Perovskite Cathode Materials for Low-Temperature Solid Oxide Fuel Cells: Fundamentals to Optimization

Abstract

Acceleration of the oxygen reduction reaction at the cathode is paramount in the development of low-temperature solid oxide fuel cells. At low operating temperatures between 450 and 600 °C, the interactions between the surface and the bulk of the cathode materials greatly impact the electrode kinetics and consequently determine the overall efficacy and long-term stability of the fuel cells. This review will provide an overview of the recent progress in the understanding of surface-bulk interactions in perovskite oxides as well as their impact on cathode reactivity and stability. This review will also summarize current strategies in the development of cathode materials through bulk doping and surface functionalization. In addition, this review will highlight the roles of surface segregation in the mediation of surface and bulk interactions, which have profound impacts on the properties of cathode surfaces and the bulk and therefore overall cathode performance. Although trade-offs between reactivity and stability commonly exist in terms of catalyst design, opportunities also exist in attaining optimal cathode performance through the modulation of both cathode surfaces and bulk using combined strategies. This review will conclude with future research directions involving investigations into the role of oxygen vacancy and mobility in catalysis, the rational modulation of surface-bulk interactions and the use of advanced fabrication techniques, all of which can lead to optimized cathode performance.

Graphic Abstract

1 Introduction

Solid oxide fuel cells (SOFCs) are emerging energy conversion devices that can transform chemical energy from various fuels such as hydrogen, ammonia [1] or hydrocarbons [2, 3] into electricity with high efficiency (i.e. ,~ 60% alone and over 85% if combined with heat and power) [4, 5]. Corresponding high operating temperatures (450–1000 °C) also allow for the feasibility of internal fuel reforming inside fuel cells, thus providing SOFC technologies with unparalleled fuel flexibility as compared with other fuel cell technologies such as proton-exchange membrane fuel cells that normally demand strict fuel requirements [6].

Typical solid oxide fuel cells (Fig. 1) consist of three major components, including a cathode, an anode and an electrolyte. Here, the function of the cathode is to electrochemically reduce oxygen molecules into oxygen anions that are subsequently transported through a dense electrolyte to oxidize fuels at the anode, resulting in the transfer of electrons from the anode to the cathode through an external circuit and do the electrical work along its way.

Fig. 1

Microstructural schematic of a typical solid oxide fuel cell with corresponding physical–chemical processes

One significant challenge faced by SOFC technologies involves high operating temperatures that are generally above 800 °C. High operating temperatures lead to significant issues including (1) thermal expansion mismatch between cell components, (2) accelerated degradation of electrode and electrolyte materials and (3) limited material selection [7, 8], all of which degrade the overall cost-effectiveness of high-temperature fuel cell technologies. Thermal expansion mismatch can slow down start-up/shut-down procedures and lower fuel cell thermal-cycling stability. The accelerated material degradation is detrimental to long-term stability, which is a key prerequisite for mature electrochemical application. The limitation of the materials that can tolerate corresponding high temperatures directly translates to higher costs.

To address these challenges, reducing operating temperatures to between 450–600 °C hold promise while maintaining merits such as fuel flexibility and sufficient ion transport kinetics. However, low operating temperatures can also significantly decrease oxygen reduction reaction (ORR) kinetics and render conventional cathodes such as (La, Sr)MnO₃ (LSM) infeasible to catalyze ORR at acceptable rates. Conventional electrode materials for operating temperature above 800 °C are generally pure electronic conductors, which limit catalytically active sites for ORR to three-phase interfaces between the electronic conducting phase (e.g., LSM phase), the ionic conducting phase (e.g., electrolyte such as yttria-doped zirconia) and gaseous oxygen molecules. Therefore, the ORR catalysts at reduced temperature should possess both (1) high surface electrochemical activity toward ORR and (2) rapid oxygen-ion diffusion in the bulk, both of which will help extend the density of ORR active sites from the limited cathode/electrolyte interfaces throughout the cathode surface. Optimal cathodes at reduced temperatures should also possess high porosity and well-matched thermal expansion coefficients as well as good chemical compatibility with contacting components such as current collectors and electrolytes.

Based on these requirements, perovskite oxides (ABO₃) such as Ba_0.5Sr_0.5Co_0.8Fe_0.2O_3−δ [9], PrBa_0.5Sr_0.5Co_1.5Fe_0.5O_5+δ [10], SrSc_0.175Nb_0.025Co_0.8O_3−δ [11] and SrCo_0.8Nb_0.1Ta_0.1O_3−δ [12] are considered in recent decades to be promising cathode candidates for operation at reduced temperatures. This is because as compared with various simple metal oxides, perovskite oxides typically show much higher mixed ionic and electronic conductivities (MIECs) that can extend active sites throughout electrode surfaces. In addition, perovskite oxides can be combined with ionic conducting materials such as samarium-doped ceria to further improve structural stability and long-term stability [13]. Despite this, highly active perovskite oxides also present new challenges, including (1) the instability of corresponding crystal structures, (2) surface enrichment and (3) susceptibility to contaminants such as CO₂, SO₂ and Cr, all of which are predetermined by the interplay between the cathode bulk and surface [14,15,16]. The interactions between cathode bulk and surface become more significant at elevated temperatures, which are different from catalysis at room temperature in which cathode surfaces dominate ORR. Because of this, tremendous efforts have been devoted to investigate the interactions between cathode surfaces and bulk, their role in ORR catalysis, and modulation of bulk and surface properties to enhance cathode performance.

This review attempts to summarize recent advancements in the understanding of mechanisms for surface-bulk interactions and ORR kinetics as well as strategies to modulate cathode bulk and surface to optimize cathode efficiency and stability. This review consists of two major sections. In Sect. 2–4, the review discusses the mechanisms underlying perovskite surface segregation, ORR kinetics and cathode susceptibility to contaminants. The strategies and approaches to tailor bulk and surface properties are summarized in Sect. 5. This review concludes with a brief summary of recent findings and provide perspective in the development of next-generation cathode materials for LT-SOFCs.

2 Surface and Bulk of Perovskite Oxides

The surface and bulk of perovskites are significantly different in composition in which the formation of perovskite surfaces can be considered to be a result of the cleavage of perovskite bulk at a specific direction. In addition, perovskite surface atoms usually possess lower coordination numbers (i.e., higher surface free energies) than bulk atoms, which will lead to the rearrangement of surface atoms and the segregation of certain cations from the bulk to the surface [17]. These excess cations as a result of segregation from the bulk will subsequently precipitate onto perovskite surfaces in the form of oxide islands or cation-rich phases. As a result, these surface composition re-distribution and cation segregation processes can dynamically alter the composition and property of corresponding perovskite oxides, including surface and bulk composition [18, 19], surface microstructures [20], crystal structures [21], electrocatalysis-related properties (e.g., oxygen surface exchange kinetics [22, 23]), bulk diffusivities [24] as well as fuel reforming and oxidation [25,26,27].

Interactions between perovskite bulk and surface as facilitated by cation segregation are prominent at elevated temperatures. The ABO₃ perovskite oxides usually contain more than two cations (i.e., relatively large A cations and smaller B cations), and cation segregation can be categorized into A-site cation (e.g., Sr) segregation and B-site cation (e.g., Co, Ni, Fe) segregation (so-called exsolution). A-site cation segregation is a common phenomenon observed in perovskite oxide cathodes [17], whereas B-site cation exsolution is usually intentionally induced to form nanoparticles at the perovskite anode surface [14]. The former process occurs with relative ease at the top surface of materials and leave sub-surfaces enriched with B-site cations [28] but may not significantly affect the structural integrity and compositions of the bulk [29].

Figure 2a illustrates the mechanism of A-site cation (Sr) surface segregation from La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ perovskite oxide, as proposed by Niania et al. [30] based on results from in situ surface characterization. The exsolution normally occurs within the bulk, driven by reducing conditions, and has profound impacts on bulk crystal structure and composition [31]. Figure 2b provides an illustration of the exsolution of Ni from perovskite as proposed by Neagu et al. [32] based on their in situ observation of the Ni exsolution from La_0.43Ca_0.37Ti_0.94Ni_0.06O₃ being annealed at 900 °C under high vacuum. Based on these proposed mechanisms, it can be seen that both segregation processes are strongly correlated with (1) perovskite composition and structure, (2) external conditions such as temperature, gas composition and partial pressure, and electrical polarization [33]. The following subsections briefly summarize the various factors governing perovskite bulk-surface interactions.

Fig. 2

Proposed mechanisms for a A-site cation Sr surface segregation, reproduced with permission from Ref. [30], Copyright 2018, Royal Society of Chemistry, and b B-site exsolution of Ni at the perovskite surface. Reproduced with permission from Ref. [32]. Copyright 2019, American Chemical Society

2.1 Perovskite Composition and Crystal Structure

2.1.1 Cation Size

Perovskite oxides are known for high flexibility with various cation dopants at A- or B-sites. The size mismatch between hosts (e.g., La³⁺ with a Shannon ionic radius of 1.36 Å) [34] and dopants (e.g., Sr²⁺ with an ionic radius of 1.44 Å) can increase the elastic energy of lattices and constitutes the thermodynamic driving force of dopant migration from the bulk to the surface to minimize this elastic energy. Typical examples include alkaline-earth element (i.e., Sr²⁺ and Ba²⁺) enrichment at the surface of many perovskite oxides such as LSM [35, 36], (La, Sr)CoO_3−δ (LSC) [20, 37] or (Sm, Sr)CoO_3−δ [38]. For example, Lee et al. [29] reported a strong correlation between cation surface segregation and the ionic size of alkaline-earth dopants such as Sr and Ba. At elevated temperatures above 430 °C, as shown in Fig. 3a, the LaMnO₃ films doped with relatively larger dopants Ba (ionic radius = 1.61 Å) and Sr (ionic radius = 1.44 Å) than the host La (ionic radius = 1.36 Å) were roughened by the formation of small particles on the surface (Fig. 3a). These researchers also found that based on angle-resolved X-ray photoelectron spectroscopy dopant surface enrichment the extent of dopant segregation increased with dopant/host size mismatch level in the order of Ba > Sr > Ca (ionic size = 1.34 Å) (Fig. 3b, c). In another example, Kwon et al. [39] applied density functional theory (DFT) calculations to compare the dopant segregation energy of alkaline-earth element-doped LaBO₃ perovskite oxides with various B-site metals including Cr_0.5Mn_0.5, Mn, Fe and Co_0.25Fe_0.75. These researchers confirmed the essential role of A-site host/dopant size mismatching in driving dopant segregation (Fig. 3d, e), and also highlighted the role of the B-site cations in the dopant segregation process. They found that increasing B-site cation size can induce extra compressive strain in host lattices and therefore promote dopant segregation (Fig. 3e). Aside from alkaline-earth dopant segregation, our group also observed the more significant surface segregation of Li (0.92 Å) and K (1.64 Å) over SrFe_0.8Nb_0.1Ta_0.1O_3−δ perovskite oxide as compared with Na (1.39 Å) as partially contributed by size mismatch between the Sr host (1.44 Å) and the alkali dopants [18].

Fig. 3

a Atomic force microscopy images of Ca-, Sr- and Ba-doped LaMnO₃ (denoted as LCM, LSM and LBM, respectively) thin films at various temperatures in air for 1 h. b Normalized dopant/Mn and c dopant/La ratios obtained from angle-resolved X-ray photoelectron spectroscopy as a function of temperature. Reproduced with permission from Ref. [29]. Copyright 2013, American Chemical Society. d Surface segregation energy as a function of size mismatch between dopant/host at the A-site and f size mismatch between A-site dopants and B-site metals. Reproduced with permission from Ref. [39]. Copyright 2016, Royal Society of Chemistry

Kinetically, large-sized dopants such as Sr, Ba or K can slow down diffusion rates from bulk to surface. For example, Lee et al. [29] performed nudged elastic band calculations for the migration of alkaline-earth metals in doped LaMnO₃ and SmMnO₃ models and reported increased energy barriers for dopant diffusion with dopant size. A high energy barrier for cation diffusion indicates slow segregation kinetics. This effect is particularly prominent for alkali metal dopants if annealing durations were short. Although the extent of the K/Sr size mismatch is more significant than Li/Sr size mismatch [18], (Sr, Li/K) Fe_0.8Nb_0.1Ta_0.1O_3−δ surfaces were much more easily enriched with Li species (Li/Fe ratio =  ~ 7) than with K species (K/Fe ratio < 1).

2.1.2 Stoichiometry

The stoichiometry of cations and anions in perovskite oxides is another critical factor that can affect the interactions between perovskite bulk and surface. By lowering the stoichiometry of relatively large A-site cations, the elastic energy of lattices can be relieved and thus weaken the driving force for A-site cation segregation from bulk to surface. This has commonly been observed in A-site-deficient (La, Sr)_1−xMnO₃ oxides [40,41,42]. For example, Pang et al. [43] very recently developed an A-site-deficient La_0.5Ba_0.5−xCoO_3−δ ( x = 0 and 0.025) and reported that Ba deficiency can significantly suppress A-site metal surface segregation, which these researchers attributed to relaxed lattice strain due to the absence of large-sized Ba. On the other hand, materials with excess A-site dopants such as Ruddlesden–Popper (RP) structured A_n+1B_nO_3n+1 can result in A-site metal-rich surfaces after annealing in oxidizing environments and can commonly be observed in RP metal oxides such as Sr_n+1Ti_nO_3n+1 [44], (La, Sr)_n+1(Co, Fe)_nO_3n+1[45] and La_n+1Ni_nO_3n+1 [46].

The introduction of deficiencies (i.e., non-stoichiometry) at A-sites can also serve as a driving force to encourage the exsolution of B-site cations from bulk to surface to revert “defect-abundant” perovskite oxides to “defect-free” oxides [31]. These exsolution processes require reducing conditions however. For example, Neagu et al. [32] recently applied environmental transmission electron microscope (ETEM) to in situ observe the exsolution of Ni nanoparticles at the surface of A-site-deficient La_0.43Ca_0.37Ti_0.94Ni_0.06O₃ oxide under vacuum at 900 °C (Fig. 4a) in which further analysis of Ni particle height and width from these snapshots revealed that exsolved Ni nanoparticles grew in a stepwise and isotropic manner (Fig. 4b, c).

Fig. 4

a Time-resolved observations of Ni particle nucleation and growth under vacuum based on TEM at the surface of La_0.43Ca_0.37Ti_0.94Ni_0.06O₃ powder at 900 °C. b Plot of Ni particle width and height shown in (a) as a function of time and c plot of height versus width of the corresponding Ni particles. Reproduced with permission from Ref. [32]. Copyright 2019, American Chemical Society

Surface oxygen vacancies can also cause surface cation rearrangement as driven by electrostatic interactions [47]. For example, Lee et al. [29] investigated seven scenarios (Fig. 5a) with varying strength of electrostatic interactions related to LaMnO₃ and SmMnO₃ doped with divalent alkaline-earth elements at the A-site. Their DFT results revealed easy divalent dopant segregation (low energy for cation segregation) for models experiencing strong electrostatic interactions [i.e., negatively charged dopants (divalent dopants at trivalent cations) are attracted by positively charged oxygen vacancies on surface dopants and pushed from negatively charged A-site vacancies in the bulk (Fig. 5b)]. Tsvetkov et al. [23] further reported the effective suppression of Sr surface segregation through the surface modification of La_0.8Sr_0.2CoO_3−δ perovskite oxide with less reducible cations (e.g., Nb⁵⁺, Ti⁴⁺, Hf⁴⁺ and Al³⁺) at the surface. The ambient pressure X-ray photoelectron spectroscopy and X-ray adsorption spectroscopy results allowed these researchers to conclude that the observed suppression of surface segregation originated from more oxidized surfaces (i.e., less oxygen vacancies) as imparted by surface modification, which can weaken electrostatic interactions between perovskite oxide surfaces and Sr cations. These results further highlighted the essential role of surface oxygen vacancies in driving surface segregation in perovskite oxides.

