any noticeable peaks of impurity structural phases. In Fig. 1e, the rigid SRO on STO exhibits a ferromagnetic to paramagnetic transition Tc at ~ 170 K by the temperature-dependent magnetic measurement, also across Tc, the transport, resistance versus temperature, of SRO shows a Fermi-liquid behavior to bad-metal transition (Fig. S1c). These physical properties of our SRO are consistent with previous reports38,39.
In Fig. 2a, we have shown the optical images of C-SRO layers (from n = 1, 2, 3, 5 and 10) on rigid SRO, and it is clear that the density of C-SRO increases with n, exhibiting a 3D stack SRO. After the water-dissolution, these samples were annealed at 400 oC for 1 hour to make the SRO connection a strong adhesion instead of a weak one (such as Van der Waals force) for the stability in OER measurements as shown in Fig. S2. In Fig. 2b, the diameter (d) of the C-SRO as a function of n is presented, the d slightly enlarges from ~ 16.8 to ~ 18.6 µm while n = 2 to 10. Before that the first C-SRO layer has a small d ~ 13.7 µm. This change may come from the difference of supporting layers for C-SROs with n = 1 and n > 1. In principle, one may get any number of C-SRO layers by stacking SRO/BTO/SRO/SAO heterostructures in the epitaxy process. Before the OER activity tests, we examined the electrochemical double-layer capacitances (Cdl) by cyclic voltammetry (CV) measurements in the Non-Faradaic current range at different scan rates (experimental details can be found in Fig. S3) to realize the C-SRO coverage effort from the view of ECSA. In Fig. 2c, we have found that Cdl as a function of n exhibits a linear dependence benefiting from the stability of C-SRO, not only the shape, and diameter but also the connection. In addition, this result shows that our stack of C-SRO is an effective method to increase the surface area of single-crystalline TMO.
Utilizing a standard three-electrode configuration, the OER catalytic activities of our samples were evaluated in 1 mol/L KOH solution at room temperature. In Fig. 3a, linear sweep voltammetry (LSV) measurements of rigid SRO and with C-SRO (from n = 1 to 10) coverage are shown. The current density (j) of 0.35 mA/cm2 at 2 V of rigid SRO is found. By stacking C-SRO on it, the j increases with n monotonously from n = 1 to 10 at a certain voltage. Further in Fig. 3b, the Tafel slopes, potential versus log(j) are plotted, where the Tafel slope reduces from 102 to 30 mV/dec with n (0 to 10), indicating better reaction kinetics in the progress via C-SRO. Then the electrochemical impedance spectroscopy (EIS) Nyquist were measured at a frequency from 105 to 0.01 Hz by 5 mV amplitude AC signal as shown in Fig. 3c, as can be seen, the charge transfer resistance between a working electrode and electrolyte decreases with n, confirming that the coverage of C-SROs provides an enhanced charge transfer kinetics. In Fig. 3d-f, we have summarized the onsetpotentials, overpotentials, and Tafel slopes as a function of C-SRO coverage to present an overall improvement of OER performance. The onsetpotential of rigid SRO is ~ 1.900 V, which is slightly larger than the previous reports of SRO thin films40,41. This is likely arisen by a faint difference of stoichiometry, it drops gradually with n and reaches the minimum of 1.233 V. Meanwhile the overpotential exhibits a similar regulation, by three-layer of C-SRO coverage, the overpotential can be ~ 93% reduced from 0.92 to 0.06 V at j = 5 mA/cm2. Throughout these overall performances, perhaps most interestingly, there is no saturation trend is seen up to n = 10. It is not far to seek that these OER activity enhancements are originated from the enlargement of active surface area in part. In Fig. 3g, the electric field dependences of CV curves of rigid SRO and C-SRO n = 3 and 10 with different scan rates are shown, detailed CV curves can be found in Fig.S3. The j increases with scan rate and reaches 204, 112, 54, and 27 µA/cm2 under 80, 40, 20, and 10 mV/s in C-SRO (n = 10), which is roughly 18.5 times bigger than the one of rigid SRO. These results show that the C-SRO on rigid SRO is indeed a feasible strategy of the ECSA increase for an enhanced OER activity.
