Fuel cells are devices for electrochemically converting chemical energy into electrical energy and heat without the need for direct combustion as an intermediate step, giving much higher conversion efficiencies than conventional energy systems. In particular, solid oxide fuel cells (SOFCs), based on an oxide-ion conducting electrolyte, have several advantages over other types of fuel cells, including relatively inexpensive materials, low sensitivity to impurities in the fuel, and high efficiency [13].

SOFCs cannot yet compete with conventional combustion systems in terms of cost and durability. In recent years, great efforts have been devoted to develop low or intermediate temperature SOFCs (IT-SOFCs) operating at 500–800 °C. Lowering the operating temperature can suppress degradation of components and extend the range of acceptable material selection; this also serves to improve cell durability and reduce the system cost. However, reducing the operating temperature decreases the electrode kinetics and results in large interfacial polarization resistances. This effect is most pronounced for the oxygen reduction reaction at the cathode. In order to lower the polarization resistance of the cathode, a favorable electronic and ionic conductivity as well as a high catalytic activity for oxygen reduction must be maintained.

A vast number of papers related to the topic of SOFC cathodes have been published in the past three decades. Regrettably, it is not possible to cover all cathode aspects and references in the literature. This review aims to provide an overview of present research progress in the field of SOFC cathodes and will focus especially on the most widely used materials and related reaction mechanisms. Further emphasis will be placed on the development of cathode materials for low and IT-SOFCs including new cathode materials and microstructure optimization.

Overview of cathode reaction mechanism

Cathode requirements

In SOFCs, the cathode functions as the site for the electrochemical reduction of oxygen. To this effect, the cathode must have: (1) high electronic conductivity (preferably more than 100 S cm−1 under oxidizing atmosphere); (2) a matched thermal expansion coefficient (TEC) and chemical compatibility with the electrolyte and interconnect materials; (3) adequate porosity to allow gaseous oxygen to readily diffuse through the cathode to the cathode/electrolyte interface; (4) stability under an oxidizing atmosphere during fabrication and operation; (5) high catalytic activity for the oxygen reduction reaction (ORR); and (6) low cost. For high temperature SOFCs operating typically at 800–1000 °C, the cathode material of choice is a composite of Sr-doped LaMnO3 (LSM) and yttria-stabilized zirconia (YSZ). Other mixed ionic-electronic conductors (MIECs) have also been used as the cathode and are discussed later.

The operating temperature of SOFCs is determined by the temperature required to achieve sufficient ionic conductivity in the electrolyte. In addition, the choice of cathode materials is largely dependent on the electrolyte materials used with care taken to match thermal expansion coefficients and avoid undesirable interface reactions. The most commonly used electrolyte material is yttria-stabilized zirconia. Other oxides, such as scandia-stabilized zirconia, samaria-doped ceria (SDC), gadolinia-doped ceria (GDC), and lanthanum strontium gallium magnesium oxide (LSGM), are also under consideration because they have higher ionic conductivities at reduced operating temperatures. Selected data on the TEC and ionic conductivity of several commonly used electrolyte materials are listed in Table 1.

Table 1 TEC and ionic conductivity of electrolyte materials in air at 800 °C

Triple-phase boundaries at cathodes

On the surface of the cathode, the oxygen reduction can be described as follows:

$$ \frac{1}{2}{{\text{O}}_{\text{2}}}\left( {\text{gas}} \right) + 2{{\text{e}}^{-} }\left( {\text{cathode}} \right) \to {{\text{O}}^{2 - }}\left( {\text{electrolyte}} \right) $$