Fig. 5

a Schematics of models based on LaMnO₃ or SmMnO₃ doped with alkaline-earth elements (denoted as blue spheres) with strengthened electrostatic dopant attraction correlating to model number from 1 to 7. b Comparison of calculated segregation energy of doped LaMnO₃ or SmMnO₃ for the seven models described in (a). Reproduced with permission from Ref. [29]. Copyright 2013, American Chemical Society

Oxygen vacancies can also affect B-site cation exsolution [48]. For example, Kwon et al. [25] investigated B-site cation segregation in layered PaBaMn₂O_5+δ perovskite oxides doped with various transition metals (e.g., Co, Ni, Mn and Fe) at B-sites and found that Co and Ni more easily migrated to the surface than Mn or Fe. These researchers also reported based on DFT calculations that Co and Ni dopants can facilitate more facile co-segregation with oxygen vacancies to the surface than Mn and Fe (Fig. 6) and that co-segregated oxygen vacancies can further contribute to the stabilization of the metallic dopant phase on the surface, which is also important for facile B-site cation exsolution.

Fig. 6

a Schematic of layered perovskite PrBaMn₂O_5+δ model doped with transition metals (denoted as T) including Co, Ni, Mn and Fe. b Comparison of co-segregation energy between different transition metal dopants. Reproduced with permission from Ref. [25]. Copyright 2017, Nature Publishing Group

2.1.3 Crystallinity and Surface Orientation

Perovskite oxide crystallinity and surface orientation both affect surface segregation. For example, Cai et al. [37] reported that low La_0.6Sr_0.4CoO_3−δ thin-film crystallinity can suppress Sr surface enrichment and suggested that the disordered structure can weaken both elastic and electrostatic driving forces by (1) allowing lattices to possess extra spacing (e.g., defects, cation vacancies) to accommodate relatively large Sr cations in the bulk and (2) increasing the oxygen vacancy content in the bulk. Surface orientation also plays a critical role in cation surface segregation. For example, Piskin et al. [49] reported the strong dependency of SrO_x species coverage on La_0.6Sr_0.4CoO₃ dense pellet surfaces on surface orientation in which the coverage of Sr increased in the order: (001) > (101) > (111). The effects of surface orientation are different for B-site cation segregation however. For example, Kim et al. [50] investigated La_0.2Sr_0.7Ti_0.9Ni_0.1O_3−δ thin films with three different surface orientations (001), (110) and (111) after chemical reduction in dry H₂ at 900 °C. They observed a strong correlation between surface orientation and the morphology and surface coverage of the exsolved Ni nanoparticles in which the size of Ni particles decreased in the order: (001) > (110) > (111), whereas the coverage of particles increased in the order: (001) < (110) < (111) (Fig. 7). These researchers further highlighted the critical role of interfacial energy between exsolved Ni particles and perovskite host lattices in the nucleation and particle growth processes. High interfacial energy can lead to higher nucleation energy barriers (i.e., slow nucleation rates and low particle populations) and larger critical nucleation sizes. Because of this, the (001) facet with relatively higher interfacial energy will result in less area coverage but larger particle sizes than the other facets.

Fig. 7

Atomic force microscopy images of the top surface structures of La_0.2Sr_0.7Ti_0.9Ni_0.1O_3−δ thin films with a (001), b (110) and c (111) surface orientations after being reduced in dry H₂ at 900 °C for 12 h. Reproduced with permission from Ref. [50]. Copyright 2019, American Chemical Society

2.1.4 Lattice Strain

Lattice strain can also affect the elastic energy of lattice cations and therefore influence the level of cation surface segregation in perovskite oxides. In general, tensile strain in lattices can cause A-site cation surface segregation, whereas compressive strain inhibits segregation [22, 51]. For example, Cai et al. [52] applied pulsed laser deposition (PLD) to deposit tensile- and compressive-strained La_0.8Sr_0.2CoO₃ (LSC) thin films over SrTiO₃ (STO) and LaAlO₃ (LAO), respectively, and reported stronger Sr surface enrichment in the tensile-strained LSC/STO as compared with the compressive-strained LSC/LAO. Lattice tensile strain can also intensify Sr surface segregation in La_0.7Sr_0.3MnO₃ as compared with compressive strain [53]. Consistently, Ding et al. [54] and Yu et al. [20] reported similar effects of lattice strain on Sr surface segregation in (La, Sr)(Co, Fe)O_3−δ (LSCF) oxides.

2.2 External Conditions

2.2.1 Temperature

Although the minimization of elastic and electrostatic energy is the main thermodynamic driving force for cation surface segregation, the rates of cation migration from perovskite oxide bulk to the outer surface are slow due to kinetic limitations. Therefore, surface cation enrichment is usually more prominent at elevated temperatures and longer annealing durations (Fig. 3b, c) [19, 22, 27, 28, 41, 55, 56]. For example, Wen et al. [57] applied in situ ambient pressure XPS to monitor Sr surface segregation in La_0.6Sr_0.4CoO_3−δ films at temperatures between 200 and 520 °C and reported that Sr segregation became increasingly pronounced with increasing temperature and quickly (within 1 h) reached equilibrium at certain temperatures. This quick response of Sr segregation to temperature change implies that temperature also contributes to the thermodynamic driving force in addition to the kinetic acceleration of surface segregation. Using isotopic ⁸⁶Sr and cation tracers, Kubicek et al. [58] further investigated cation diffusivity over La_0.6Sr_0.4CoO_3−δ thin films and reported a less than one order of magnitude difference in diffusivity between A-site and B-site cations at temperatures above 600 °C (relatively small B-site cations diffuse faster than A-site cations), demonstrating that kinetic contributions of temperature are minor as compared with corresponding thermodynamic contributions to cation segregation.

2.2.2 Gas Atmosphere

Gas molecules and partial pressures have significant impacts on the interactions between perovskite oxide surfaces and bulk. Oxygen partial pressures can also influence the oxidation state and size of reducible cations as well as oxygen vacancy content, which in turn can affect elastic and electrostatic interactions within perovskites. As mentioned, the segregation of A-site cations such as Sr normally occurs under oxidizing environments (Fig. 3), whereas the exsolution of B-site cations such as reducible transition metals is facilitated through reducing environments (H₂ gas or high vacuum) [26, 27, 31, 59] (Figs. 4, 7). The perovskite oxide surfaces are sensitive to gas molecules such as moisture and CO₂ to induce cation segregation. For example, Sharma et al. [60] reported pronounced alkaline-earth dopant surface segregation under humid atmosphere and suggested that the promoted cation segregation was related to interactions between moisture and surface oxygen vacancies. Similarly, Niania et al. [30] reported more significant Sr segregation for La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ in the presence of water than in air or pure O₂.

The presence of acidic CO₂ can further drive the surface segregation of basic cations such as alkaline-earth and alkali metals. These CO₂-cation interactions can also promote the carbonation of cations at the surface [18, 61] and induce the destruction of crystal structures in the subsurface region [21]. Our previous work on SrCo_0.85Ta_0.15O_3−δ cathodes for SOFCs achieved more significant Sr surface enrichment if annealed under 10% CO₂ in air as compared with treatment in air alone [18]. In the presence of CO₂, additionally, the Sr_0.95A_0.05Fe_0.8Nb_0.1Ta_0.1O_3−δ (A is the alkali metal dopant) showed significant alkali metal dopant segregation to the surface as compared with ones without CO₂ treatment (Fig. 8). Moreover, interactions between basic alkali metals and acidic CO₂ can serve as a key driving force for alkali metal segregation in addition to elastic interactions due to dopant/host size mismatch. These combined driving forces were confirmed by XPS in our study, which showed that cation surface segregation increased following the trend: K > Li > Na.

Fig. 8

SEM micrographs of dense pellets of SrFe_0.8Nb_0.1Ta_0.1O_3−δ doped with alkali metals (i.e., Li, Na and K) at A-sites before a–d and after e–h treatment in the presence of CO₂ at 600 °C. Reproduced with permission [18], Copyright 2019, American Chemical Society. SEM images of particles exsolved from La_0.8Ce_0.1Ti_0.6Ni_0.4O₃ perovskite oxide i–j in the presence of 5% CO at 900 °C for 10 h and k–l under 2.5% H₂O/5% H₂ at 1000 °C for 10 h. Reproduced with permission from Ref. [32]. Copyright 2019, American Chemical Society

Despite the essential role of reducing atmospheres in facilitating B-site cation exsolution, gas molecules can also influence the morphology of exsolved particles. For example, hydrogen reduction normally leads to quasi-spherical shaped particles [25, 27]. In contrast, Neagu et al. [32] reported that high vacuum conditions can lead to nanoparticles on the surface of La_0.8Ce_0.1Ti_0.6Ni_0.4O₃ with well-defined facets, whereas CO reduction-produced cubic shaped exsolved Ni particles and humidified hydrogen led to the formation of heterostructures.

2.2.3 Electrochemical Polarization

Electrical polarization (or bias) can vary oxygen chemical potential, cation oxidation states and lattice size, all of which have profound impacts on the interactions between perovskite oxide surfaces and bulk. Typically, cathodic polarization can lower oxygen chemical potentials, create oxygen vacancies and increase the ionic size of reducible cations in lattices, whereas anodic polarization results in the opposite. These properties are related to the elastic and electrostatic driving forces for cation segregation in which cathodic polarization can increase A-site cation segregation degrees over perovskite oxides such as LSC [56, 62,63,64] and LSM [35, 65, 66]. Anodic polarization can impart effects similar to high oxygen partial pressure that can promote A-site cation segregation. Despite this, inconsistent results have been reported in the literature. For example, Huber et al. [67, 68] reported that the degree of Sr enrichment decreased under cathodic polarization and increased under anodic polarization in LSM and (La, Sr)(Cr, Mn)O₃ perovskite oxides. Kim et al. [35] recently also reported that both cathodic and anodic polarizations can lead to significant alkaline-earth dopant segregation on the surface over deposited LaMnO₃ perovskite oxides. These inconsistencies may be due to the various synthesis methods and annealing processes used to obtain oxides prior to electrochemical polarization in the different studies. Despite this, these results do highlight the essential role of balance between elastic and electrostatic driving forces. According to results by Kim et al. [35], elastic driving forces are dominant if oxides are under anodic polarization, whereas electrostatic interactions become crucial if oxides are cathodically polarized. This conclusion is reasonable because anodic polarization tends to shrink lattices through the oxidization of cations, which promotes elastic interactions due to dopant/host size mismatch. Cathodic polarization tends to alleviate elastic interactions by expanding lattices to create more oxygen vacancies and therefore increases electrostatic interactions that attract cation segregation.

3 Oxygen reduction and Mixed Conductivity

3.1 ORR Mechanism

The function of a cathode is to reduce oxygen molecules into oxygen ions as represented by Eq. (1). This reaction primarily proceeds through (1) O₂ diffusion into a porous cathode, (2) O₂ adsorption onto the cathode surface, (3) the partial or complete reduction of O₂ into oxygen ions such as O²⁻ and O⁻ and (4) oxygen-ion migration to the electrolyte along the cathode surface, the cathode-electrolyte interface and the cathode bulk phase (Fig. 9) The rate of ORR is mainly controlled by either oxygen surface exchange (including oxygen dissociative adsorption and lattice incorporation) or oxygen-ion diffusion.

Fig. 9

Diagrams of typical mechanisms believed to govern ORR in cathode materials. α, β and γ represent electronic, gas and ionic phases, respectively. a Incorporation of oxygen molecules into the electronic bulk phase (if α is a mixed ionic-electronic conductor); b adsorption and/or partial reduction of oxygen molecules on the electronic phase surface; c bulk phase or surface phase migration of oxygen ions (O²⁻ and Oⁿ⁻, respectively) toward the α/γ interface; d charge transfer of O²⁻ or Oⁿ⁻ across the α/γ interface; e one or multiple mechanisms wherein the electrolyte is active for generating and transporting oxygen species. Reproduced with permission from Ref. [69]. Copyright 2004, American Chemical Society

$$\frac{1}{2}{\mathrm{O}}_{2} (\mathrm{g})+{2\mathrm{e}}^{-}\to {\mathrm{O}}^{2-}$$

(1)

In general, oxygen surface exchange can be interpreted as three consecutive stages involving (1) oxygen adsorption, (2) oxygen dissociation and (3) oxygen species incorporation into crystal lattices (Fig. 10). The rate-determining step of surface exchange is highly dependent on the availability of surface oxygen vacancies. For example, for cathode surfaces with limited oxygen vacancies such as LSM, oxygen migration coupled incorporation may be the rate-controlling step for oxygen surface exchange [71], whereas for cathode surfaces with sufficient oxygen vacancies such as LSCF and BSCF, surface exchange processes are generally dominated by the dissociation of oxygen ions (O₂⁻ or O₂²⁻).

Fig. 10

Schematic of possible pathways for oxygen incorporation into mixed ionic and electronic conductors (MIECs) with a few oxygen vacancies and b abundant oxygen vacancies according to results from ab initio calculations. Here, red arrows indicate the rate-limiting step, green arrows are the fast processes and dashed arrows represent the slower alternative parallel step. Reproduced with permission from Ref. [70]. Copyright 2012, Materials Research Society

Oxygen diffusion is widely considered to be a hopping process involving oxygen ions from one equilibrium site to an adjacent site along the surface or within the bulk of a material through interstitial or vacancy mechanisms. For example, oxygen-ion migration in materials such as RP structured oxides (Fig. 11a) [72] usually relies on the interstitial mechanism. The oxygen ions located at one interstitial site (i.e., free space) migrate to an adjacent interstitial site [73, 74]. Alternatively, oxygen-ion transport in fluorite-/perovskite-type materials (Fig. 11b, c) occurs through the vacancy mechanism [75] in which oxygen diffusion is facilitated through oxygen-ion hopping from occupied oxygen sites to neighboring vacancy sites [76]. The oxygen vacancy concentration in corresponding materials is critical for oxygen diffusion and overall ORR kinetics. Because the vacancy mechanism is more common and suitable for oxygen ionic conduction, this review will mainly provide an overview of vacancy-driven progress in oxygen-ion diffusion. Detailed information concerning oxygen vacancy formation and oxygen-ion mobility is discussed in Sect. 3.2.

