Introduction

The pursuit of highly active and cost-effective catalysts is of prime significance for sustainable energy conversion and storage in order to develop renewable energy production. Implementing these emerging clean energy solutions, such as water splitting, direct solar and metal air batteries et al., highly relies on a variety of electrocatalytic reactions, such as oxygen evolution reaction (OER). Given that OER is in general impeded by its intrinsically sluggish kinetics due to a multistep four electron process, and consequently requires a considerable overpotential relative to its thermodynamic potential of 1.23 V vs. reversible hydrogen electrode (RHE). To address this, precious metal-based materials, e.g. RuO2 and IrO2, are currently employed as the efficient state-of-art OER catalysts1. However, high cost, low abundance as well as poor durability during long term operation badly hinder their widespread application.

Over these years, a tremendous number of alternatives based on non-precious metals materials have been intensively explored, including oxides (hydroxides)2,3,4,5, carbon (carbon based hybrid)6,7, metal oxides/carbon composite catalysts8,9,10, chalcogenides11,12 and phosphates13,14. Among these candidates, perovskite oxides have drawn much attention as competitive electrocatalysts due to their flexible states of transition metals, fascinating oxygen ion mobility and exchange kinetics, and unique tailorable properties15,16,17. For example, Suntivich et al.18,19 reported that Ba0.5Sr0.5Co0.8Fe0.2O3 (BSCF) perovskite provided a higher intrinsic OER activity in alkaline solution comparable to commercial IrO2 electrocatalyst, A site of Ba/Sr displays the fastest oxygen transport kinetics, while the Co-rich compositions on the B site offers faster oxygen exchange kinetic. The rational design of B-site transition metal ratio could tune the eg orbital filling (σ*-orbital occupation), and thus benefited OER process18.

Although BSCF is known to have superior intrinsic activity, large particle sizes rendered from traditional bottom-up method18 limit their large-scale practical application due to low gravimetric mass activity. The poor stability caused by surface amorphization of BSCF particles under OER conditions is another tough issue for industrial application20. It is known that the surface area could be increased through a nanostructure strategy or surface modification, while the stability of perovskites could be improved by intrinsic methods resort such as chemical substitution or electronic configuration modification. Jung et al.19 introduced a nano-sized Lax(Ba0.5Sr0.5)1−xCo0.8Fe0.2O3 with superior activity and stability, where the electronic states was managed by chemical substitution of A-site cations with lanthanum and particle growth was controlled by sintered temperature. Ln(Ba0.5Sr0.5)0.5Co0.8Fe0.2O3 (Ln = Nd, Sm, Gd)21 and Pr0.5(Ba0.5Sr0.5)0.5Co0.8Fe0.2O322 have also been developed as highly active catalysts base on A-site chemical substitution strategy.

In this work, we highlight a feasibility of Bi introduced into A site of BSCF lattice (Bi0.1(Ba0.5Sr0.5)0.9Co0.8Fe0.2O3−δ, denoted as Bi0.1BSCF) to modify the electronic structure of perovskites. As known, bismuth oxide is usually used as an oxygen-ion conducting electrolyte in solid oxide fuel cells. Bi could also be introduced to A site of BSCF perovskite and tailor oxygen electrocatalysis ability23. Furthermore, surface modification is another effective strategy to improve the activity and durability of electrocatalysts. It is reported that the cold plasma process is an efficient approach for surface modification and functionalization, which can generate roughed surface, surface vacancies, defects and other active functional groups24,25,26,27,28. Therefore, for the first time, we propose the pollution-free and facile plasma modification on perovskite oxides to tailor the near-surface structure. The optimized 5% H2/Ar plasma-treated Bi0.1BSCF catalyst, which denoted as P-Bi0.1BSCF, demonstrates a higher intrinsic activity and durability towards OER, relative to the well-known BSCF perovskite as well as commercial IrO2 catalyst. An electrochemically active amorphous layer created by plasma engraving is achieved, and its effect is furthermore elaborated for OER.