The electrochemical reactions are quite different from normal heterogeneous catalytic reaction in many aspects [11]. For instance, it is widely believed that the electrochemical reactions can only occur at the triple-phase boundaries (TPBs), which are defined as the confluence of sites where the oxygen ion conductor, electronic conductor, and the gas phase come in contact. A schematic illustration of the region between the electrolyte and the cathode where the TPB exists is shown in Fig. 1. If there is a breakdown in connectivity in any one of the three phases, the reaction cannot occur [12]. Furthermore, a hindrance of access for ions, gasses, or electrons to the reaction site renders it inactive. Microstructure and composition clearly affect the size and distribution of the TPBs. One compositional design option employed is to provide a single-phase electrode with mixed conductivities permitting both oxide ion and electron mobility within the cathode material, e.g., \( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{C}}{{\text{o}}_{1 - y}}{\text{F}}{{\text{e}}_y}{{\text{O}}_{3 - \delta }}\left( {\text{LSCF}} \right) \). Thus, electrochemical oxygen reduction can occur at the electrode surface as well as within the bulk electrode. Another strategy is to use a porous composite consisting of an electronic conducting cathode material and an appropriate amount of ionic conducting electrolyte material, e.g., LSM-YSZ. By using these strategies, the electrochemically active reaction sites can be orders of magnitude greater than that of porous cathodes exhibiting only electronic conductivity. The degree of this extension depends critically on the rate of defect transport through the solid MIECs, gas transport through the pores to promote surface coverage in the MIECs, and the catalytic activity of interfaces.

Fig. 1
figure 1

Schematic diagram of the cathode triple-phase boundary

Kinetics and reaction mechanisms of cathodes

Although the molecular species involved in the overall electrochemical reactions at the cathode is a single diatomic species (O2), it must first be converted to some “electroactive” intermediate form via one or more processes. Usually, this electrochemically kinetic step is restricted to an area close to TPBs [13, 14]. The electrochemical processes going on in the cathode involve different bulk and surface steps [12, 14, 15]. The elementary reactions in the overall electrode reaction are usually considered as follows [16, 17]: (1) the reduction of O2 molecules involving adsorption, dissociation, reduction, and incorporation of the oxygen anion into the lattice of the cathode materials; (2) ionic transport through the porous cathode toward the electrolyte; and (3) the ion jumping into the electrolyte lattice. Among all three elementary reaction steps, several steps could be rate-limiting for the oxygen reduction process. The oxygen reduction step is the biggest contributor to total cell resistance, and improvements on the catalytic activity of the cathode have a strong impact on the final cell performance [18, 19].

For electron-conducting perovskite-type materials, Fleig [20] summarized three possible paths for the cathodic reaction (Eq. 1), that is, the electrode surface path, the bulk path, and the electrolyte surface path as shown in Fig. 2. These three paths are in sequence suitable for the cases of an electronic conductor (e.g., LSM), an MIEC conductor (e.g., LSCF), and a composite (e.g., LSM-YSZ), respectively. It should be noted here that only at higher partial pressure of oxygen (p O2) that corresponds to the cathode operating conditions of SOFCs, LSM has a p-type conductivity. However, with decreasing p O2 and increasing temperature, oxygen vacancies are formed in LSM.

Fig. 2
figure 2

Sketches of the three reaction paths of the oxygen reduction and incorporation reaction and some possible rate-determining steps. Modifications of the paths (e.g., adsorption of a molecular rather than an atomic species or diffusion along the cathode/electrolyte interface) and a combination of electrode and electrolyte surface paths (adsorption on cathode and surface diffusion onto the electrolyte surface) are also possible [20]. Copyright © 2003 Annual Reviews

The cathodic reaction can simultaneously occur via all three paths, and for each path, one or more elementary steps determine the corresponding reaction rate. Which elementary reaction will be rate-limiting step is not always predictable and may depend on local conditions like temperature and oxygen partial pressure or on microstructural conditions. In addition, there may be parallel reaction pathways and a crossover of these various reaction pathways may cause interference. For example, the incorporation rate into the cathode (bulk path) depends on the surface concentration of adsorbed oxygen and thus can be influenced by the surface diffusion occurring via the electrode surface path mechanism. Despite the complicated nature of the reaction mechanism, oxygen reduction mechanistic studies provide very insightful understanding information, especially on well-defined systems [20]. The use of dense thin films [21, 22] and microelectrodes [23, 24] with well-defined geometry has provided insights into the kinetics of ORR by quantitative correlation of cathode geometry with ORR impedance.

Reducing the operating temperature of SOFCs has advantages for many practical applications. However, a decrease in temperature is concurrent with an increase in polarization losses. This phenomena is typified by the example shown in Fig. 3 [25]. When the temperature was decreased, the cathodic overpotentials of two cathode materials markedly increased. Thus, the key technical challenge is how to minimize electrode polarization losses especially at the cathode/electrolyte interface at lower operating temperatures.