Fig. 11

Crystal model of a a RP structure. Reproduced with permission from Ref. [72]. Copyright 2016, Royal Society of Chemistry. Oxygen migration pathway in b fluorite and c perovskite structures, arrows in (b) and (c) represent the migration orientation of oxygen ions

3.1.1 Oxygen Surface Exchange Versus Oxygen-Ion Diffusion

The oxygen surface exchange coefficient (k_chem) and the oxygen-ion diffusion coefficient (D_chem) are two parameters that can be used to quantify ORR reaction rates on MIEC material surfaces and bulk, respectively. The k_chem reflects electrode surface oxygen kinetics (i.e., oxygen adsorption, dissociation and incorporation [77]), whereas D_chem describes bulk oxygen-ion diffusion rates. Several techniques such as electrical conductivity relaxation (ECR) [78, 79] and isotope exchange [80] measurements can be applied to obtain k_chem and D_chem parameters. For example, ECR experiments can be used to calculate k_chem and D_chem values through the following formula [81, 82]:

$$ \begin{aligned} \sigma_{n} = \frac{\sigma (t) - \sigma (0)}{{\sigma (\infty ) - \sigma (0)}} & = 1 - \sum\limits_{p = 1}^{\infty } {\frac{{2L_{x}^{2} }}{{\beta_{p,x}^{2} (\beta_{p,x}^{2} + L_{x}^{2} + L_{x} )}}} \exp \left( {\frac{{\beta_{p,x}^{2} }}{{l_{x}^{2} }}D_{{{\text{chem}}}} t} \right) \times \sum\limits_{q = 1}^{\infty } {\frac{{2L_{y}^{2} }}{{\beta_{q,y}^{2} (\beta_{q,y}^{2} + L_{y}^{2} + L_{y} )}}} \exp \left( {\frac{{\beta_{q,y}^{2} }}{{l_{y}^{2} }}D_{{{\text{chem}}}} t} \right) \\ & \quad \times \sum\limits_{r = 1}^{\infty } {\frac{{2L_{z}^{2} }}{{\beta_{r,z}^{2} (\beta_{r,z}^{2} + L_{z}^{2} + L_{z} )}}} \exp \left( {\frac{{\beta_{r,z}^{2} }}{{l_{z}^{2} }}D_{{{\text{chem}}}} t} \right) \\ \end{aligned} $$

(2)

in which σ(0), σ(t) and σ (∞) are the initial, time-dependent and final conductivities, respectively, whereas L_x, L_y and L_z parameters are defined as:

$$ L_{x} = \frac{{l_{x} K_{{{\text{chem}}}} }}{{D_{{{\text{chem}}}} }},\;\;L_{y} = \frac{{l_{y} K_{{{\text{chem}}}} }}{{D_{{{\text{chem}}}} }}\;\;{\text{and}}\;\;L_{z} = \frac{{l_{z} K_{{{\text{chem}}}} }}{{D_{{{\text{chem}}}} }} $$

(3)

in which 2 lx, 2l_y and 2l_z are the side lengths of the bar sample. In addition, β_p,x, β_q,y and β_r,z coefficients are the roots of transcendental equations:

$$ \beta_{p,x} \tan \beta_{p,x} = L_{x} ,\;\;\beta_{q,y} \tan \beta_{q,y} = L_{y} \;\;\beta_{r,z} \tan \beta_{r,z} = L_{z} $$

(4)

Here, the magnitudes of p, q and r provide partial influence on calculation accuracy.

Tsvetkov et al. [23] further proposed that k_chem can be obtained based on surface polarization resistance (i.e., low-frequency impedance response) as measured using electrochemical impedance spectroscopy (EIS) through Eq. (5):

$$ k_{{{\text{chem}}}} = \frac{{k_{\text{B}} T}}{{4e^{2} c_{\text{o}} R_{\text{s}} }} $$

(5)

in which k_B is the Boltzmann constant, T is the temperature, e is the electronic charge, c_o is the total lattice oxygen concentration, and R_s is the low-frequency impedance of surface polarization resistance.

A summary of the k_chem and D_chem values of common cathode materials recently reported in the literature is given by Fig. 12. The reported coefficients, especially k_chem, can vary significantly over the same material, which may be due to surface sensitivity arising from segregation (e.g., SrO, BaO), reconstruction and contamination [5]. The values of k_chem and D_chem (Fig. 12a, b) were also reported to decrease with temperature, which consequently reduces ORR kinetics at lower temperatures. Furthermore, the values of k_chem were ~ 0.2–2 orders higher than those of D_chem over the same material at the same temperature, which is a trend that is consistent with findings by Souza and Kilner [91] in (La, Sr)(Mn, Co)O₃.

Fig. 12

Oxygen transport data of perovskite-type and other oxides from literature published from 2016 to 2020 [83,84,85,86,87,88,89,90], including the temperature dependence of a k_chem and b D_chem and c the correlation between k_chem and D_chem. Materials with different symbols are also shown in the figure

3.2 Electrical Conductivity

Electrical conductivity typically consists of electronic and ionic conductivity [12]. A high electronic conductivity can ensure electron transport to active sites to facilitate charge transfer to adsorbed oxygen species, whereas high oxygen ion conductivity is vital for reduced oxygen-ion transport to electrolyte. Because of this, high electronic and ionic conductivities are essential for efficient ORR catalysis, particularly at reduced temperatures. However, the formation of oxygen vacancies can reduce charge carriers such as holes for electron conduction, meaning the elucidation of which conductivity dominates ORR kinetics is crucial in cathode material optimization, especially at reduced temperatures. Based on the correlation summary provided by Wang et al. [70] between electronic conductivity and surface exchange coefficient (Fig. 13a), no apparent relationship exists between these two parameters, suggesting that electronic conductivity is not a dominant factor for ORR catalysis. This also explains why BSCF with relatively low electronic conductivity is more ORR active than La_0.74Sr_0.18MnO_3−δ. In contrast, a clear trend exists between oxygen-ion conductivity and surface exchange kinetics (Fig. 13b), highlighting the dominant role of oxygen-ion conductivity over electronic conductivity in ORR catalysis.

Fig. 13

a Correlation between electronic conductivity and the oxygen exchange rate constant (k) at 750 °C and ${P}_{{\mathrm{O}}_{2}}$ =1 bar. b Bulk ionic conductivity as a function of surface exchange coefficient as determined by isotope exchange (k^*) and chemical relaxation (k^q) experiments. Reprinted with permission [70], Copyright 2012, Materials Research Society

3.2.1 Electron Conduction

Electron conduction in most perovskite cathodes is believed to experience small polaron hopping, which is a diffusion-type process that involves electrons (or holes) hopping between B-site transition metal ions and mixed valences (e.g., TM³⁺/TM⁴⁺, so-called redox couples) such as Fe [92], Cu [93], Co [89], Ce [94] and Mn [95, 96]. Charge carrier migration generally occurs in conjunction with local structural distortion, and polarons are defined as quasiparticles consisting of charge carriers and associated local structural distortion [97]. A high concentration of charge carriers (e.g., electrons or holes) and availability of redox couples as hopping sites are prerequisites for fast electron conduction. The small polaron hopping process can generally be described by Eq. (6) [97, 98]:

$$ \rho (T) = AT^{m} \exp \left[ {\left( {\frac{{T_{{0}} }}{T}} \right)^{p} } \right] $$

(6)

in which ρ(T) is the electrical resistivity, A is a constant, T is the temperature, T₀ is the characteristic temperature, and m and p are interrelated parameters that depend on the exact mechanism of the hopping process.

In particular, hopping sites can be created by altering material composition (i.e., incorporating aliovalent cations into perovskite structures). Using Pr_1−xSr_xFeO₃ [99] as an example, the partial substitution of divalent Sr cations into Pr-sites can increase electronic conductivity and Fe cation charge compensation from + 3 to + 4, which will leads to increased Fe³⁺/Fe⁴⁺ coupling for electron hopping. Besides, the charge disproportionation of B-site cations from intermediate to high or low valences can also increase the level of hopping sites. One typical example is the disproportionation of Fe³⁺ into Fe²⁺ and Fe⁴⁺ in Ba_1−xLa_xFeO_3−δ (x = 0.1–0.4), which can increase electronic conductivity [100]. Note that these composition alterations can further affect the structural symmetry of materials, which can be reflected by p-d orbital overlap between B-site cations and oxygen ions (Fig. 14) [97, 101]. With a perfect cubic perovskite structure, this p–d overlap can be maximized with an optimal B–O–B bond angle of 180° to facilitate the mobility of electrons (or holes) from t_2g orbitals to neighboring p_π orbitals.

Fig. 14

Schematic of a an ABO₃ pseudo-cubic (or cubic) structure with oxygen-ion p_σ and p_δ orbitals and B-site cation e_g (${d}_{{{x}}^{2}-{{y}}^{2}}$, ${d}_{{{z}}^{2}}$) and t_2g (d_xy, d_yz, d_xz) orbitals. Reproduced with permission from Ref. [101], Copyright 1967, American Physical Society. b Covalent bonds of p_σ–p_σ and p_π–d_xy between an oxygen ion and a B-cation

Temperature is a significant factor for electronic conductivity based on Eq. (6). Temperature can also affect the number of hopping sites by influencing the electronic structure of B-site cations. Using Co-containing oxides as an example, Co³⁺ at low temperatures exhibiting low-spin configuration (${\text{t}}_{2\mathrm{g}}^{6}{\mathrm{e}}_{\mathrm{g}}^{0})$ can be transformed into intermediate-spin (${\text{t}}_{2\mathrm{g}}^{5}{\mathrm{e}}_{\mathrm{g}}^{1})$ and subsequent high-spin (${\text{t}}_{2\mathrm{g}}^{4}{\mathrm{e}}_{\mathrm{g}}^{2})$ configurations with increasing temperature [102, 103]. Because of this, the reduction of temperature will result in electron transfer from high-spin Co³⁺ to low-spin cations that are prone to transformation into Co⁴⁺ and Co²⁺ [104].

Gas atmospheres (e.g., increasing or decreasing oxygen partial pressure (${P}_{{\mathrm{O}}_{2}}$)) can also introduce redox couples and promote electron conduction [105]. For example, Mizusaki et al. [106] reported that increases in ${P}_{{\mathrm{O}}_{2}}$ for (La, Sr)FeO_3−δ at temperatures above 900 °C can transform n-type conductivity (Fe valence between + 2 and + 3) into p-type conductivity (Fe valence between + 3 and + 4) to increase electrical conductivity. Similar results have also been observed in other ion-containing materials such as La_0.6Sr_0.4Co_0.8Fe_0.2O_3−δ [107], La_1−xSr_xFeO_3−δ (x = 0, 0.1) [108] and SrFe_1−xSi_xO_3−δ (x = 0.05–0.20) [109].

3.2.2 Oxygen-Ion Conduction

As discussed in Sect. 3.1, the ionic conduction of perovskite-type materials occurs through an oxygen vacancy mechanism that is related to oxygen vacancy formation and oxygen-ion migration in lattices.

3.2.2.1 Formation of Oxygen Vacancy

Oxygen vacancies, commonly described using oxygen non-stoichiometry, are anion defects in lattices as a result of doping or external conditions (e.g., temperature and gas atmosphere change). In most perovskite oxides, these two factors can lower the overall positive charge of cations and create oxygen vacancies to compensate for the induced charge imbalance. A typical example in B-site cations is described in Eq. (7):

$$ 2{\text{B}}_{{\text{B}}}^{ \times } + {\text{O}}_{{\text{O}}}^{ \times } \to {\text{V}}_{{\text{O}}}^{ \bullet \bullet } + 2{\text{B}}_{{\text{B}}}^{\prime } + {}_{2}^{1} {\text{O}}_{2} $$

(7)

in which “ × ” represents neutrality and “‧” and “′” represent the positive and negative charges , respectively.

Although the incorporation of low oxidation state dopants can reduce overall positive charge in corresponding lattices, the formation of oxygen vacancies depends on the reducibility of B-site transition metals (e.g., Mn, Fe, Co and Ni) because induced charge imbalance can also be compensated by reducing the average valence of B-sites. Here, the creation of oxygen vacancies is easier in the case of Co- and Ni-based perovskite oxides as compared with Mn- and Fe-based perovskites, possibly due to the poorer reducibility of Co and Ni than Mn and Fe [110,111,112,113].

Lattice geometry is another important factor in the formation of oxygen vacancies. For example, doping with small-sized B-site cations can force corresponding B-site hosts to increase in size to maintain a perovskite structure, which is normally accompanied by the reduction of valence states and thus the formation of oxygen vacancies. For example, Lee et al. [114] reported that smaller ionic radius Ge⁴⁺ doping at B-sites in PrBa_0.5Sr_0.5Fe_2−xGe_xO_5+δ (PBSFG, x = 0 and 0.1) perovskite can facilitate the reduction of Fe⁴⁺ to Fe³⁺ and the generation of oxygen vacancies by reducing the coordination state from FeO₆ to FeO₄ octahedra. Doping with large ionic sized A-site cations can also expand lattices and therefore increase the size of B-site reducible cations to create oxygen vacancies. For example, Merkle et al. [115] applied DFT calculations to investigate oxygen vacancy formation and migration in Ba_1−xSr_xCo_1−yFe_yO_3−δ (x = 0.5, 1, y = 0, 0.2, 0.25, 0.75 and 1) perovskites and found that Ba (x = 0.5) doping at A-sites can reduce the formation energy of oxygen vacancies by 0.3 eV due to the larger Ba ions as compared with Sr ions, which can result in lattice expansion and low-valence Co.

External conditions such as temperature and gas atmosphere are essential to the equilibrium of oxygen species in gas and perovskite solid phases. At elevated temperatures, especially above 400 °C, oxygen is prone to escape from the solid phase to the gas phase, which causes the solid phase to possess lowered overall lattice positive charge and experience the formation of oxygen vacancies. Similarly, lowering oxygen partial pressure in gas atmospheres can shift equilibriums to create lattice oxygen vacancies [116]. Different from doping effects, external condition effects largely depend on equilibrium conditions, meaning that corresponding oxygen vacancy changes are reversible if temperatures and gas atmospheres revert back to original conditions. Such behavior provides opportunities to evaluate oxygen surface exchange kinetics by varying oxygen partial pressure [12, 117, 118].

High oxygen vacancy content that is above optimal can also destabilize beneficial perovskite structures. For example, the transformation of SrCoO_3−δ-based oxides into an oxygen-ordering brownmillerite-type structure with significantly increased oxygen vacancy content is easily achievable through exposure to either high temperature or reduced oxygen partial pressure [119,120,121,122] and will significantly reduce oxygen condition rates in the corresponding lattice and therefore degrade overall cathode performance. In addition, the formation of oxygen vacancies based on Eq. (7) can reduce the concentration of charge carriers such as holes for hopping processes and therefore degrade electronic conductivity. This is a common phenomenon in many perovskite materials such as SrCoO_3−δ-based perovskite oxides [12, 18, 19, 123, 124].

3.2.2.2 Mobility of Oxygen Ions

In addition to oxygen vacancy content, the ease of oxygen-ion migration in corresponding lattices can also determine overall oxygen diffusion rate. Moreover, oxygen-ion migration is naturally accompanied by the formation and breakage of metal–oxygen bonds and the diffusion of oxygen-ions through lattice voids. Because of this, the bond strength between metals and oxygen and the geometry of lattices are important factors determining oxygen-ion mobility. Weakened interactions further favor oxygen-ion migration [125]. For example, Mayeshiba and Morgan [126] applied ab initio methods to investigate the energy barriers of oxygen vacancies in more than 40 perovskite oxide models and compared corresponding oxygen-ion migration energy with oxygen vacancy formation energy (Fig. 15) to confirm that low metal–oxygen bond strength as measured by oxygen formation energy and oxygen p-band centers generally led to facile oxygen-ion migration in lattices.