The forming of the cylindrical shape SRO comes from the BTO interlayer via the support effect, which enables us to achieve the stack of single-crystalline SROs. First of all, BTO is ferroelectric with a polarization ~ 30 µC/cm2 42, the carrier of SRO will be depleted or accumulated at the BTO/SRO interface. According to the dielectric constant, and the carrier density of SRO, the screening length of SRO is ~ 1.6 Å which is smaller than the half unit cell43,44. So in such a scenario, the bound charge effect can neglect. Secondly, the BTO interlayer provides a lattice mismatch between two thickness asymmetric SROs for curling. Then the non-flat SRO may also contribute to the enhanced OER by extra electronic reconstruction, such as the appearance of a high-spin state, due to the strain effect. To clarify this, we compared the overpotential at 1 mA as a function of the SRO surface area in Fig. 4a. Here the surface area of C-SRO is evaluated by the value in the rigid heterostructure, say, before the water-dissolution. The overpotential decreases with the surface area in both rigid and C-SRO covered SRO, however, in the case of C-SRO, the overpotential is lower than epitaxial rigid SRO at a certain surface area, suggesting that this phenomenon is given by the strain effect, this difference is marked by grey area. In Fig. 4b, schematically the outer and inner SROs in C-SRO endure tensile and compressive strain respectively, the bending strain ε in the bending direction is given by ε = t/R, where t and R are half of the thickness of SRO/BTO/SRO and the corresponding radius of curvature33. Roughly the outer and inner SROs bear ± 0.72% strain, in addition to an altering of the dxy, dxz, and dyz occupancy by the modification of RuO6 octahedra, the possible emerging high spin-state would contribute to a benefit of an enhanced OER activity. In Fig. 4c, we examine the magnetic field dependent magnetization of rigid SRO and C-SRO (n = 3 and 10). Based on the ideal stoichiometry, theoretical saturated magnetization (MS) of Ru is two with the electron configuration of t2g(3↑, 1↓), usually, the MS of bulk SRO is always from ~ 1.1 to ~ 1.6 µB per Ru owing to the electron delocalization associated with itinerancy27,45,46. In our rigid SRO, the MS equals ~ 1.61 µB per Ru under a parallel magnetic field, however, in contrast, the MS of C-SRO (n = 3 and 10) exhibits a value of ~ 2.24 and 2.42 µB per Ru which indicates that the high-spin state occurs in our C-SRO. Then the occupancy of extra eg orbital provides an additional charge transfer path which could boost a further enhancement of OER activity. This conclusion is also testified by the X-ray photoelectron spectroscopy (XPS) measurement on the C-SRO and rigid SRO as shown in Fig. 4d. The binding energy of Ru 3d3/2 shifts from ~ 285.9 eV of rigid SRO to ~ 285.4 eV of C-SRO. This indicates that the adsorption energy between C-SRO and oxygen intermediates can be lowered benefiting the OER performance40. Meanwhile, it implies that the electron occupancy of C-SRO is partially driven to a high-spin state33,47.
By achieving the enlarged ECSA and the high-spin state in stacked single-crystalline C-SRO, the enhanced OER performance is found, the overpotential of C-SRO (n = 10) is ~ 60 mV at j = 5 mA/cm2, which is one order of magnitude smaller than the one of promising OER candidate TMO thin films, such as LaNiO312, LaCoO310, Ba0.5Sr0.5Co0.8Fe0.2O3−δ48, and IrO249. Besides, the Tafel slope of our C-SRO (n = 10) can be triggered to ~ 30 mV/dec, which is almost reduced by half comparing to theirs.