Fig. 3
figure 3

Steady-state polarization curves of LSM and LSCF electrodes measured at different temperatures in air [25]. Copyright © 2002 Elsevier B.V.

The electrode polarization losses are usually associated with the generation and transport of oxygen ions within the porous cathode structure [26]. The Adler or ALS model [27] could provide a strategy for optimizing the composition and structure of electrodes to obtain good performance at intermediate temperatures. Providing that values for the oxygen self-diffusion coefficient (D*, cm2 s−1), oxygen surface exchange coefficient (k, cm s−1), and appropriate microstructural parameters are available, the heterogeneous chemical reaction contribution (R chem) can be calculated by the ALS model as shown in Eq. 2.

$$ {R_{chem}} = \frac{{RT}}{{2{F^2}}}\sqrt {\frac{\tau }{{\left( {1 - \varepsilon } \right)a{\text{C}}_{\text{0}}^{\text{2}}{D^*}k}}} $$

where τ, ε, and a are tortuosity, fractional porosity, and internal surface area/unit volume, respectively, C0 is the surface concentration of oxygen.

Isotopic exchange is an important technique to determine the oxygen diffusion in mixed ionic-electronic conducting materials [28, 29]. For example, Kilner et al. [29] measured the oxygen surface exchange and oxygen tracer diffusion coefficients of \( {\text{L}}{{\text{a}}_{0.8}}{\text{S}}{{\text{r}}_{0.2}}{\text{Mn}}{{\text{O}}_{3 + \delta }} \) using the 18O/16O isotope exchange depth profile method, obtaining \( k = 1.0 \times {10^{ - 8}}{\text{cm }}{{\text{s}}^{ - 1}}{\text{and}}\,{D^*} = 3.2 \times {10^{ - 13}}{\text{c}}{{\text{m}}^{\text{2}}}{{\text{s}}^{ - 1}}{\text{at }}900^\circ {\text{C}} \).

Usually, the polarization losses of the cathode can be decreased by two approaches. On one hand, it can be achieved by choosing an appropriate composition to improve the kinetics of oxygen exchange and diffusion [30] bearing in mind that the TEC of cathodes should be matched with those of other components. On the other hand, it can also be achieved by optimizing the microstructure at the cathode/electrolyte interface to extend the TPB length. Endo et al. [31] studied the electrochemical properties of three microstructural types (dense, porous, and porous/dense double layer) of cathode materials based on La1−x Sr x CoO3 (LSC) and La1−x Sr x MnO3 (LSM). Their results revealed that surface area is an important factor for reaction rate on the LSC electrode and high performance electrodes can be obtained by using a mixed ionic and electronic conducting electrode with a large surface area. For the MIEC cathodes, the role of surface area in improving their performance lies in the fact that a higher surface area leads to more active sites for the oxygen reduction reaction because the electrochemical reactions can only occur at the TPBs.

An alternative method to enhance cathode activity is utilizing cathodic biasing or polarization [3234], which has been shown to significantly reduce the oxygen reduction reaction overpotential on porous LSM [35, 36] and LSM-YSZ composites [37, 38]. This phenomenon is called the activation effect [39] or hysteretic behavior of cathodes [40]. It not only influences the surface microstructural and compositional changes of the cathode but also causes a morphological change at the interface between the cathode and electrolyte under the polarization/biasing treatment. The ORR activity was enhanced via oxygen vacancy formation associated with surface and compositional changes adjacent to TPBs [33] or by promoting direct incorporation of oxygen from the gas into the electrolyte due to partially reduced manganese oxide [32]. However, until now, there is no general consensus about the mechanism responsible for electrode activation or for the hysteretic behavior.