Fig. 15

Relationship between oxygen migration barrier and oxygen vacancy formation energy calculated at 900 °C and 0.1 atm (1 atm=101.325 kPa) ${{P}}_{{\mathrm{O}}_{2}}$. Reproduced with permission from Ref. [126]. Copyright 2016, Elsevier

High symmetry can further promote high oxygen-ion conduction, particularly in polycrystalline materials, because corresponding oxygen ions can migrate isotopically to adjacent vacancy sites. This is well exemplified by the inferior oxygen-ion conductivity of oxygen-vacancy ordering brownmillerite structures as compared with simple cubic perovskites with less oxygen vacancy content [127,128,129]. In addition, the oxygen ions need to permeate through a trigonal planar phase established by two A-site and one B-site ions (Fig. 16a), whereas the large space surrounding an A₂B triangle phase enables facile oxygen-ion migration. This surrounding space can be referred to as a “saddle point” in which the critical radius (r_c) can be defined to describe the size of the space (Fig. 16b) based on Eq. (8):

$$ r_{\text{c}} = \frac{{r_{\text{B}}^{2} - r_{\text{A}}^{2} + a\left( {\frac{3}{4}a - \sqrt 2 r_{\text{B}} } \right)}}{{2(r_{\text{A}} - r_{\text{B}} ) + \sqrt 2 a}} $$

(8)

in which r_A and r_B are the radii of A- and B-site cations, whereas a represents the lattice constant and the pseudo-cubic structure and can be calculated by using Eq. (9) [130]:

Fig. 16

a Schematic of oxygen-ion migration from an occupied site to an adjacent vacant site through permeation through the A₂B trigonal plane (red area). b Determination of the critical radius (r_c) among the A₂B trigonal plane

$$ a \approx \sqrt[3]{{V_{{{\text{Cell}}}} }} = 2.15r_{B} + 2.72 - 1.40(\tau^{ - 1} - 1) $$

(9)

in which τ_o is the Goldschmidt tolerance factor as determined based on Eq. (10) [131]:

$$ \tau = \frac{{r_{\text{A}} + r_{{\text{o}}} }}{{\sqrt {2(r_{\text{B}} + r_{{\text{o}}} )} }} $$

(10)

in which r_o is the radius of the anion (usually the oxygen ion).

Despite these calculations, Mogensen et al. [132] found that calculated r_c values were usually less than 1.05 Å for typical perovskite oxides and were far smaller than the size of oxygen anions (1.4 Å), suggesting that oxygen ions cannot travel through the A₂B trigonal plane as defined by the hard-sphere model. Such inconsistencies highlight the role of lattice relaxation (or constant expansion [133]) in facilitating oxygen-ion migration.

Lattice-free volume (V_f) is another parameter used to describe geometrical effects on oxygen-ion mobility. Here, V_f is defined as the volume of unoccupied ions in lattice unit cells and can be calculated using Eq. (11) [134]:

$$ V_{\text{f}} = a^{3} - \frac{4}{3}\uppi \left[ {r_{\text{A}}^{3} + r_{\text{B}}^{3} + (3 - \delta )r_{\text{o}}^{3} } \right] $$

(11)

in which δ is the oxygen non-stoichiometry in the lattice.

Here, Cook et al. [135] reported that increasing V_f values from ~ 7 ų to ~ 36 ų can effectively reduce oxygen-ion migration activation energy from ~ 1.6 eV to ~ 0.4 eV (Fig. 17a) and Lu et al. [134] reported that the reduction of In doping content from 0.2 to 0.1 in BaFe_1−xIn_xO_3−δ (x = 0.1–0.2) perovskite oxide increased free volume by ~ 0.4 ų and therefore lowered oxygen-ion migration activation energy by nearly 0.12 eV at 900 °C (Fig. 17b). Despite these results, geometric effects on oxygen migration remain controversial in which theoretical calculations from Mayeshilba and Morgan [126] revealed no significant correlation between oxygen migration energy barrier and geometric factors such as lattice volume.

Fig. 17

a Activation energy of oxygen-ion migration in various perovskite oxides as a function of free volume. Reproduced with permission [135], Copyright 1992, The Electrochemical Society. b Influence of doping content (x) in BaFe_1−xIn_xO_3−δ (x = 0.1–0.2) at 900 °C on oxygen-ion migration energy and free volume. Reproduced with permission from Ref. [134]. Copyright 2015, Royal Society of Chemistry

4 Cathode Degradation Mechanisms

In contrast to conventional power generation devices such as internal combustion engines, fuel cell stacks cannot be easily repaired through the simple replacement of parts, meaning extended lifetimes (i.e., more than 40,000 h operation [136] with degradation rates less than 0.2% per 1000 h operation) are essential in practical SOFC applications. Lowering the operating temperature of SOFC is an effective strategy to alleviate the potential degradation due to particle coarsening and volatile contamination at elevated temperatures. Cathode materials still face challenges in terms of (1) sustaining original structures and surfaces optimal for ORR catalysis and (2) tolerating poisoning effects from contaminants (e.g., CO₂, SO₂, humidity, chromium, boron and silicon) in ambient air and/or from other contacting fuel cell components such as interconnects and sealants. In addition, these two degradation modes are highly related to the bulk and surface of perovskite cathodes as well as their interaction with gas and solid phases in contact with cathodes and can lead to increased ohmic and kinetic loss (Fig. 18) [137].

Fig. 18

Brief summary of cathode degradation modes in the presence of common contaminants

4.1 Crystal Structure and Surface Degradation

4.1.1 Crystal Structure

The crystal structures of perovskites are essential for cathode performance because they predetermine mixed electron and oxygen-ion conductivity, cation segregation (e.g., through elastic energy minimization) and material stability. Because of this, seminal efforts have been devoted to the investigation of correlations between cathode performance and crystal structure [12, 19, 117,118,119, 138, 139]. Typically, high symmetry perovskite structures such as simple cubic perovskite oxide and A-site or B-site-ordered perovskites render sufficiently mixed electronic and ionic conductivities for ORR catalysis in which the Goldschmidt tolerance factor [140] is a common descriptor that can be used to predict the stability of perovskite structures with given composition. Here, the occupation of A-, B- and X-sites in perovskite materials is mainly dependent on elemental ionic size (Fig. 19) with corresponding ionic radii being easily obtained from the Shannon effective ionic radii [34]. And despite the Goldschmidt tolerance factor (Eq. 10) being an over-simplified approximation [141], it is still a practical tool to predict the trend of perovskite stability in which (1) τ > 1 prefers the formation of hexagonal or tetragonal phases; (2) τ close to 1 likely leads to cubic structures; and (3) τ < 0.9 allows for the easy formation of orthorhombic or rhombohedral structures.

Fig. 19

a Schematic of a simple cubic perovskite structure of ABX₃ and b a summary of elements that can occupy A, B, and X-sites of perovskite materials

The crystal structure of perovskites is also closely related to the interatomic distance between ions in lattices. Because of this, variations in cation oxidation state and cation and oxygen stoichiometry can lead to structural transformation at elevated temperatures and varied oxygen partial pressures. For example, the oxidation states of B-site reducible cations such as Co and Fe can easily be lowered by elevating temperatures and/or lowering oxygen partial pressures. These cations with lowered oxidation states possess increased ionic sizes, which will further decrease tolerance factors and induce phase transition from cubic phase to hexagonal (e.g., BSCF-based oxides [142]) or orthorhombic phases (e.g., brownmillerite-type structure for Sr(Co, Fe)O_3−δ-based oxides [119, 143, 144]).

4.1.2 Surface

Driven by elastic and electrostatic interactions, cation segregation from surface to bulk generally impacts surface oxygen exchange kinetics with many studies reporting oxygen surface exchange kinetics degradation in perovskite oxides with significant Sr enrichment [17, 23, 28, 37, 145]. Here, because B-site redox cations (e.g., Co ions) are active centers facilitating oxygen surface kinetics [146], degraded surface oxygen exchange rates can be attributed to increased coverage of A-site cations and/or insulating A-site cation surface precipitates (e.g., SrO and Sr(OH)₂) [147]. As mentioned in Sect. 2, A-site cation segregation typically occurs at the near-surface region of perovskite oxides and thus may not significantly affect bulk oxygen-ion transport in which despite observing degraded oxygen surface kinetics, Baque et al. [148] in their study found that Sr surface enrichment did not significantly impact bulk oxygen-ion conductivity in porous La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ perovskite oxide after annealing at 800 °C for 50 h. As the cathode performance is closely related to both bulk oxygen-ion conductivity and oxygen surface exchange kinetics, exact relationships remain unclear between surface enrichment and overall cathode performance degradation. In addition, surface enrichment in most cases is partially driven by abundant oxygen vacancies that are also critical for efficient oxygen surface and bulk kinetics. This has been well exemplified by the volcano-shaped dependency observed by Tsvetkov et al. [23] in terms of the surface exchange kinetics of LSC-based thin films with the formation of oxygen vacancies in which although oxygen vacancies are key active sites for ORR, excessive concentrations of surface oxygen vacancies can degrade surface reactivity by accelerating Sr surface segregation. Moreover, surfaces predetermine the tolerance of cathodes against contaminants in gas phases and/or contacting fuel cell components [149]. More details are provided in the following subsections.

4.2 Susceptibility to Gas Contaminants

4.2.1 Water Vapor

Moisture is a common compound in ambient air and can adversely affect the durability of cathodes through the direct degradation of ORR and/or promotion of chromium [150,151,152] and silicon poisoning [153]. Based on this, this subsection will mainly focus on the direct effects of water vapor.

Nielsen, Hagen and Liu [154, 155] reported the strong dependency of water–vapor effects on moisture content, cathode polarization and temperature for LSM and LSCF cathodes in single-cell configurations in which for LSM-yttria-stabilized zirconia (YSZ) composite cathodes, the degradation rate of corresponding fuel cells increased with moisture content [155]. These researchers also found that cell performance degradation was more prominent at lower temperatures (e.g., 750 °C) under a high current load at 0.75 A cm⁻² [154] but that such degradation could partially be recovered if humid air was replaced with dry air.

Water vapor can also compete against O₂ for the limited surface oxygen vacancies of perovskite oxides [156,157,158,159]. For example, by applying operando ¹⁸O isotope exchange techniques and using deuterium oxide (D₂O) as the water vapor, Huang et al. [157] reported the existence of competitive adsorption between water and oxygen at the surface of La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ in which at 200–350 °C, increasing water vapor content promoted water surface exchange (Fig. 20a) but prohibited oxygen surface exchange (Fig. 20b). Alternatively, these researchers reported that this competitive adsorption phenomenon was not significant in (La_0.8Sr_0.2)_0.95MnO₃ oxide, indicating a higher tolerance for humidity as compared with La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ. This finding was consistent with results obtained by Liu et al. [160], who reported the higher tolerance of LSM as compared with LSCF in the presence of 10% humidity. In addition to degrading oxygen surface exchange rates, the competitive adsorption of water vapor can further influence oxidizing conditions and therefore the lattice geometry and electrochemical property of perovskite oxides. For example, Wang et al. [161] reported that water vapor absorption can prevent the oxidization of La_0.4Sr_0.6Co_0.2Fe_0.7Nb_0.1O_3−δ perovskite oxide, which consequently decreased lattice parameters, electrical conductivity and oxygen vacancy content. And because properties such as lattice geometry, conductivity and oxygen vacancy are dominant factors affecting surface cation segregation and ORR kinetics, the presence of water vapor can lead to complex degradation mechanisms through competitive surface adsorption with O₂.

Fig. 20

a D₂O exchange signal and b O₂ exchange signal over La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ as a function of temperature and water content [157]

4.2.2 CO₂ Poisoning

CO₂ is commonly present in air at concentrations of ~ 300 cm³ m^-3 and can strongly interact with defects such as oxygen vacancies and basic alkali and alkaline-earth element cations on the surface of perovskite oxides [13, 18, 162,163,164,165,166,167] due to acid–base interactions between acidic CO₂ and the basic nature of defects and alkali and alkaline-earth elements. As a result, even small amounts of CO₂ can compete against O₂ for limited surface reactive oxygen vacancies [168] and react with cations to form catalytically inert carbonates that can block active sites and adversely affect the crystal structure integrity of sub-surfaces [21]. Here, cathode performance degradation due to acid–base interactions can be recovered by lowering CO₂ partial pressures [13, 18, 169]; however, degradation due to carbonate formation is irreversible. In addition, the poisoning effects of CO₂ on cathode materials are more prominent at lower temperatures (< 650 °C) [158], likely arising from the increased stability of carbonate phases at reduced temperatures.

Cations are another factor that can determine anion stability. For example, Stern and Weise [170] reported that anion stability decreased with cation polarization power and suggested that cation polarization power can be used as a term to describe cation capability to distort anions and is associated with the size, charge and electronegativity of cations. In general, cations with strong polarization power possess (1) small size, (2) large positive charge and (3) high electronegativity that is close to the anion [171]. Based on this, Stern [172] proposed r^1/2/Z* (r is the cation size, Z* is the effective nuclear charge) to describe the polarization power and enthalpy of carbonate deposition at 298 K as a method to describe the thermal stability of carbonates. As a result, these researchers found a near-linear behavior between these two parameters (Fig. 21) in which relatively large-sized A-site cations were consistently more reactive with CO₂ than small-sized B-site cations and perovskite oxides containing cations such as Sr and Ba were more susceptible to CO₂ poisoning as compared with perovskite oxides containing Ca [173,174,175,176] or cations with higher oxidation states (Nb⁵⁺ [177, 178], Sb⁵⁺ [179, 180], Ta⁵⁺ [181, 182], Ti⁴⁺ [183] and Cr⁶⁺ [184]).

Fig. 21

Decomposition enthalpy of various carbonates against cationic property r^1/2/Z* at 298 K in which r denotes cation radius and Z* is the effective nuclear charge calculated from Slater’s rules [185]. Reproduced with permission from Ref. [172]. Copyright 1972, American Institute of Physics

Due to strong interactions between CO₂ and A-site cations such as Sr and Ba, the lower surface coverage of those reactive cations can improve the CO₂ tolerance of perovskite oxides. The surface coverage of these cations is associated with surface segregation (Sect. 2), and the weakening of the two main driving forces (i.e., elastic and electrostatic interactions) can help to suppress surface segregation and therefore limit exposure to CO₂. Based on this, perovskite properties such as cation size, stoichiometry, crystallinity and lattice strain and cathode operating conditions such as polarization can all affect the tolerance of cathodes against CO₂ poisoning.

Based on the strong dependency of anion stability on cation polarization power, a conclusion can be made that cations possessing strong interactions with oxygen ions should also possess strong interactions with carbonate ions in which the former can affect the stability of perovskite oxides, whereas the latter can determine reactivity with CO₂, meaning that CO₂-tolerant cathodes should contain cations with optimal interactions with O²⁻ and CO₃²⁻ (i.e., strong bonding with O²⁻ but weak bonding with CO₃²⁻). Here, interactions between metal and oxygen can be described by using average metal–oxygen energy (ABE) and an example of calculating the ABE of an ABO₃ perovskite oxide is provided by Eq. (12) [186] in which $\Delta H$ denotes formation enthalpy for A_mO_n, B_xO_y and sublimation enthalpy of A or B metals and ${D}_{({\mathrm{O}}_{2})}$ is the dissociation energy of O₂. Possibly due to the balance of cation interactions with oxygen and carbonate, more negative ABE values strengthen metal–oxygen interactions and therefore indicate cathodes that are more CO₂ tolerant [21, 165]. For example, by varying the ratio of Co to Fe at B-sites, Zhang et al. [165] reported that the thermal stability of adsorbed CO₂ species at their perovskite surface decreased with more negative ABEs (Fig. 22).

$$ {\text{ABE = }}\frac{1}{12}\left( {m \left( {\Delta H_{{{\text{A}}_{m} {\text{O}}_{n} }} } \right) - m\Delta H_{{\text{A}}} - \frac{n}{2}D_{{{{\text{O}}_{2} }}} } \right) + \frac{1}{6}x\left( {\Delta H_{{{\text{B}}_{x} {\text{O}}_{y} }} - x\Delta H_{{\text{B}}} - \frac{y}{2}D_{{ {{\text{O}}_{2} } }} } \right) $$

(12)

in which ${\Delta H}_{\mathrm{A}{\left(\mathrm{B}\right)}_{m}{\text{O}}_{n}}$, ${\Delta H}_{\mathrm{A}\left(\mathrm{B}\right)}$ and ${D}_{{\mathrm{O}}_{2}}$ represent the formation enthalpy of A(B)_mO_n oxide, sublimation enthalpy of A(B) metal at 25 °C and dissociation energy of oxygen (${D}_{{\mathrm{O}}_{2}}=500.2 \, \mathrm{kJ}\cdot {\mathrm{mol}}^{-1}$), respectively.