Methods

Catalysts synthesis

Ba0.5Sr0.5Co0.8Fe0.2O3 and Bi0.1(Ba0.5Sr0.5)0.9Co0.8Fe0.2O3−δ (denoted as BSCF and Bi0.1BSCF, respectively) perovskite oxides were synthesized by using a EDTA-citrate sol-gel process17. Briefly, the starting materials of Ba(NO3)2, Sr(NO3)2·4H2O, Bi(NO3)3·6H2O, Fe(NO3)3·9H2O, and Co(NO3)2·6H2O (Sinopharm Chemical Reagent Co., Ltd.) were mixed in deionized water in accordance with their stoichiometric amounts, employing EDTA and citric acid as chelating agents. With the introduction of aqueous ammonium hydroxide solution (NH3, 28%, Sinopharm Chemical Reagent Co., Ltd.) into above solution, the pH value was tuned at ~7. Such solution was stirred and then heated at 100 °C to yield a gel, then calcination at 250 °C overnight to form a solid precursor. The solid precursor of BSCF was sintered in air at 1000 °C, whereas Bi0.1BSCF was calcined in air at 850 °C for 4 h to obtain the powders. The commercial IrO2 (99.5%, Aladdin Industrial Corporation) catalysts was studied for comparison.

The freshly-prepared Bi0.1BSCF catalyst was subsequently conducted via a plasma cleaner (MING HENG, PDC-MG) with a gas flow of air, Ar or 5% H2/Ar for the plasma ignition (commercial 13.56 MHz RF source). Different irradiation time (0 min, 2 min, 3 min, 5 min and 8 min) with powers of ~130 W and pressure of 90 Pa was applied during plasma process. The optimized 5% H2/Ar plasma-treated Bi0.1BSCF for 3 min was used for detailed studies, which denoted as P-Bi0.1BSCF.

Materials characterization

Phase structures of the as-prepared catalyst powders were determined by XRD on Bruker (D8 Focus, Cu Kα radiation). Program FullProf was employed for the diffraction refinement. SEM images were performed on a SU-8010 SEM, whereas high resolution TEM images equipped with EDS were conducted on a Tecnai G2 F20 U-TWIN TEM. XPS measurements of the catalysts were carried out on a Kratos Axis Ultra DLD instrument. The obtained XPS spectra were calibrated by referencing C 1 s to 284.6 eV, and simulated using the XPSPEAK41 software. We analyzed the specific surface areas by Brunauer Emmet Teller (BET) system with N2 as the adsorptive medium. Approximately 2.0 g samples were weighed and degassed at 250 °C for 4 h before nitrogen physisorption at the temperature of liquid nitrogen (77 K). The ability of perovskites to adsorb OH− after exposure to water (100% humidity) at room temperature for 2 h was estimated from the Fourier transform infrared spectra (FTIR, Nicolet iS50, Thermo Scientific America). Oxygen temperature programmed desorption (O2-TPD) measurement was surveyed to analysis the oxygen desorption properties.

Electrochemical evaluation

The as-prepared BSCF, Bi0.1BSCF, P-Bi0.1BSCF and commercial IrO2 (~5 μm, Sigma Aldrich 99.9%) catalysts were, respectively, ground in mortar to disperse well. The mixture, including 40 mg perovskite catalyst, 8 mg Ketjen black (KB, EC-600JD), 5 mL ethanol and 0.25 mL 5 wt.% Nafion solution were ultrasonicating to obtain a homogeneous ink. A 2 μL catalyst ink was then deposited onto the polished glassy carbon electrode surface with a uniform loading of ~0.2832 mg cmdisk−2 (~0.2266 mgcatalyst cmdisk−2). Electrodes containing IrO2 were also prepared with a similar loading of 0.2587 mg cmdisk−2. The OER electrochemical characteristics were conducted in N2-saturated 0.1 M KOH electrolyte with a standard three electrode cell configuration (Pine Research Instrumentation). 1 M Hg/Hg2Cl2 electrode and platinum sheet were applied as the reference and the counter electrode, respectively. The potentials in this work were iR corrected, where the value of R is the ohmic electrolyte resistance via high frequency AC impedance. Cyclic voltammetry (CV) curves for OER were recorded on the rotating disk electrode from 1.3 to 2.0 V vs. RHE at 1600 rpm, with the scan rate of 10 mV s−1.