In recent years, new in situ experimental technologies are leading to new insight into the cathode reaction mechanism. Conventional electrochemical techniques such as impedance spectroscopy and voltammetry can provide detailed information about the rates of processes occurring in situ. However, these techniques are not informative for directly assigning which molecular species participate in the electrochemical reactions. In contrast, optical spectroscopy is capable of identifying the molecular structures present on the SOFC electrode surfaces. Lu et al. [41] developed a potential dependent Fourier Transform Infrared Spectroscopy technique (pd-FTIRES) to obtain information on oxygen species adsorbed directly onto the functioning cathode surface as a function of temperature, overpotential, and oxygen partial pressure. Although they believed that evidence for two or possibly three distinct dioxygen species present on the electrode surface were found, the separation of bulk phonon modes from surface species is usually difficult. Horita et al. [42] observed by secondary-ion mass spectrometry imaging technique that the active sites for oxygen reduction on the O2/LSM/YSZ were spots around the triple-phase boundaries. The width of the active sites for oxygen reduction was estimated to be less than 1 μm under the examined conditions.

Quantum chemical calculations could also play a vital role in rational design of better electrode materials for SOFCs. Wang et al. [43] studied the interactions between oxygen molecules and a silver surface or a CeO2(111)-supported atomic layer of silver using first-principle calculations. Their results showed that the O2 dissociation process on a monolayer of silver supported by CeO2(111) (TPB) with oxygen vacancies has a lower reaction barrier than on silver surfaces (two-phase boundary), and the dissociated oxygen ions can quickly bond with subsurface cerium atoms via a highly exothermic reaction with no barrier. The computed energies of the reactions show the O2 reduction process, and the incorporation of the dissociated O2- ions in the oxide electrolyte prefer to take place at the TPB region with oxygen vacancies, consistent with experimental observations, suggesting that the rate of oxygen reduction may be enhanced by increasing the TPB areas. Recently, Choi et al. [44] studied the catalytic mechanism of oxygen reduction on La0.5Sr0.5MnO3 (110) surfaces in solid oxide fuel cells using density functional theory (DFT) and molecular dynamics calculations. Their results suggest that O2 species are preferentially adsorbed on the Mn site rather than the La site according to the difference in adsorption energy, which supports the fact that B-site cations are more active than A-site cations for oxygen reduction on several perovskite-type ABO3 cathode materials. The atomic and molecular oxygen adsorption and diffusion on LaMnO3 (001) surfaces were also studied by Kotomin et al. [45] using ab initio DFT plane–wave supercell calculations. They concluded that the dissociative adsorption of O2 molecules from the gas phase is energetically favorable on the surface Mn ions even on a defect-free surface. The adsorbed O atoms could penetrate the electrode plane first when much more mobile surface oxygen vacancies approach the O ions strongly bound to the surface Mn ions. Direct incorporation of the O2 molecule into an O vacancy is also energetically favorable.

Development of cathode materials

Physical properties of perovskite cathode materials

Perovskite materials have been widely used as cathode materials in SOFCs [46]. In order to better design and optimize the cathode materials, it is necessary to first understand the fundamentals of the perovskite structure. A perovskite-type oxide has the general formula ABO3, in which A and B are cations with a total charge of +6. The lower valence A cations (such as, La, Sr, Ca, and Pb, etc.) are larger and coordinated to twelve oxygen anions while the B cations (such as, Ti, Cr, Ni, Fe, Co, and Zr, etc.) occupy the much smaller space and are coordinated to six oxygen anions. Full or partial substitution of A or B cations with cations of different valence is possible. When the overall valence of the A-site and B-site cations (n + m) adds up to less than six, the missing charge is made up by introducing vacancies at the oxygen lattice sites [47, 48]. Figure 4 shows the typical structure of the cubic perovskite ABO3 [47].

Fig. 4
figure 4

Unit cell of the ABO3 perovskite structure [47]. Copyright © 2003 Nature Publishing Group

Many perovskite structures are distorted and do not have cubic symmetry. Common distortions such as cation displacements within the octahedra and tilting of the octahedra are related to the properties of the A and B substituted atoms. The degree of distortion in ABO3 perovskites can be determined according to the Goldschmidt tolerance factor (t) as follows:

$$ t = \left( {{r_{\text{A}}} + {r_{\text{B}}}} \right)/\sqrt 2 \left( {{r_{\text{B}}} + {r_{\text{O}}}} \right) $$

where r A, r B, and r O are the effective ionic radii of A, B, and O ions, respectively. Usually, this factor is evaluated from Shannon's ionic radii [49] for respective coordination numbers. When the tolerance factor is near unity, the structure is the ideal cubic one. Smaller A or bigger B cations give rise to a decrease in tolerance factor with a cation displacement and tilting of corner sharing BO6 octahedra to an orthorhombic symmetry [50].