Fig. 22

(a) Comparison of calculated average metal–oxygen bond energy and (b) CO₂ temperature-programmed desorption profiles of SrNb_0.1Co_0.9−xFe_xO_3−δ oxides. Reproduced with permission from Ref. [165]. Copyright 2015, Elsevier

4.2.3 SO₂ Poisoning

Trace amounts of SO₂ (e.g., in the 1×10⁻³ cm³ m⁻³ level) present in air can exert adverse effects (i.e., increased ohmic and polarization resistances) on the long-term stability of cathodes [187, 188]. Similar to CO₂, SO₂ is also an acidic gas that is reactive to basic cations such as Sr and Ba to form sulfate species on the surface. The formation of sulfates is an oxidization reaction (i.e., S⁴⁺ needs to be oxidized to S⁶⁺) as represented by Eq. (13) that can easily occur at high SO₂ concentrations on cathode surfaces with abundant basic terminated AO phases, oxygen vacancies and dissociated oxygen species. For example, Wang et al. [189] reported that increasing SO₂ concentrations from 0.1 cm³ m⁻³ to 100 cm³ m⁻³ led to the more pronounced cathode degradation of La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ and that high SO₂ levels led to the further formation of La₂O₂SO₄ in addition to SrSO₄. These researchers also compared the stability of (La, Sr)Co_0.2Fe_0.8O_3−δ cathodes with varying La/Sr ratios in the presence of 1 cm³ m⁻³ SO₂ at 800 °C and found higher coverage of the SrSO₄ phase over the Sr-rich perovskite oxide as compared with La-rich samples as well as immediate poisoning effects with the introduction of SO₂ [190]. Here, this trend can be explained by the promoted Sr segregation and oxygen vacancy as imparted by increasing divalent and large-sized Sr content.

$$\mathrm{AO}+{\mathrm{SO}}_{2}+\frac{1}{2}{\mathrm{O}}_{2}\left(\mathrm{g}\right)\to \mathrm{AS}{\mathrm{O}}_{4}$$

(13)

Cathode/electrolyte interfaces can also play a role in SO₂ tolerance. For example, De Vero et al. [191] reported the improved SO₂ tolerance of La_0.6Sr_0.4Co_0.2Fe_3−δ thin film after annealing at 800 °C for 300 h over a BaZr_0.8Y_0.2O_3−δ (BZY) electrolyte substrate as compared with over a Gd_0.1Ce_0.9O_1.95 (GDC) electrolyte substrate in which these researchers observed the significant interdiffusion of cations across the cathode/BZY interface (over the 1-µm-thick interdiffusion zone) but not across the cathode/GDC interface (Fig. 23a, b). Here, these researchers proposed that significant cation interdiffusion with BZY can weaken the thermodynamic driving force of Sr surface segregation and therefore hinder further SrSO₄ formation with SO₂ (Fig. 23 d), whereas due to limited Sr interdiffusion into GDC, Sr species in the corresponding cathode can only diffuse to the cathode/gas interface, which consequently promoted unwanted SrSO₄ precipitation (Fig. 23c). Based on this, these researchers postulated that easy Sr interdiffusion between cathodes and electrolytes is likely related to the similar lattice symmetries (i.e., perovskite structures for cathodes and BZY) of the cathode and the electrolyte, which can reduce cation migration energy barriers across different phases. This easy cation interdiffusion has also been commonly reported for materials with similar structures by other researchers [192,193,194,195].

Fig. 23

Scanning transmission electron microscopy bright-field images and elemental distributions of the cross sections of La_0.6Sr_0.4Co_0.2Fe_3−δ thin films over a BZY and c GDC electrolytes. Schematics of the proposed mechanisms of La_0.6Sr_0.4Co_0.2Fe_3−δ interaction with b BZY and d GDC and their role in SO₂ poisoning. Reproduced with permission from Ref. [191]. Copyright 2018, Materials Research Society. In (a), label 1 denotes the SrSO₄ phase, label 2 denotes nanopores, label 3 denotes the interdiffusion region rich in Y and Sr, and label 4 denotes the other interdiffusion region rich in La, Co and Fe

4.3 Susceptibility to Volatile Contaminants

Cathode materials are also susceptible to volatile contaminants such as Cr, Si and B-based species from fuel cell components (e.g., interconnects, balance of plants, glass sealants). Although Cr, Si and B play essential roles in the tailoring of fuel cell component properties to tolerate SOFC operating conditions, they can become volatile at high temperature and humidity conditions. Therefore, one of the main motivations to lower SOFC operating temperatures is to suppress the vaporization of these elements and achieve improved long-term stability of the electrodes.

4.3.1 Chromium Poisoning

Chromium is an important element that can prevent the oxidization of metallic alloys at SOFC operating temperatures through the formation of Cr₂O₃ layers on alloy surfaces. However, chromium can also rapidly degrade cathode performance if evaporated as highly oxidized (i.e., Cr⁶⁺) volatile species such as CrO₂(OH)₂ and CrO₃. Aside from temperature, the volatility of Cr₂O₃ is also related to water vapor and oxygen content in gases in which Gindorf et al. [196] reported that the vaporization of CrO₂(OH)₂ increased with increasing humidity in air at 950 °C. Water vapor partial pressure in air can also affect Cr volatile species vaporization from Cr₂O₃ in which CrO₃ (g) is dominant at low water vapor pressures (Eq. 14), whereas CrO₂(OH)₂ is dominant at high water vapor pressures (Eq. 15):

$${\mathrm{Cr}}_{2}{\mathrm{O}}_{3}\left(\mathrm{S}\right)+1.5{\mathrm{O}}_{2}\left(\mathrm{g}\right)\to 2{\mathrm{CrO}}_{3}\left(\mathrm{g}\right)$$

(14)

$${\mathrm{CrO}}_{3}\left(\mathrm{g}\right)+1.5{\mathrm{O}}_{2}\left(\mathrm{g}\right)+2{\mathrm{H}}_{2}\mathrm{O}(\mathrm{g})\to 2{\mathrm{CrO}}_{2}{(\mathrm{OH})}_{2}\left(\mathrm{g}\right)$$

(15)

In general, the nucleation theory can be used to explain the deposition of Cr on LSM and LSCF surfaces [197,198,199] and involve the deposition of volatile Cr species through chemical reactions between high-valent Cr and nucleation agents such as Mn, Co and Sr to form Cr₂O₃, (Cr, Mn)₃O₄, SrCrO₃ or CoCr₂O₄ on cathode surfaces. Here, the general reaction routes as summarized by Jiang and Chen [197] can be represented by Eqns. (16–18) in which N denotes Mn, Co and/or SrO species at cathode surfaces and indicate that Cr deposition kinetics is highly dependent on cathode surface chemistry, which is dominated by factors discussed in Sect. 2.

$${\mathrm{CrO}}_{3}\left(\mathrm{g}\right)+\mathrm{N}\left(\mathrm{s}\right)\to {\mathrm{Cr}-\mathrm{N}-\mathrm{O}}_{(\mathrm{nuclei})}\left(\mathrm{s}\right)+{\mathrm{O}}_{2}\left(\mathrm{g}\right)$$

(16)

$${\mathrm{Cr}-\mathrm{N}-\mathrm{O}}_{(\mathrm{nuclei})}\left(\mathrm{s}\right)+{\mathrm{CrO}}_{3}\left(\mathrm{g}\right)\to {{\text{Cr}}_{2}\mathrm{O}}_{3}\left(\mathrm{s}\right)$$

(17)

$${\mathrm{Cr}-\mathrm{N}-\mathrm{O}}_{(\mathrm{nuclei})}\left(\mathrm{s}\right)+\mathrm{N}+{\mathrm{CrO}}_{3}\left(\mathrm{g}\right)\to {\mathrm{N}-\mathrm{Cr}-\mathrm{O}}_{x}\left(\mathrm{s}\right)$$

(18)

4.3.2 Silicon Poisoning

Glass-based sealants used in the manufacturing of SOFCs contain silicon-based materials such as boroaluminosilicate, borosilicate, aluminosilicate and silicate [200]. In addition, silicon species are present in alloys used for high-temperature operations [201, 202]. Similar to Cr, Si species can easily generate Si(OH)₄ vapor at elevated temperatures, particularly with increasing water vapor pressure [203], the deposition of which on cathode surfaces can significantly degrade cathode performance by blocking oxygen-ion transport [202] and promoting perovskite phase decomposition at the near-surface region [204, 205]. For example, Perz et al. [204] observed significant performance degradation of 5.9 times in La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ with a cathode surface coated with 10-nm-thick SiO₂ at 700 °C for 1340 h as compared with a cathode surface without SiO₂ coating (degradation of 4.5 times in the same annealing conditions). Such degradation is related to the formation of a continuous layer of La-Sr-Si–O species on the surface and Co-Fe–O nanoparticles as promoted by SiO₂ deposition. Bucher et al. [206] also observed the decomposition of surface phases over La_0.6Sr_0.4CoO_3−δ oxides due to Si species deposition and highlighted the role of water vapor (relatively humidity = 30%–60%) in exacerbating Si poisoning effects in which such surface phase decompositions can degrade oxygen surface exchange kinetics and bulk diffusivities and therefore degrade cathode performance [206, 207].

4.3.3 Boron Poisoning

Boron is commonly used to tune the viscosity and soften the temperature of glass sealants [208,209,210]. Similar to Cr and Si contaminants however, borates become volatile at high temperatures and water pressures [211, 212] in which Zhang et al. [212] reported that borates mainly vaporize in the form of BO₂ in dry atmosphere but B₃H₃O₆ in humid atmosphere. Here, the deposition of volatile borate species can degrade the performance of perovskite oxides such as LSM and LSCF-based materials by inducing surface phase decomposition and accelerating particle coarsening. Based on this, Chen et al. comprehensively studied the effects of boron poisoning on LSM- [213], LSCF- [214] and BSCF-based [215] electrodes by annealing perovskite oxides in the presence of borosilicate glass at 700–800 °C for 7–30 days and found that volatile boron was highly reactive with surface La species to form LaBO₃, which led to the destruction of the perovskite structures. Alternatively, these researchers reported that La-free BSCF showed much-improved tolerances toward surface boron deposition, but that boron can promote A-site cation (e.g., Ba and Sr) segregation and alter the microstructure of cathodes [215]. These researchers further reported that deposited boron can serve as a sintering agent to cause the severe particle coarsening of infiltrated Gd-doped ceria nanoparticles in LSM composite cathodes and significantly degrade overall cathode performance [216].

5 Design Strategies to Develop Perovskite Cathode Materials

At SOFC operating conditions, both the surface and bulk properties of cathode materials are essential for efficient and stable ORR activity. In addition, surface and bulk interactions such as surface cation segregation are significant for the catalytic activity and long-term stability of corresponding cathodes. Considering the importance of cathode surface and bulk properties in ORR catalysis, cathode performance can be enhanced through several methods, including (1) optimizing cathode composition and structures and (2) incorporating additional functional phases. As such, this section will mainly focus on reviewing recent advancements in cathode development through doping (Sect. 5.1) and surface functionalization (Sect. 5.2).

5.1 Doping Strategy

The structural and compositional flexibility of perovskite oxides enable the tuning of corresponding bulk and surface properties through the incorporation of heterogeneous atoms at A-sites and B-sites (Table 1). Here, common doping methods in cathode development include (1) solid-state reactions and (2) sol–gel methods in which the solid-state reaction route involves the mechanical mixing of metal precursors (e.g., metal oxides or carbonates), the pelletizing of powder mixtures and long-term (more than 10 h) calcination at high temperatures (e.g., 1200 °C) to enable sufficient interdiffusion between constituent ions. Despite its simplicity and versatility in perovskite synthesis, the long-term sintering procedures of solid-state reactions have limited application in obtaining products with well-controlled particle sizes. Alternatively, the sol–gel route (the so-called Pechini method) [217] requires relatively low temperatures (normally below 1000 °C) to calcine metal precursors and typically involves dissolving water-soluble metal precursors such as nitrates, complexation with agents such as EDTA and citric acid, gel formation through water evaporation, firing to form char and calcination at elevated temperatures to obtain targeted oxides. As compared with solid-state reactions, the sol–gel reactions are more uniform dispersions of metal precursors than mixtures of solids, meaning that the sol–gel method requires lower calcination temperatures and shorter calcination periods. Despite this, a major limitation of the sol–gel method is the lack of water-soluble salts such as Ta-containing precursors. Nevertheless, the sol–gel method is preferred for perovskite synthesis due to low calcination temperatures and easy control of synthesis conditions [218].

Table 1 Summary of doping strategies at A-site, B-site and A-/B-sites in typical perovskites reported in the literature

5.1.1 A-Site Doping

A-site cations in perovskites typically possess relatively large ionic sizes as compared with B-site cations and are coordinated with 12 oxygen atoms in the lattice. Common A-site cations include lanthanide elements (e.g., La, Pr, Nd, Sm and Gd) [231] and alkaline-earth elements (e.g., Ca, Sr and Ba) [232]. A-site cations can further be partially occupied by alkali metals (e.g., Li, Na and K) [18] and post-transition metals such as Bi metals [233]. And although most A-site cations are redox inactive and are not directly involved in ORR, they play an essential role in determining cathode performance and long-term stability. For example, A-site cations due to their relatively large ionic size can influence lattice spacing for B-site cations that normally sit within voids formed by A-site cations and therefore can affect the overall oxidation state of B-site cations. Perovskite surfaces are also usually enriched with A-site cations (Sect. 2 and 4) that can reduce the number of active sites for oxygen surface exchange and are prone to react with gaseous and volatile contaminants such as CO₂ and Cr species. The migration of A-site cations from bulk to surface at elevated temperatures can further dynamically change the bulk and surface properties of cathodes. Considering these effects, the rational modulation of perovskite oxide properties through A-site doping can allow for optimized cathode catalytic activity and stability.

5.1.1.1 A-Site Cation Size

The ionic size of A-site cations determines tolerance factor (τ) (Eq. 10) and thus affect perovskite structural stability [100, 234]. In addition, perovskite structures with high symmetry such as simple cubic perovskite structures generally exhibit higher mixed ionic and electronic conductivity (Sect. 3 and 4) in which high mixed conductivity is important for efficient ORR activity, especially at reduced operating temperatures (Sect. 3.2 and 3.3). Therefore, the consideration of A-site cation effects on structural symmetry is important. One typical example is our recent work [19] involving the incorporation of Li into Co-free SrFe_0.8Nb_0.1Ta_0.1O_3−δ perovskite oxide in which because Li (~ 0.92 Å) is much smaller than host Sr (1.44 Å), less than 5% of Li can be incorporated into the lattice to maintain the beneficial cubic structure of the perovskite. As a result, the stabilized cubic structure of Sr_1−xLi_xFe_0.8Nb_0.1Ta_0.1O_3−δ with x < 0.05 together with optimal levels of oxygen vacancies and A-site deficiencies as imparted by Li doping led to improved ORR activities at 600 °C by ~ 1.6 times.