Results and Discussion

Fig.1a,b show the Rietveld refined powder XRD pattern of pristine Ba0.5Sr0.5Co0.8Fe0.2O3 and Bi0.1(Ba0.5Sr0.5)0.9Co0.8Fe0.2O3−δ (denoted as BSCF and Bi0.1BSCF, respectively). Upon Rietveld refinements, the fitted lattice parameters of BSCF and Bi0.1BSCF are summarized in Table S1 (ESI). One can see that both BSCF and Bi0.1BSCF present a cubic perovskite structure with the space group of Pm-3m. Schematic representation of A-site doping perovskite structure of Bi0.1BSCF is described in Figure S1 (ESI). The phase structure of Bi0.1BSCF in this study agrees well with the reported literature23. The typical diffraction peak (110) of Bi0.1BSCF moves to a higher angle compared to pristine BSCF, indicating that the introduction of Bi into BSCF lattice causes lattice contraction. This might be attributed to the substitution of the larger Sr2+ (0.158 nm) and Ba2+ (0.175 nm) ions by smaller Bi3+ (0.124 nm). Additionally, upon XRD analysis (Fig. 1c), no obvious structural change of P-Bi0.1BSCF is observed after plasma treatment for 3 min using 5% H2/Ar as the generating gas.

Figure 1
figure 1

(a) Refined diffraction patterns of BSCF powder (b) Refined diffraction patterns of Bi0.1BSCF, (c) Diffraction patterns of P-Bi0.1BSCF.

Full size image

Shown in Fig. 2 is the SEM morphology of the as-prepared BSCF, Bi0.1BSCF, and P-Bi0.1BSCF catalysts. It is clearly seen that the particle size of Bi0.1BSCF is 0.5~1 um (Fig. 2c,d) and is smaller than that of initial BSCF (Fig. 2a,b), which is probably due to its much lower sintering temperature. Accordingly, Bi0.1BSCF has a larger surface area of 3.0771 m2 g−1, which is 4.3 times larger than that of BSCF (0.7132 m2 g−1) as estimated via BET method (Fig. S2, ESI). Although the SEM morphology of P-Bi0.1BSCF does not change obviously after 5% H2/Ar plasma engraving (Fig. 2e,f), the BET surface area of P-Bi0.1BSCF increase by 14% compared to the freshly prepared Bi0.1BSCF. This demonstrates that the plasma treatment could engrave the surface of Bi0.1BSCF perovskite and thereby increase the specific surface area.

Figure 2
figure 2

SEM images of (a,b) BSCF, (c,d) Bi0.1BSCF, and (e,f) P-Bi0.1BSCF.

Full size image

To further investigate the crystalline structural change of BSCF after Bi-doping and plasma treatment, HR-TEM are performed on BSCF, Bi0.1BSCF and P-Bi0.1BSCF catalysts, as shown in Fig. 3. The cubic perovskite structure of BSCF and Bi0.1BSCF is also confirmed by Fourier transformed pattern. The lattice diffraction fringes in Fig. 3b are 0.397 and 0.176 nm, which are well indexed to the (100)BSCF and (021)BSCF interplanar spacing, respectively. In the case of Bi0.1BSCF, the lattice diffraction fringes in Fig. 3d is 0.395 nm, which agrees well with the (100) interplanar spacing. One can see that the surfaces of BSCF and Bi0.1BSCF particles are highly crystalline with little amorphous region. After exposure in 5% H2/Ar-plasma for 3 min, the perovskite structure of P-Bi0.1BSCF is still observed in Fig. 3f (d (110) spacing of 0.281 nm). Together with XRD characterization of plasma-engraved Bi0.1BSCF, the results reveal that no crystalline changes after plasma treatment. In addition, STEM and the corresponding EDS mapping analysis suggests that the elements of P-Bi0.1BSCF are uniformly distributed without surface segregation phenomenon after plasma engraving process. Moreover, it is apparent that P-Bi0.1BSCF displays a unique core-shell structure, a disordered amorphous shell with a thickness of approximately 8–10 nm is generated after plasma modification. Similar phenomenon is reported by Li and his co-workers that the TiO2 nanosheets shows a crystalline TiO2 core/amorphous TiO2−δ shell structure via NH3-plasma surface modification with oxygen deficient on shell region29.

Figure 3
figure 3

Bright-field TEM, High-resolution TEM image and corresponding fast-Fourier transformed pattern of (a,b) BSCF; (c,d) Bi0.1BSCF; (e,f) P-Bi0.1BSCF; (g) EDS mapping of P-Bi0.1BSCF.