For most of the perovskite materials used as cathodes in SOFCs, the A-site cation is a mixture of rare and alkaline earths (such as La and Sr, Ca or Ba), while the B-site cation is a reducible transition metal such as Mn, Fe, Co, or Ni (or a mixture thereof). Therefore, in most cases, a redox catalytic mechanism is usually provided by B-site cations [51]. The octahedral symmetry around the transition metal often promotes a metallic or semiconducting band structure at high temperatures leading to high electronic conduction. With a rational choice of A- and B-site cations, a large and stable number of oxygen ion vacancies can be introduced at SOFC operating conditions, thus facilitating significant bulk ionic oxygen transport [12].

For \( {{\text{A}}_{1 - x}}{{\text{A}}_x}^\prime{\text{B}}{{\text{O}}_3} \) perovskite-type oxides resulting in substoichiometric oxygen-to-metal regions, an oxygen vacancy scenario results [52, 53], e.g., in \( {\text{L}}{{\text{n}}_{1 - x}}^{3 + }{\text{S}}{{\text{r}}_x}^{2 + }{{\text{M}}^{3 + }}{{\text{O}}_{3 - }}_\delta \), the oxygen vacancies arise either from substitution of A3+ by A2+ or from the partial reduction of B3+ or B4+ to B2+ and B3+, respectively. Oxygen mobility through vacancies is the basis of oxide ion conductivity. In some compositions of perovskite-type oxides, the oxide ion conductivities are as high as those observed in solid electrolyte materials, e.g., the ionic conductivity of \( {\text{S}}{{\text{r}}_{0.9}}{\text{C}}{{\text{e}}_{0.1}}{\text{Co}}{{\text{O}}_{3 - }}_\delta \) reaches 0.133 S cm−1 at 800 °C in air [5456]. In general, most cathode materials rely on doping of both the A- and B-sites to improve electrical conductivity and electrocatalytic performance.

Besides the electrical and structural characteristics mentioned above, thermodynamic data are essential for evaluating and predicting the long-term stability of perovskite cathode materials and their compatibility with other components at SOFC operating temperatures. The thermodynamic stability of perovskite cathode materials was discussed in terms of the stabilization energies and the valence stabilities by Yokokawa et al. [57, 58]. By determining the enthalpies of formation for the perovskite LaMnO3 (M = Cr, Fe, Co, and Ni) at room temperature using high-temperature oxide melt solution calorimetry, Navrotsky et al. [59] concluded that the relative thermodynamic stability of the perovskite LaMnO3 decreases in the order of Cr, Fe, Co, Ni. Tanasescu et al. correlated the thermodynamic properties with the compositional variables of perovskite ABO3 (A = La, Sr; B = Mn, Fe, Co) by a coulometric titration technique coupled with EMF measurements [6062]. The thermodynamic properties were represented by the relative partial molar free energies, enthalpies, and entropies of oxygen dissolution in the perovskite phase.

Lanthanum manganite-based cathodes

LSM cathode

Oxygen nonstoichiometry and oxygen defects have a great effect on the ionic and electronic transport properties of cathode materials. LaMnO3-based oxides have either oxygen-excess or oxygen-deficient nonstoichiometry. This is generally formulated by \( {\text{L}}{{\text{a}}_{1 - x}}{{\text{A}}_x}{\text{Mn}}{{\text{O}}_3}_{ \pm \delta } \) (A is divalent cation, such as Sr2+ or Ca2+; “+” denotes oxygen excess, and “−” denotes oxygen deficiency). Mizusaki et al. [53, 63, 64], Anderson [65, 66], and Nowotny et al. [67] studied in detail the oxygen nonstoichiometry of \( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_3}_{ \pm \delta } \) as a function of oxygen partial pressure, temperature, and composition. Various defect models have been proposed to explain the defect structure of the doped LaMnO3 oxides.