5.1.1.2 A-Site Cation Surface Segregation

A-site dopant/host ionic size mismatch constitutes a major driving force (i.e., elastic energy minimization) for A-site cation surface segregation in SOFC operating conditions in which in most cases, surface-enriched A-site cation species pose negative impacts on cathode performance and long-term stability (Sects. 2, 4). Because of this, the minimization of dopant/host size mismatch can suppress unwanted surface cation segregation [235]. For example, Pang et al. [220] highlighted the role of A-site cation size mismatch in determining ORR activity and stability for double perovskite PrBa_1−xCa_xCoCuO_5+δ and reported that relatively low A-site cation size mismatch as achieved by PrBa_0.7Ca_0.3CoCuO_5+δ led to an improved ORR activity of 0.052 Ω cm² at 650 °C as compared with the undoped sample (0.072 Ω cm²) and a peak power density of 2.04 W cm⁻² at 800 °C. These researchers also found that the incorporation of Ca can effective suppress Ba surface enrichment (Fig. 24) during 100-h stability testing under 0.7 V at 700 °C. As a result, a single-cell using PrBa_0.7Ca_0.3CoCuO_5+δ as the cathode showed better stability (the degradation rate = 0.053% h⁻¹) than a cell using the parent oxide as the cathode (the degradation rate = 0.178% h⁻¹). And because surface-enriched A-site cations such as Sr are susceptible to contaminants such as water, vapor, CO₂ and SO₂ (Sects. 4.2 and 4.3), the suppression of surface segregation through the minimization of A-site dopant/host size mismatch can also effectively improve cathode tolerance toward these contaminants.

Fig. 24

a Energy-dispersive X-ray (EDX) images of PrBa_1−xCa_xCoCuO_5+δ, including Pr, Ba, Co, Cu and O elements. Metal cation concentrations on the surface of PrBa_1−xCa_xCoCuO_5+δ with Ca contents of b x = 0 and c x = 0.3 as calculated from X-ray photoelectron spectroscopy results. Reproduced with permission from Ref. [220]. Copyright 2020, Elsevier

Alternatively, A-site cation surface segregation can create A-site deficiencies in lattices and induce the formation of oxygen vacancies to compensate charge imbalance in which A-site deficiencies and oxygen vacancies can enhance electronic and ionic conductivities [236]. For example, Celikbilek et al. [221] recently designed A-site deficiency by manipulating the composition of (La_0.71Sr_0.29)_0.95Co_0.17Fe_0.83O_3−δ. By introducing 5 mol.% A-site deficiency, the authors observed a low polarization resistance of 0.037 Ω cm² at 650 °C and 0.1 Ω cm² at 600 °C in a symmetrical cell, and a peak power densities of 1.4 W cm⁻² at 650 °C and 1.0 W cm⁻² at 600 °C in a Ni/YSZ anode-supported cell were obtained, which were much better than values obtained from a pristine La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ analogue. Furthermore, if enriched A-site cations such as alkali metals or noble metals do not passivate electrochemically active surfaces, A-site cation surface segregation can promote ORR activity. This was well exemplified by our recent work on alkali metal-doped SrFe_0.8Nb_0.1Ta_0.1O_3−δ [18, 19] in which significant size mismatch between the Li dopant and the Sr host allowed for the easy migration of the Li dopant from the bulk to the cathode surface, leaving a bulk with increased A-site deficiency and oxygen vacancy. Interestingly, alkali metal segregation can also suppress problematic Sr surface enrichment to further contribute to improved ORR activity and contaminant tolerance. Similarly, Zhu et al. [222] slightly reduced Sr_0.95Ag_0.05Nb_0.1Co_0.9O_3−δ with hydrogen to induce Ag exsolution from the bulk to the surface (Fig. 25) and created A-site cation deficiency in the bulk and ORR-active Ag nanoparticles on the surface to result in a remarkably low polarization resistance of 0.214 Ω cm² at 500 °C. Similar improvements resulting from A-site cation surface segregation were also reported for an Ag-doped LSM cathode [223].

Fig. 25

a High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of the e-SANC material with b corresponding liner scan file from A to B. c HAADF-STEM image of the s-SANC material with d EDX spectra at P1 and P2 positions. Reproduced with permission from Ref. [222]. Copyright 2015, American Chemical Society

Enriched A-site dopants can further functionalize cathode surfaces to resist contaminant poisoning in which our recent findings [18] highlighted alkali metals such as Na and K that can form peroxide or sesquioxide species in cathode operating conditions that can react with CO₂ to release oxygen molecules as a product and therefore increase oxygen partial pressures close to the cathode surface (Fig. 26c). Moreover, these formed alkali metal carbonates at the cathode surface are good oxygen-ion conductors that can accelerate oxygen surface exchange in the presence of CO₂ and constituted one of the major reasons for remarkable cathode ORR activity and stability improvements at 600 °C in the presence of CO₂, showing area-specific resistance (ASR) values as low as 0.12 Ω cm² at 600 °C and over fourfold improvement in tolerance to CO₂ poisoning (Fig. 26a, b).

Fig. 26

Impedance spectra of a SrFe_0.8Nb_0.1Ta_0.1O_3−δ and b Sr_0.95A_0.05Fe_0.8Nb_0.1Ta_0.1O_3−δ measured in air, 10% CO₂ balanced air and the subsequent removal of CO₂ after 2 h. c Mechanism of oxygen cloud generation that enhanced the CO₂ tolerance of the Sr_0.95A_0.05Fe_0.8Nb_0.1Ta_0.1O_3−δ cathode. Reproduced with permission from Ref. [18]. Copyright 2019, American Chemical Society

5.1.1.3 Polarization Power of A-Site Cations

Cation polarization power reflects the capability of cations to stabilize anions (Sect. 4.2) in which large cations with low oxidation states generally possess weaker polarization power and strong bonds with anions such as oxygen or carbonate anions. And because A-site cations (e.g., Sr and Ba) are larger in size and have lower oxidation states than B-site cations, they are more prone to poisoning from gaseous contaminants such as CO₂ than B-site cations. Based on this, increases in A-site cation oxidation state (e.g., replacing alkaline-earth elements with lanthanides) or reductions in A-site cation size (e.g., replacing Ba or Sr with Ca) can improve the durability of corresponding cathodes in the presence of gaseous contaminants. For example, Chen et al. [224] developed a PrBa_0.8Ca_0.2Co₂O_5+δ double perovskite that exhibited stable ORR activity with ASR values of ~ 0.024–0.028 Ω cm² at 750 °C in the presence of 1 vol.% CO₂ for over 1000 h and attributed the improved durability and ORR activity to the Pr³⁺ and Ca²⁺ cations at the A-sites (high oxidation state and/or small size). Other rare-earth elements such as Eu [237], Nd and Sm [238] with high polarization powers have also been reported to be able to destabilize bonding with anions such as sulfates and carbonates to improve the contaminant tolerance of corresponding cathodes.

5.1.1.4 Inducing Effects on B-Site Cations

Doping A-site cations can also impact B-site cations in perovskite oxides. A common strategy to introduce oxygen vacancies into perovskite oxides is to reduce the overall positive charge of lattices through the partial replacement of high-valence A-site cations such as La³⁺ with relatively low oxidation state dopants such as Sr²⁺ [10, 239, 240], Ba²⁺ [234, 241] and alkali metals [18, 242]. Such induced charge imbalance can be compensated by either increasing the oxidation state of B-site cations or creating oxygen vacancies. The response of perovskite oxides to the induced charge imbalance likely depends on the ease of B-site cation redox transition. For B-site cations with easy redox transitions such as Mn [243, 244] and Fe [245, 246], increases in B-site cation oxidation state are preferred and will generally lead to increased electronic conductivity due to the increased density of redox couples as charge hopping sites. As for B-site cations that are difficult to oxidize such as Co and Ni, induced charge imbalances can lead to the formation of oxygen vacancies that can enhance ionic conductivity. Increasing the size of A-site cations can also facilitate the expansion of B-site cations (i.e., be slightly reduced) to further create oxygen vacancies with a typical example being benchmark cathode BSCF in which large-sized Ba can promote oxygen vacancy formation [247]. Interestingly, although A-site cations with lower oxidation states and large ionic sizes can enhance mixed conductivity, they can also lead to weak polarization powers that increase susceptibility toward contaminants. Therefore, an optimal balance between activity and stability needs to be considered in the selection of A-site cations.

The electronegativity of A-site dopants can also affect the oxidation state of B-site cations in which relatively low electronegativities can weaken the capability of neighboring oxygen to attract electron clouds from adjacent B-site cations and slightly reduce B-site valences to promote oxygen vacancy [18] as confirmed by our recent investigation into the effects of alkali metals (i.e., Li, Na and K) on Fe valence in doped SrFe_0.8Nb_0.1Ta_0.1O_3−δ oxide. In this study, monovalent alkali metal doping not only promoted oxygen vacancies, but also reduced the average valence of B-site Fe cations in which we believed the lower electronegativity of Li (0.99), Na (0.98) and K (0.93) as compared with host Sr (1.03) [248] was the main reason for Fe valence state reduction.

5.1.2 B-Site Doping

As main ORR active centers, B-site cations play an essential role in ORR catalysis. In general, typical B-site cations in perovskite oxides are six-coordinated transition metals that include Mn [249, 250], Fe [245, 251], Ni [252] and Co [253, 254]. B-sites can also be partially occupied by a variety of elements while maintaining perovskite structural integrity, allowing for the tailoring of perovskite properties through B-site doping. Similar to A-site cations, B-site properties such as cation size, oxidation state and electronegativity can all influence the overall performance of perovskite cathodes.

5.1.2.1 Selection of B-Site Dopant Size

The selection of suitable B-site dopant size is a prerequisite for successful doping in which B-site dopants should generally be comparable in size to B-site host cations to sustain perovskite structures. Here, the Shannon ionic radii of common host cations such as Mn, Fe, Co and Ni are within 0.4–0.8 Å and can serve as a criterion for the selection of potential B-site dopants. For example, Ling et al. [225] reported that the partial replacement of Fe with Sb³⁺ in Ba_0.5Sr_0.5Fe_1−xSb_xO_3−δ (x = 0.0, 0.05, and 0.1) perovskite can stabilize the beneficial cubic perovskite structure by reducing the tolerance factor close to unity, resulting in the Ba_0.5Sr_0.5Fe_0.9Sb_0.1O_3−δ displaying significantly lowered polarization resistances from 0.213 Ω cm² down to 0.120 Ω cm² at 700 °C.

The ionic size of B-site dopants can also affect specific free volume and critical saddle point with large sizes leading to facile oxygen-ion transport (Sect. 3.1). However, the ionic size of redox-active B-site dopants can vary with oxidation state depending on A-site cations and operating conditions (e.g., temperature and gas atmosphere) and oxidation state is strongly correlated with oxygen vacancy and mobility. As a result, the understanding of the specific effects of B-site cations on saddle point free volume and size through experiments is challenging.

5.1.2.2 Doping with High Oxidation State B-Site Dopants

The oxidation state of B-site cations is an important factor governing cathode ORR activity and long-term stability, and numerous studies have shown that doping with high oxidation state cations such as W⁶⁺ [255], Mo⁶⁺ [256], Sb⁵⁺ [257], Nb⁵⁺ [118, 258], Ta⁵⁺ [13, 117], Zr⁴⁺ and V⁵⁺ [259] can effectively enhance cathode performance. According to Pauling’s rules [260], high-valence B-site cations with low coordination numbers can increase electrostatic repulsion between polyhedra and stabilize high symmetry perovskite structures. Nagai and Sakon [261] also studied the effects of dopants including Ni, Cu, Zn, Cr, Fe, Al, Ga, In, Ce, Ti, Zr, Sn, V and Nb on the structural stability of SrCoO_3−δ and found that high-valence cations with higher solubility in lattices such as Nb⁵⁺ and Ti⁴⁺ were more effective in stabilizing the cubic structure than low-valence dopants. Interestingly, researchers have also reported that non-metals with high valences and suitable ionic sizes such as P⁵⁺ [262], S⁶⁺ [263, 264] and Si⁴⁺ [265] can also be applied as B-site dopants to enhance the structural stability and ORR activity of materials. We have further reported that P cation doping at the B-sites of SrCo_0.85Fe_0.1P_0.05O_3−δ can stabilize the cubic perovskite structure and prevent oxygen-vacancy ordering brownmillerite-type structure formation [119], resulting in the P-doped cathode exhibiting a low polarization resistance of 0.097 Ω cm² at 589 °C and no performance degradation at 600 °C for 40 h. Similar effects of these non-metal dopants were also reported by Xu et al. [266] and Luo et al. [267]

Despite their role in stabilizing structural stability, high-valence B-site dopants can reduce oxygen vacancy content in lattices due to increases in overall cation positive charge. And because oxygen vacancies are major carriers for oxygen-ion conduction (Sect. 3), high-valence B-site doping can have negative impacts on oxygen kinetics at both the surface and the bulk lattice. One method to alleviate this negative effect is to slightly lower the electronegativity of high-valence dopants. For example, one of our studies investigated the effects of Nb⁵⁺ and Ta⁵⁺ electronegativity by constraining potential effects from structural geometry and the oxidation state based on the fact that although Nb⁵⁺ and Ta⁵⁺ possessed similar sizes with the same Shannon ionic radius of 0.64 Å, they possessed different electronegativity (Ta⁵⁺ is slightly lower than Nb⁵⁺ in electronegativity) [118]. In this study, both experimental results and theoretical calculations revealed that the low electronegativity of the Ta⁵⁺ dopant can move electron clouds closer to neighboring Co centers to result in relatively low oxidation state Co and therefore slightly higher oxygen vacancy content, which led to enhanced oxygen surface and bulk kinetics (Fig. 27a, b) as well as cathode performance as compared with the Nb-doped counterpart. Similarly, our recent results [138] revealed that Sc³⁺ and Ta⁵⁺ co-doped SrCoO_3−δ perovskite can increase oxygen vacancy content as compared with Sc³⁺ and Nb⁵⁺ co-doped analogues to result in a notably lower ASR of 0.233 Ω cm² at 500 °C.

Fig. 27

a Oxygen surface exchange coefficient (k) and b oxygen-ion diffusivity (D) as a function of temperature for SrCo_0.8Nb_0.2O_3−δ (SCN20) and SrCo_0.8Ta_0.2O_3−δ (SCT20). Reproduced with permission from Ref. [118]. Copyright 2015, Royal Society of Chemistry

Reduced oxygen vacancy content should also alleviate unwanted A-site cation segregation by weakening electrostatic interactions between A-site cations and oxygen vacancies. This effect has recently been confirmed by Tsvetkov et al. [23] (Sect. 2.1 and Fig. 5). And because A-site cation segregation is a major reason for cathode susceptibility to contamination (Sect. 4), high-valence dopants can further enhance resistance to poisoning. Zhao et al. [227] also highlighted the role of Nb⁵⁺ doping in Sm_0.5(1−x)Sr_0.5(1−x)Ba_xCo_1−0.3xFe_0.2xNb_0.1xO_3−δ (x = 0.4, 0.5, and 0.6) in the suppression of surface Ba segregation if in contact with Fe–Cr metallic interconnects under 200 mA cm⁻² at 900 °C for 20 h in which suppressed Ba segregation can significantly improve cathode tolerance to Cr contamination.