Full size image

Figure 4a presents the CV curves of pristine BSCF, Bi0.1BSCF and P-Bi0.1BSCF catalysts under the OER potential window. Similar measurements are performed on the IrO2 and KB for comparison. The catalytic activity contribution of KB is subtracted according to the composition of electrode. The performance of commercial IrO2 catalyst is comparable to that reported elsewhere30,31. The CV of Bi0.1BSCF shows a comparably lower onset potential (1.49V) and a higher current density than the IrO2 and BSCF catalyst, indicating a better OER catalytic activity of Bi0.1BSCF. After plasma engraving, P-Bi0.1BSCF catalyst manifests an apparent enhancement in OER activity with a significantly improved current density. The optimized processing parameters for plasma engraving is 5% H2/Ar working atmosphere only for 3 min (Fig. S3, ESI). Moreover, P-Bi0.1BSCF catalyst exhibits a quite small overpotential (η) at a current density of 10 mA cmdisk−2 of 370 mV, which is superior to that of Bi0.1BSCF (411 mV), BSCF (525 mV), and IrO2 (464 mV). Notably, the overpotential of P-Bi0.1BSCF favorably outperforms the reported perovskite-based OER catalysts21,32,33,34,35,36,37,38,39, e.g. BaCo0.7Fe0.2Sn0.1O3−δ (410 mV)36, SrNb0.1Co0.7Fe0.2O3−δ (SNCF) (500 mV)39 summarized in Table S2 (ESI).

Figure 4
figure 4

(a) Cyclic voltamograms of the BSCF, Bi0.1BSCF, P-Bi0.1BSCF, IrO2, and KB in 0.1 M KOH at 1600 rpm. (b) Tafel plots of these catalysts. (c) OER mass activity (MA) and specific activity (SA) at an overpotential of 0.4 V. (d) Degradation of electrochemical performance of as-prepared catalysts, inset: LSV curves of P-Bi0.1BSCF at 2nd, 200th, 400th and 600th cycles.

Full size image

Corresponding Tafel plots of the investigated samples are compared in Fig. 4b to evaluate the kinetics performance of OER. The Tafel slopes are 68, 74, 87, and 78 mV dec−1 for P-Bi0.1BSCF, Bi0.1BSCF, BSCF, and IrO2, respectively. Furthermore, the mass activity (MA) and specific activity (SA) are illustrated to evaluate the intrinsic activity, as shown in Fig. 4c. For instance, at an overpotential of 0.4 V, P-Bi0.1BSCF displays 8.9 times higher MA and 2.7 times higher SA relative to BSCF, respectively. The results confirm the positive contribution from the Bi-dopant and plasma engraving to the intrinsic activity. In addition,

The durability of electrocatalysts is another vital parameter for OER. In current study, we performed continuous CV curves for P-Bi0.1BSCF, Bi0.1BSCF, BSCF, and commercial IrO2 catalysts for 600 cycles, as shown in Fig. 4d. BSCF and IrO2 show a 25% and 27% reduction of its 2nd activity over 600 cycles, respectively, and the poor stability of BSCF results from the leaching of A site cations to the alkaline medium20. In contrast to BSCF, only 10% and 12% reduction is observed in the Bi0.1BSCF and P-Bi0.1BSCF catalysts under the same condition, respectively. The TEM image of post-OER P-Bi0.1BSCF in Fig. S4 shows that the structure of P-Bi0.1BSCF is maintained and the thickness of amorphous shell did not change much after continuous CV tests compared to as-synthesized P-Bi0.1BSCF. In addition, the XPS spectra before and after long-term OER testing shows that the oxidation state of Bi3+ does not change40,41 (Fig. S5), which also reveals the durability of P-Bi0.1BSCF catalyst.

To understand the source responsible for the excellent OER activity of P-Bi0.1BSCF, the surface state of BSCF, Bi0.1BSCF and P-Bi0.1BSCF catalysts is intensively studied by XPS analysis. The O1s XPS spectra for BSCF, Bi0.1BSCF and P-Bi0.1BSCF is deconvoluted to four characteristic peaks (Fig. 5a). The first one at a lower binding energy (529.6 eV) represents the lattice oxygen (O2−), followed by the surface oxidative oxygen O22−/O− (530.5 eV), hydroxyl groups OH− or adsorbed O2 (531.5 eV), and adsorbed molecular H2O of the oxide surface (532.9 eV)42. The relative content of the lattice and surface oxygen species derived is listed in Table S3 (ESI). The content of surface oxidative oxygen O22−/O− increases from 35.68% with BCSF to 42.17% with Bi0.1BSCF. Moreover, the O22−/O− concentration is further increased to ~49.23% by plasma engraving, indicating the existence of abundant active oxygen species on the amorphous region of P-Bi0.1BSCF oxide surface. Importantly, the electrochemically active O22−/O− are proved to be beneficial for catalyzing OER43,44. Also, the highly oxidative oxygen species on the surface are reported to be closely relevant to the oxygen vacancies45, and surface oxygen vacancies play a key role for perovskite oxides in catalyzing OER46. It reveals that the plasma engraving could create more electrochemical active sites on the amorphous shell. The similar phenomenon of surface amorphization and simultaneous increased activity for efficiently electro-catalyzing OER is also found in other catalysts47,48,49,50.