For lanthanum manganite, the most commonly used dopant is strontium because its size matches with lanthanum. The Sr dopant in \( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_3}_{ \pm \delta }\left( {x \leqslant 0.5} \right) \) does not increase the oxygen vacancy concentration, a common phenomenon in most of the other perovskite cathode materials studied but rather oxidizes the manganese ion according to Eq. 4 [68, 69].

$$ {\text{Mn}}_{{\text{Mn}}}^ \times + {\text{SrO}}\overset{{{\text{LaMn}}{{\text{O}}_3}}}{\longleftrightarrow} {\text{Sr}}_{{\text{La}}}^\prime + {\text{Mn}}_{{\text{Mn}}}^ \cdot + {\text{O}}_{\text{O}}^ \times $$

This reaction effectively increases the electron-hole concentration and improves the electrical conductivity. The electronic conductivity of LSM increases approximately linearly with increasing Sr concentration up to a maximum around 50 mol% [69]. At high temperature, LaMnO3 undergoes a solid-phase reaction with YSZ to form La2Zr2O7 (LZ) at the electrode–electrolyte interface [35, 7072]. A little amount of Sr substitution decreases the reactivity of LSM compound with YSZ. However, SrZrO3 (SZ) forms when Sr concentration is above about 30 mol% [7275]. Therefore, an Sr content of 30 mol% is considered as optimal against the formation of unwanted electronically insulating phases. Incorporating a slight A-site deficiency in the materials can further decrease unwanted reactivity.

The relative amounts of LZ and SZ depend on the La/Sr ratio in the LSM [73, 76] and increase with increasing reaction time and sintering temperature [73, 77, 78]. The conductivity of these zirconates is two to three orders of magnitude lower than that of YSZ and therefore must strictly be avoided in order to prevent high ohmic loss. In addition, it was found that nonstoichiometry at the La site in LaMnO3 plays an important role in the reaction [79]. Diffusion of Mn into YSZ leads to increases in La activity at the interface and promotes the interfacial reaction.

Mitterdorfer et al. [80] studied in detail nucleation and growth of LZ between porous \( {\text{L}}{{\text{a}}_{0.85}}{\text{S}}{{\text{r}}_{0.15}}{\text{M}}{{\text{n}}_y}{{\text{O}}_{3 + }}_\delta \) with (001) oriented 9.5 mol% doped YSZ single crystals using many characterization approaches. The insulating LZ phase between LSM and YSZ will interfere with two possible processes, which induce undesired polarization. One is surface diffusion of electroactive oxygen along the LSM surface to the TPBs; the other is the electrochemical formation of O2- at the TPB due to limited availability of electronic charge carriers.

Other manganite cathodes

Generally, the electrical properties of manganite-based perovskite compounds are not sufficient for operation at temperatures below 800 °C. The absence of oxygen vacancies in LSM restricts the reduction of oxygen to the three-phase boundary regions. This limitation is the primary reason why LSM does not have acceptable performance at lower temperatures. Usually, two approaches have been taken to improve the performance of LSM cathodes so that they may be used at lower temperatures [68]. The first is to add a second ionically conducting phase to LSM. As mentioned above, this extends the surface area over which the oxygen reduction can occur. The second approach has been to replace La with other rare earth elements or dope LSM with a cation (such as, Co, Fe, or Ni) that promotes the formation of oxygen vacancies when strontium is doped on the A-site. In this subsection, the latter is mainly discussed.

The cathodic overpotentials of manganite cathodes can be modified greatly by replacing different rare-earth cations at the A site [81, 82]. Ishihara et al. [81] systematically studied the cathodic overpotentials of \( {\text{L}}{{\text{n}}_{0.6}}{\text{S}}{{\text{r}}_{0.4}}{\text{Mn}}{{\text{O}}_3}\left( {{\text{Ln}} = {\text{La,Pr,Nd,Sm,Gd,Yb,orY}}} \right) \). The Sr-doped PrMnO3 cathode showed low overpotential values even at decreased operating temperatures. Sr-doped PrMnO3 also exhibits a compatible thermal expansion coefficient with YSZ. Consequently, they concluded that doped PrMnO3 is a promising cathode material for IT-SOFCs. In addition, the formation of undesired pyrochlore phases Ln2Zr2O7 between LSM cathode and YSZ electrolyte can be suppressed for those having smaller lanthanoids especially the \( {\text{P}}{{\text{r}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_3}{\text{,N}}{{\text{d}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_3}{\text{,and S}}{{\text{m}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_3}\,{\text{systems}} \) [83].