Although it is generally unlikely that high-valence dopants are directly involved in oxygen-ion transport, dopants can affect neighboring redox-active B-site cations. For example, Zhou et al. [11] developed a Sc³⁺ and Nb⁵⁺ co-doped SrSc_0.175Nb_0.025Co_0.8O_3−δ (SSNC) perovskite that showed a remarkably low ASR value of 0.95 Ω cm² in a symmetrical cell at 450 °C and 0.91 W cm⁻² in a Ni-SDC|SDC based single cell at 550 °C. Here, DFT calculations revealed that oxygen vacancy formation energy was 1.007 eV for Sr₄Co₂, 2.164 eV for Sr₄CoNb and 1.906 eV for Sr₄CoSc clusters, highlighting the negative effects of Nb⁵⁺ on oxygen vacancy formation. Alternatively, these researchers found that oxygen migration energy barriers were significantly reduced within Sr₄Co₂ clusters close to Nb dopants (Fig. 28a-d), suggesting that improved lattice oxygen migration is a main contributor to enhanced ORR activity. These results also highlighted the fact that easy oxygen vacancy formation does not necessarily represent high ORR activity, especially at reduced operating temperatures. Li et al. [12] also studied Nb⁵⁺ and Ta⁵⁺ co-doped SrCo_0.8Nb_0.1Ta_0.1O_3−δ (SCNT) perovskite that showed low ASR values of ~ 0.16 Ω cm² at 500 °C and ~ 0.68 Ω cm² at 450 °C in which DFT calculations revealed that oxygen could only migrate along Co sites in the lattice (Fig. 28f). These researchers further reported that Nb⁵⁺ doping can enhance the density of state at the Fermi level of the next nearest Co to the dopant, which can contribute to facile charge transfer to adsorbed oxygen species. And together with the beneficial effects of Ta⁵⁺ in oxygen vacancy formation, the enhanced oxygen mobility and charge transfer as a result of Nb⁵⁺ doping allowed for an optimal balance between oxygen vacancy, oxygen-ion mobility and surface charge transfer in the resulting cathode.

Fig. 28

Schematics of oxygen vacancy migration from a V_O(Sr₄Co₂), b V_O(Sr₄CoNb) and c V_O(Sr₄CoSc) in the d SSNC perovskite structure. MB represents oxygen vacancy migration barrier. Reproduced with permission from Ref. [11]. Copyright 2013, Wiley–VCH. e Oxygen ionic conductivities of SCN20, SCT20 and SCNT from 475 to 600 °C. f Schematic of oxygen vacancy migration with minimum energy [12]

Although volatile Si species can cause ORR degradation (Sect. 4.3), Si doping in perovskite lattices can effectively improve cathode performance due to benefits from its high valence. For example, Porras-Vazquez et al. [228] found that Si doping into a Sr_0.75Ca_0.25Fe_0.85Si_0.15O_3−δ perovskite cathode can allow for a lower polarization resistance of 0.51 Ω cm² than undoped Sr_0.75Ca_0.25FeO_3−δ (0.91 Ω cm²) at 700 °C and that doping with high-valence Si⁴⁺ can reduce the average valence of Fe to lower values to introduce more oxygen vacancies close to Fe ions. These researchers also highlighted the role of cathode surfaces and bulk in determining the effects of particular elements on cathode performance in which Si interaction with surface-enriched A-site cations can result in the formation of unwanted phases that block active sites for oxygen surface exchange, whereas high-valence Si doping into cathode bulk can optimize neighboring B-site hosts to enhance cathode performance.

5.1.2.3 Modulation of ABE

B-site doping can also impact average metal–oxygen bonding energy (Eq. 12) in which the strengthening of average metal–oxygen interactions can effectively improve perovskite tolerance to contaminants such as CO₂ (Sect. 4.2) [21, 165]. For example, the replacement of Co with Fe can significantly shift ABE to more negative values (Fig. 22). Gao et al. [230] also evaluated the effects of Ti⁴⁺ doping on the structural and electrocatalytic properties of Fe-based Bi_0.5Sr_0.5Fe_1−xTi_xO_3−δ (x = 0.05–0.20) cathodes and found by increasing CO₂ content from 5 mol.% to 15 mol.%, the overall ABE value of the sample decreased from − 282.05 kJ mol⁻¹ to − 292.79 kJ mol⁻¹. These researchers also reported that the strengthened metal–oxygen bonding and increased overall acidity as imparted by Ti⁴⁺ doping did not lead to the formation of carbonates after a BSFT0.15 cathode was treated at 700 °C for 12 h in the presence of 15% CO₂.

5.2 Surface Functionalization

The surface of perovskite oxide cathodes is an important factor determining oxygen surface exchange rate and contaminant tolerance and can potentially affect bulk properties (Sect. 2). And although the modulation of material property and operating condition through doping can indirectly modify cathode surface chemistry and microstructure, surface functionalization is a more direct method. Table 2 provides a summary of discussed strategies for cathode surface functionalization.

Table 2 Summary of cathode surface functionalization strategies

5.2.1 Infiltration

Infiltration (also called wet impregnation) is a popular technique to prepare cathodes and modify cathode surfaces (Fig. 29). In typical infiltration processes, the substrate is normally a porous backbone bonded with the electrolyte (Fig. 29a), in which during infiltration, a liquid solution is introduced into the substrate through coating techniques such as drop-casting (Fig. 29b). Here, liquid solutions contain precursors with desired concentrations, surfactants or additives for easy wetting and complexing agents to enhance precursor dispersion in the solution. A vacuum is also commonly introduced after coating to assist with the penetration of the solution inside the scaffold. Infiltrated samples are subsequently exposed to thermal treatment to evaporate solvents, decompose precursors and additives and strengthen backbone/infiltrate bonds to allow for the deposition of discrete particles or continuous thin films on target surfaces (Fig. 29c, d).

Fig. 29

Schematic of a typical infiltration process. Reproduced with permission from Ref. [282]. Copyright 2014, Royal Society of Chemistry

Advances in the application of infiltration techniques in cathode development have been comprehensively reviewed by Jiang et al. [283], Vohs and Gorte [284], Jiang et al. [285] and Ding et al. [282]. Because of this, this review will provide a brief summary of advancements in the improvement of cathode ORR activity and stability through infiltration.

Large triple-phase interfaces are key to effectively lowering SOFC operating temperatures and the infiltration of either mixed-conducting materials into oxygen-ion-conducting porous scaffolds (usually made of electrolyte materials) [268, 286] or ionic-conducting materials into porous cathode backbones can extend ORR-active interfaces [287, 288]. In the case of cathode materials with sufficiently high mixed conductivities such as BSCF and SrCo_0.85Ta_0.15O_3−δ (SCT15), infiltrating with conventional ionic conducting materials such as doped ceria can degrade cathode ORR activity however. For example, our previous results [13] revealed that although the incorporation of SDC through both mechanical mixing and infiltration degraded the ORR activity of SCT15 in instrument air, slight improvements in activation energy from 102 kJ mol⁻¹ to 96–97 kJ mol⁻¹ for the SDC-infiltrated cathode were obtained (Fig. 30a). In addition, the CO₂ tolerance of the infiltrated cathode was improved by over five-fold at 550 °C in the presence of 10% CO₂ (Fig. 30b), which can be attributed to the extended ORR-active interface formed by the SCT15 and the infiltrated SDC nanoparticles in which SCT15 can sustain its electronic conductivity and SDC can sustain its ionic conductivity in 10% CO₂ (Fig. 30c).

Fig. 30

Comparison of a polarization resistance and b tolerance against 10% CO₂ of SCT15, SCT15 + SDC and SCT15 + SDC infiltrated with 4.2% and 8.4% SDC nanoparticles in a symmetrical cell. c Back-scattered electron images of the microstructures of SCT15 + SDC, SCT15 + SDC + 4.2% infiltrated SDC and SCT15 + SDC + 8.4% infiltrated SDC cathodes. Reproduced with permission from Ref. [13]. Copyright 2017, American Chemical Society

The selection of infiltrates, scaffolds and microstructures is limited by chemical incompatibility between scaffolds and infiltrates such as La/Zr [289, 290], Sr/Zr [291, 292] and BSCF/La₂NiO_4+δ at temperatures higher than 900 °C [270] in which the lowering of annealing temperatures or the shortening of annealing durations are common methods used to address this challenge. For example, Zhou et al. [270] developed a novel BSCF-based cathode protected by a hierarchical conformal La₂NiO_4+δ dense coating through the use of microwave plasma heating that allowed for rapid heating up to 1800 °C within seconds. As a result, the infiltrated coating only required 10 min for densification and did not experience the formation of additional phases, both of which are not achievable through conventional heating processes (i.e., 850 °C for 5 h). This densification of the La₂NiO_4+δ shell was further demonstrated to be able to effectively improve the tolerance of the pristine BSCF cathode, resulting in only a 1.7-fold increase in polarization resistance in 10% CO₂ for the first hour (likely due to reduced oxygen partial pressure) and no degradation for the following 24 h. Similarly, Ai et al. [293] reported that their fast annealing procedure involving placing samples into a preheated furnace at 800 °C followed by rapid cooling led to a continuous shell of La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ infiltrate over a La_0.8Sr_0.2MnO₃ backbone (Fig. 31) and that the resulting dense La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ thin shell significantly improved the cathode performance of La_0.8Sr_0.2MnO₃ by reducing polarization resistances from 88.2 Ω cm² to 2.1 Ω cm² at 700 °C.

Fig. 31

a Schematic of the effects of annealing in determining cathode microstructure. SEM images of the LSCF infiltrate microstructure annealed through b conventional heating and c rapid heating treatments. Reproduced with permission [293], Copyright 2016, Elsevier

Another strategy to address potentially adverse effects arising from multiple cycles of infiltration and thermal treatment is to apply electrodeposition at room temperature [252, 253, 271] in which the facilitation of electrochemical infiltration into the scaffold and the selection of suitable substrates for infiltrate deposition are key. Based on this, Lee et al. [252, 253, 271] incorporated 3D-structured carbon nanotubes into a porous scaffold through catalytic chemical vapor deposition in the presence of C₂H₄ and N₂ (50:50 vol%) at 750 °C (Fig. 31) using infiltrated cobalt nitrate as the catalyst for carbon formation. In one study involving LaNi_0.6Fe_0.4O_3−δ deposition for example [271] (Fig. 32), these researchers deposited Ni and Fe through co-electrodeposition in a single chamber electrolysis cell and La (OH)₃ through chemically assisted electrodeposition (CAED) in a typical H-cell with the cathode and anode chambers being separated by an anion exchange membrane in which the CAED technique involved the use of generated hydroxide ions (OH⁻) from electrochemical nitrate reduction (Eq. 19) to form precipitates with soluble metal ions (Mⁿ⁺) such as La and Co ions (Eq. 20). Here, the deposited precursors were thermally converted into desired oxides, which avoided the long-term sintering processes of conventional infiltration techniques and therefore prevented the potential formation of insulating phases such as La₂Zr₂O₇. And as a result of the absence of Sr in LaNi_0.6Fe_0.4O_3−δ, the resulting LaNi_0.6Fe_0.4O_3−δ-GDC composite showed significant improvements against Cr poisoning and exhibited stable operations at 750 °C for 300 h in the presence of Cr vapor [271].

$${\mathrm{NO}}_{3}^{-}+{\mathrm{H}}_{2}\mathrm{O}+2{\mathrm{e}}^{-}\to {\mathrm{NO}}_{2}^{-}+2{\mathrm{OH}}^{-}(E=0.1 \, \mathrm{V}\, \mathrm{vs}.\mathrm{ RHE})$$

(19)

$${\mathrm{M}}^{n+}+n{\mathrm{OH}}^{-}\to \mathrm{M}{(\mathrm{OH})}_{n}\downarrow $$

(20)

Fig. 32

Schematic of electrochemical deposition to prepare nano-structured LaNi_0.6Fe_0.4O_3−δ on a Sc_0.1Ce_0.01Zr_0.89O₂ scaffold. Reproduced with permission from Ref. [271]. Copyright 2020, American Chemical Society

5.2.2 Chemical Vapor Deposition

Chemical vapor deposition is an effective method to functionalize substrate surfaces with heterogeneous thin films through the thermally activated decomposition or reaction of precursors in vapor or gas form on substrates. However, typical chemical vapor deposition methods are not commonly used to decorate SOFC cathodes because of limited precursor penetration into cathode active surfaces inside pores. Nevertheless, Zhou et al. [272] demonstrated the effectiveness of chemical vapor deposition in promoting the cathode performance of SrSc_0.2Co_0.8O_3−δ by depositing 5-nm-thick amorphous iron oxide on the surface in which these researchers found that the decorated cathode demonstrated reduced polarization resistances of 50% at 700 °C and can stably operate in a fuel cell configuration for at least 250 h at 650 °C. To further address precursor transport limitations, atomic layer deposition involving evacuation (or vacuum) during deposition is also an emerging technique to decorate porous cathode surfaces.

5.2.3 Atomic Layer Deposition

Atomic layer deposition (ALD) is a thin-film deposition technique that can introduce uniform heterogeneous coatings onto cathode surfaces with similar thicknesses (so-called conformal coating). In typical ALD processes (Fig. 33), deposition usually occurs in a flow reactor in which pristine cathodes are placed in a chamber. Here, coating precursors are introduced into the chamber in vapor form and can form monolayers on the top of cathode surfaces in which as the surface becomes saturated with the precursor, excess precursor can be pumped away and other gas species can be introduced (e.g., oxidants such as water vapor or ozone) to further adsorb and condense the deposited monolayer. This process can subsequently be repeated to achieve desired film thicknesses over cathode surfaces [294]. And because precursor vapors can easily access the inner surface of porous cathodes to a depth of at least 50 μm [295] using ALD, cathode with uniform 3D hierarchal structures can be obtained [276, 295].

Fig. 33

Schematic of atomic layer deposition for cathode surface functionalization

Based on these advantages, researchers have attempted to deposit several metal oxides onto cathodes, such as ZrO₂ [274, 276], Fe₂O₃ [296] and CoO_x [295, 297]. Table 3 summarizes common metal precursors for ALD over cathode surfaces. Here, deposition temperatures are normally kept at between 150 and 400 °C to facilitate the thermal decomposition of vapors in which insufficient temperatures may slow down reaction rates, whereas excessive temperatures may prevent the adsorption of gas species onto cathode surfaces [294].

Table 3 Summary of common metal precursors used for ALD

Not all deposited films lead to enhanced ORR activity. Instead, cathodes deposited with metal oxides such as Al₂O₃ [300], ZrO₂ [273], La₂O₃ and Fe₂O₃ [296] show worse initial cathode performance than pristine cathodes, possibly due to degradation arising from either ORR-inert overlays that can block oxygen adsorption sites (e.g., ZrO₂ and Fe₂O₃ over La_0.8Sr_0.2FeO_3−δ) [296] or chemical incompatibility between overlayers and pristine cathode materials [291]. Because of this, the careful selection of materials for overlay is important to sustain cathode performance in which because surface overlay dominates surface oxygen kinetics, materials with mixed electron and oxygen-ion conductivities that provide easy access to gases are preferred for thin film deposition. For example, Rahmanipour et al. [296] achieved improved ORR activity by co-depositing La₂O₃ and Fe₂O₃ thin films over La_0.8Sr_0.2FeO_3−δ, whereas Chen et al. [276] introduced a conformal Pt coating onto a nanoscale ZrO₂ mesoporous medium to improve the performance of an LSM-YSZ cathode by 1.5–1.7-fold at 750 °C, which these researchers attributed to the promotion of mixed conductivity by the Pt-ZrO₂ nanocomposite. Choi et al. [275] further deposited a conformal thin LSC layer (1–12 nm) over a La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ cathode and reported enhanced cathode performance (0.45 lower polarization resistance than pristine) that contributed to a 1.8-fold increase in fuel cell power density at 600 °C. These researchers also found that the deposited LSC film was partially amorphous and therefore possessed an oxygen p-band center close to the Fermi level that can contribute to improved surface oxygen kinetics.