Figure 5
figure 5

XPS spectra of (a) O1s, (b) Co 2p & Ba3d, (c) Fe 2p for BSCF, Bi0.1BSCF and P-Bi0.1BSCF samples.

Full size image

The core-level spectra of Co 2p & Ba 3d for these three catalysts are presented in Fig. 5b. The overlapping between Co 2p and Ba 3d spectra makes it different to identify the surface Co valence by peak fitting, while the Co oxidation state could be inferred from the shift of peak position. Compared to BSCF, the positions of Co peaks for the Bi0.1BSCF shift to lower binding energies, suggesting that the surface valence of Co slightly decrease owing to high valence Bi-substitution. After 5% H2/Ar plasma engraved, the two main peaks of P-Bi0.1BSCF shift to higher binding energies obviously, which indicates the presence of surface cobalt in a higher oxidation state. It is reported that the high valence of Co could also play a positive role in the improvement of OER activity51. No distinctive variation was observed in the Fe state of catalyst surface (Fig. 5c). On the basis of the XPS results, the increased O22−/O− concentration together with electronic structure tuning of Co induced by plasma engraving might be responsible to the enhancement in OER performance.

The redox property of BSCF, Bi0.1BSCF, and P-Bi0.1BSCF is studied by O2-TPD, as shown in Fig. 6a. The desorption peak of the Bi0.1BSCF catalyst, associated with the reduction of Co iron, occur at approximately 251 °C, which is lower than that of BSCF (275 °C). It’s worth noting that, after 5% H2/Ar plasma engraved, the desorption temperature of P-Bi0.1BSCF catalyst reduce remarkably to nearly 198 °C, this indicates an excellent oxygen desorption capability35 and reflects a higher oxygen surface exchange ability for P-Bi0.1BSCF perovskite. The result is consistent with the EIS analysis that P-Bi0.1BSCF has better ionic and charge transfer abilities. Besides O2 desorption capability, the P-Bi0.1BSCF catalyst also shows good OH− adsorption. As observed by FTIR spectroscopy (Fig. 6b), a broad IR band centers at approximately 3466 cm−1 appeared, which corresponds to H-bonded OH− stretching vibration52. Obviously, Bi0.1BSCF exhibited a stronger OH− absorption ability than initial BSCF, and the OH− absorption ability is further improved via plasma treatment. According to the adsorbate evolution mechanism53, OH− absorption on the active sites of perovskites is the prerequisite for OER, and thereby the large OH− absorption can continually offer raw materials for the following OER process36. Due to the complex interplay between oxygen deficiency and surface-active redox centre of perovskite oxides, detailed mechanistic insights into the OER electrocatalysis of P-Bi0.1BSCF materials in alkaline media are currently limited. Nevertheless, with the superior activity, stability and cost-effectiveness, Bi0.1BSCF promises a novel precious-metal-free catalyst for the alkaline OER, and plasma engraving proved to be a facile and effective surface modification method to further improve the activity.

Figure 6
figure 6

(a) O2-TPD profiles of BSCF, Bi0.1BSCF and P-Bi0.1BSCF samples. (b) FTIR spectra of BSCF, Bi0.1BSCF and P-Bi0.1BSCF samples after exposure to 100% humidification.

Full size image

Conclusion

In summary, a plasma engraving strategy is, for the first time, proposed for the surface modification of perovskite materials towards electrocatalyzing OER. After 5% H2/Ar plasma engraving only for 3 min, an amorphous layer with abundant electrochemical active oxygen species is generated on Bi0.1BSCF perovskite surface. Moreover, the increased oxygen surface exchange ability and OH- absorption ability of P-Bi0.1BSCF are achieved. Owing to the unique properties via Bi-doping as well as plasma engraving, P-Bi0.1BSCF exhibits a fast kinetics process with a small Tafel slope of 68 mV dec−1 and high stability for OER, that is superior to most of the state-of-the-art perovskite-based electrocatalysts. The foregoing results open a new avenue for engineering perovskite oxides via plasma method for highly efficient OER, which is promising for a variety of electrochemical energy storage applications.