ALD can also suppress unwanted cation segregation through cation interdiffusion at the interface between the backbone and the deposited layer and prevent the backbone cathode from thermal agglomeration. For example, Gong et al. [273] deposited a nanostructured ZrO₂ overcoat over a La_0.6Sr_0.4Co_0.8Fe_0.2O_3−δ-Gd_0.2Ce_0.8O_1.9 composite cathode using a flow reactor in which tetrakis(dimethylamido)zirconium(IV) and water were dosed as Zr and O sources, respectively, to the cell at 250 °C and found that the ZrO₂ overcoat (~ 20 nm thick with nanopores, Fig. 34a) can significantly improve the overall thermal stability (ohmic and polarization resistances) of the cathode by fourfold for 1100 h at 800 °C despite the cathode with the overcoat initially showing inferior performance as compared with the pristine one. Here, these researchers proposed that cation interdiffusion (e.g., Co, Fe into ZrO₂ and Zr into the cathode) across the interface between La_0.6Sr_0.4Co_0.8Fe_0.2O_3−δ and the ZrO₂ overcoat can suppress Sr surface segregation over long-term annealing by alleviating electrostatic driving forces (i.e., attraction between electron holes and Sr_La, Fig. 34b). These researchers further used the same technique to coat 5-nm-thick ZrO₂ onto nanostructured La_0.6Sr_0.4CoO_3−δ infiltrated in a La_0.8Sr_0.2Ga_0.83Mg_0.17O_3−δ porous medium and reported significantly enhanced thermal stability at 700 °C for 4000 h (Fig. 34c) [274]. In this study, these researchers similarly claimed that interfacial cation interdiffusion can promote mixed conductivity in ZrO₂, suppress Sr surface segregation and stabilize the nanostructured cathode over long-term operation. Despite this, the potential formation of additional phases such as insulating SrZrO₃ [291, 292] and BaZrO₃ [301] between cathodes and the porous ZrO₂ film was overlooked by these studies and likely contributed to the reported degradation of initial cathode performance [273]. For example, Schmauss et al. [291] reported the formation of an unwanted SrZrO₃ phase between SmCoO_3−δ and the ZrO₂ overcoat after annealing at 750 °C for only 10 h and attributed this phase as the main reason for initial cathode performance degradation. However, these researchers also highlighted the potential benefits of SrZrO₃ in the suppression of Sr surface enrichment [277] and the limitation of cathode coarsening over long-term operation [291]. This significant improvement to cathode performance over long-term operation highlights the importance of nanostructure confinement and suppression of Sr surface enrichment toward cathode thermal stability.

Fig. 34

a TEM image of ZrO₂ porous thin film deposited using ALD on La_0.6Sr_0.4Co_0.8Fe_0.2O_3–δ. Reproduced with permission from Ref. [273]. Copyright 2013, American Chemical Society. b Proposed mechanism of a ZrO₂ overcoat in the enhancement of La_0.6Sr_0.4CoO_3−δ thermal stability. c Comparison of the polarization resistance between ZrO₂ coated and pristine La_0.6Sr_0.4CoO_3−δ cathodes in a symmetrical cell at 700 °C for 4000 h. Reproduced with permission from Ref. [274]. Copyright 2013, American Chemical Society

5.2.4 Pulsed Laser Deposition

Pulsed laser deposition (PLD) is a commonly used deposition technique to prepare thin films that is particularly useful to probe the fundamental mechanisms of cathode materials at operating conditions. PLD involves the application of pulsed laser beams to excite targets in ultra-high vacuum to generate energy such as thermal, chemical and mechanical energy that can in turn generate fumes from metal targets [302]. These generated fumes can subsequently condense and form dense thin films on substrates (e.g., electrolyte or cells). PLD can also allow for the co-evaporation of multiple materials including refractory materials (e.g., Nb and Ta) and rare-earth elements in short durations [303] to synthesize ORR-active materials that contain multiple elements.

Dense thin films fabricated by using PLD can also provide a platform for the investigation of intrinsic ORR activities, surface exchange kinetics and surface cation enrichment by ruling out potential effects from the grain boundaries and microstructures of conventionally fabricated cathodes in which the surface segregation mechanisms as discussed in Sect. 2 were uncovered based on thin-film perovskite materials prepared by using PLD [20, 22, 29, 51, 54]. The selection of substrates or modifications for PLD can further affect the surface orientation of thin films [304, 305]. By varying the electrolyte substrate for example, Zhu et al. [305] deposited (110) faceted Sr₂Sc_0.1Nb_0.1Co_1.5Fe_0.3O_6−2δ thin films onto (100)-oriented YSZ and the (100) facet on the (100)-oriented SDC/YSZ substrate using PLD, followed by annealing at above 600 °C. The authors reported a much higher ORR activity (ASR = 30.9 Ω cm⁻²) over (110) facets than over (100) facets (ASR = 45.6 Ω cm⁻²) at 650 °C. The improvement of the ORR activity arises from a less Sr surface enrichment and a higher concentration of redox-active Co ions over (110) facets as compared to over (100) facets. The surface composition of the thin film cathode can also be modified after PLD deposition. For instance, Tsvetkov et al. [23] modified the deposited La_0.8Sr_0.2CoO₃ thin films into aqueous solutions containing metal (i.e., Co, V, Nb, Zr, Ti, Hf and Al) chloride precursors. Interestingly, the authors found an increase of cobalt oxidations states at the thin film when the surface was doped with cations such as Ti and Hf, which likely arise from the diminishment of surface oxygen vacancies. The reduced oxygen vacancy level, as imparted by surface doping, alleviates the Sr surface segregation and thus benefit the oxygen surface exchange kinetics.

The PLD also enables the formation of well-defined hetero-interfaces, which allows one to explore the roles of the heterostructures such as perovskite (e.g., LSC/RP (e.g., (La, Sr)₂CoO₄) phases in boosting the ORR activity [279, 306,307,308]. For more detailed advances in heterointerface engineering, we refer readers to the very recent review by Zhao et al. [15]. Well-defined interfaces are also beneficial in the study of the role of interfaces in long-term cathode stability. For example, Bagarinao et al. [309] studied the oxygen-ion blocking effects of SrZrO₃ that contributed to cathode performance degradation by an order of magnitude by intentionally depositing a ~ 20-nm-thick SrZrO₃ in between La_0.6Sr_0.4CoO_3−δ and GDC, which further clarified the negative effects of SrZrO₃ formation at the interfaces between Sr-containing cathodes and Zr-containing materials.

5.2.5 Sputtering

Sputtering is another deposition technique that involves the use of plasma to bombard target atoms to form vapors that can subsequently condense onto desired substrates [302]. Commonly used sputtering techniques include direct-current, radio-frequency and magnetron sputtering in which direct-current sputtering is the simplest, involving the use of direct current power to create plasma over flowing Ar. Alternatively, radio-frequency sputtering alternates current to sputter at typical frequencies between 0.5 and 30 MHz [302], whereas magnetron sputtering uses magnetic fields to create high-density plasma to accelerate sputtering rates.

Unlike PLD, sputtering can easily manipulate cathode microstructures. For example, Lee et al. [280] recently obtained a power density of ~ 1.7 W cm⁻² at 600 °C for a 3-μm-thick all-sputtered SOFC cell with nanostructured La_0.6Sr_0.4Co_0.2Fe_0.8O_2.95/YSZ composite cathodes and suggested that the direct-current sputtering of Y/Zr alloys can allow for the easy formation of a porous columnar structured thin film (Fig. 35a), whereas the sputtering of ceramic La_0.6Sr_0.4Co_0.2Fe_0.8O_2.95 usually results in a dense thin film (Fig. 35b). These researchers also reported that the co-sputtering of both the Y/Zr alloy and the ceramic material led to a highly porous columnar structured nanocomposite (Fig. 35c). And because of the low temperatures required for the sputtering processes and the relatively low annealing temperature (600 °C), these researchers further reported the lack of insulating phases such as SrZrO₃ or La₂Zr₂O₇ forming, highlighting the advantages of sputtering in the fabrication of nanostructured composite cathodes over conventional procedures that involve high-temperature thermal treatments. In another example, Yun et al. [281] recently studied the interaction between Cu cations of a PrBaCo_1.8Cu_0.2O_5+δ substrate and sputtered Au nanoparticles at SOFC operating temperatures that involved Cu localization at deposited Au sites as revealed by EDS elemental mapping. And based on their results, these researchers suggested that this interaction can confine surface Au nanoparticles to ~ 25 nm to promote oxygen surface kinetics and therefore led to a ~ 60% improvement in cathode performance. In addition, Choi et al. [310] sputtered Sc stabilized ZrO₂ to successfully prevent their porous sputtered Ag electrode from thermal coarsening at below 500 °C, allowing this confinement effect as a result of the sputtered coating to contribute to a significantly improved peak power density of 59.3 mW cm⁻² and much higher cathode stability at 0.4 V for 25 h at 450 °C as compared with bare Ag and Pt electrodes.

Fig. 35

Schematics of sputtering a metals and b ceramics and c co-sputtering metals and ceramic as well as SEM images of the corresponding cross-sections and top-view microstructures of the sputtered materials. Reproduced with permission from Ref. [280]. Copyright 2020, American Chemical Society

6 Conclusions and Perspective

The surface and bulk properties of perovskite oxides are essential for ORR activity and stability in cathodes at intermediate temperatures. The surface of cathodes determines the availability of active sites for oxygen adsorption, charge-transfer and oxygen surface migration and react dynamically to external conditions such as contacting materials (e.g., electrolytes) and contaminants (e.g., CO₂, water vapor and Cr vapor). Alternatively, cathode bulk can determine both electron and oxygen-ion conductivities that are important for efficient ORR activity at reduced temperatures. Interplay also exists between the bulk and surface of cathodes. For example, bulk properties such as increased lattice strain, cation dopant/host size mismatch and cation non-stoichiometry can all contribute to cation levels (e.g., Sr surface enrichment) and surface oxygen vacancies. Alternatively, surface oxygen vacancies as determined by lattice structures, composition and operating conditions constitute a major driving force for cation segregation from bulk and surface. Moreover, surface cation segregation can alter bulk lattice composition and consequently significantly impact bulk properties.

As such, current strategies to improve cathode performance and stability involve either the manipulation of bulk properties through doping or the introduction of heterogeneous phases through mechanical mixing or surface deposition. The key information presented in this review is as follows:

• 1. Cation size is essential for crystal structures, oxygen transport kinetics and cation surface segregation.

  • Increases in A-site cation size and/or decreases in B-site cation size can shift the Goldschmidt tolerance factor toward higher values to alter crystal structures from orthorhombic/rhombohedral symmetry to cubic or even hexagonal/tetragonal phases.

  • Relatively large A-site cations can increase lattice sizes to expand the ionic size of B-site redox-active cations such as Co ions and increase oxygen vacancies. Enlarged lattices can also enhance oxygen-ion migration in lattices.

  • Dopant/host size mismatch, particularly for relatively large A-site cations, can induce lattice strain that can drive dopant segregation from bulk to the surface.

• 2. Cation oxidation state determines crystal structure, cation surface segregation and oxygen vacancy concentration.

  • High oxidation state B-site cations (such as Nb⁵⁺ and Ta⁵⁺) can exert strong electrostatic repulsion between adjacent octahedra and can therefore effectively stabilize beneficial cubic perovskite structures.

  • Dopant/host oxidation state mismatch (e.g., Sr²⁺ dopants at La³⁺ sites) constitutes the electrostatic driving force for cation segregation.

  • The introduction of low oxidation state A-site dopants such as alkaline-earth or alkali metals can either increase the oxidation state of B-site cations or induce more oxygen vacancies to compensate for overall charge imbalance. The effectiveness of this strategy to enhance oxygen vacancies is believed to depend on the redox activity of B-site cations, which has a general trend: Mn > Fe > Co > Ni.

• 3. Dopant electronegativity is a key factor for B-site cation states and therefore plays a vital role in ORR catalysis in which low electronegativity B-site dopants (e.g., Ta) generally lead to increased oxygen vacancies by affecting neighboring hosts such as Co.

• 4. The polarization power of cations is essential in determining cathode stability. Here, large-sized cations with low oxidation states normally show weaker polarization power, which will lead to strong bonding with anions such as CO₃²⁻ and explains the susceptibility of alkali-earth metals at A-sites to contaminants. However, large-sized A-site cations with low oxidation states can enhance oxygen vacancies at B-sites and therefore, increases to the bond energy between B-site cations and oxygen can enhance perovskite stability and contaminant poisoning resistance.

• 5. Surface modification is an effective approach to address the limitations of doping and can both increase oxygen vacancy and promote cation surface segregation in which various surface deposition methods such as infiltration, atomic layer deposition and sputtering can coat heterogeneous layers over pristine structures to result in the constraining of unwanted surface segregation, enhancement of oxygen kinetics and resistance to negative impacts from contaminants.**

In general, these strategies have multiple impacts on cathode properties, making the investigation of underlying mechanisms challenging. Nevertheless, they provide opportunities to overcome the limitations of catalyst design due to trade-offs between cathode activity and stability if single strategies are applied. A typical example involves the deposition of thin films over pristine Sr-doped perovskite surfaces to suppress Sr surface enrichment and take advantage of the beneficial effects of Sr on bulk oxygen kinetics.

Despite seminal efforts in cathode development however, the detailed relationship between ORR mechanisms and oxygen-ion transport in cathode surfaces and bulk is underexplored. However, several recent studies have reported the nonlinear relationship between ORR activity and oxygen vacancy concentration. For example, Tsvetkov et al. [23] found that surface exchange kinetics does not increase with oxygen vacancy concentration and attributed this nonlinear trend to Sr surface enrichment. Mayeshiba and Morgan [126] further calculated oxygen vacancy and mobility over an array of doped perovskites and found local nonlinear trends for samples with ORR-active elements including cobalt, nickel, iron and manganese ions at B-sites, meaning that the easy formation of oxygen vacancies may not necessarily lead to high oxygen mobility. Our results [12, 117, 118, 138] on Nb and Ta-doped perovskites also presented similar trends in which perovskite activities were much higher than BSCF but with significantly lower bulk oxygen vacancies. Overall, all of these recent experimental and theoretical findings highlight the fact that these nonlinear behaviors likely originate from both the surface and bulk properties of perovskites. Despite this, the role of cathode surfaces and bulk in overall performance remains unclear.

The rational modulation of interactions between cathode surfaces and bulk also shows promise to further improve cathode performance, particularly for cobalt-free perovskite oxides in which our team [18] demonstrated an effective method to in situ promote A-site alkali metal surface segregation so as to create A-site deficiencies in the bulk to improve electron and oxygen-ion conductivity and surface CO₂ poisoning resistance. Alkali metals themselves also present interesting behaviors at SOFC operating temperatures in which peroxides and sesquioxides of Na and K can release local oxygen partial pressures if reacting with gaseous contaminants such as CO₂.

Recent advancements in perovskite deposition techniques (e.g., atomic layer deposition, electrochemical deposition and sputtering) can further enable the practical application of nanostructured cathodes at elevated temperatures. More importantly, these techniques can be tailored to avoid potential side reactions between different phases through either the shortening of annealing durations or the reduction of annealing temperatures, both of which are not easily achievable through conventional cathode preparation methods. These techniques also hold great promise in the development of next generation composite cathodes with high density active sites and optimal porous structures.