More Sr substitution and an increase of the atomic number of lanthanides are beneficial to adjust the TEC of Ln1−x Sr x MnO3 to match with those of the electrolytes. Kostogloudis et al. [84] found that \( {\text{N}}{{\text{d}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Mn}}{{\text{O}}_{3 \pm x}}\left( {x = 0.4,0.5} \right) \) with a TEC of 12.3 × 10−6 K−1 demonstrated thermal and chemical compatibility with GDC electrolytes. Pr0.5Sr0.5MnO3 showed high electrical conductivity of 226 S cm−1 at 500 °C [85], which is expected to increase with temperature indicating a potentially promising cathode material for SOFCs.

The majority of work performed on the lanthanide manganite materials has concentrated on the cases where the A site is Sr-doped. In terms of chemical compatibility and electrical conductivity, however, Ca-doped ones have proven more promising. Pr0.7Ca0.3MnO3 showed chemical stability and was thermally compatible (11.9 × 10−6 K−1) with YSZ electrolyte materials [86]. Its conductivity is close to that of the previously mentioned Sr-doped PrMnO3 with the same composition.

As for the doping at the B-site, Sc-doped manganite \( {\text{L}}{{\text{a}}_{0.8}}{\text{S}}{{\text{r}}_{0.2}}{\text{M}}{{\text{n}}_{1 - x}}{\text{S}}{{\text{c}}_x}{{\text{O}}_{3 - }}_\delta \left( {\text{LSMS}} \right) \) is a potential cathode material for IT-SOFCs [87]. Nonstoichiometric defects are introduced into the perovskite lattice of LSMS samples by Sc substitution, which leads to an increased oxygen ion mobility in the samples containing Sc. However, high-level doping of Sc (>10 mol%) results in the segregation of Sc2O3 secondary phases at elevated temperatures. The cells with LSMS cathodes exhibit higher performance, especially at lower temperatures, which can be ascribed to the increased oxygen vacancies in the LSMS. For this cathode material, the high cost of scandium may be a concern for applications.

\( {\text{S}}{{\text{r}}_{1 - x}}{\text{C}}{{\text{e}}_x}{\text{Mn}}{{\text{O}}_{3 - }}_\delta \,{\text{system}} \) are also potential cathode materials for IT-SOFCs. The TECs of \( {\text{S}}{{\text{r}}_{1 - x}}{\text{C}}{{\text{e}}_x}{\text{Mn}}{{\text{O}}_{3 - }}_\delta \left( {0.1 \leqslant x \leqslant 0.3} \right) \) varies between 12.4 × 10−6 and 10.1 × 10−6 K−1, close to that of CeO2-based electrolytes [88]. By doping Co in the B-site, the area-specific resistance of \( {\text{S}}{{\text{r}}_{0.8}}{\text{C}}{{\text{e}}_{0.2}}{\text{M}}{{\text{n}}_{0.8}}{\text{C}}{{\text{o}}_{0.2}}{{\text{O}}_{3 - }}_\delta {\text{is }}0.10{\text{ }}\Omega {\text{ c}}{{\text{m}}^2}{\text{at }}750^\circ {\text{C}} \), which is about 20 times lower than that of \( {\text{S}}{{\text{r}}_{0.8}}{\text{C}}{{\text{e}}_{0.2}}{\text{Mn}}{{\text{O}}_{3 - }}_\delta \) [89]. The \( {\text{S}}{{\text{r}}_{0.8}}{\text{C}}{{\text{e}}_{0.2}}{\text{M}}{{\text{n}}_{0.8}}{\text{C}}{{\text{o}}_{0.2}}{{\text{O}}_{3 - }}_\delta \,{\text{cathode}} \) shows high catalytic activity for oxygen reduction in the temperature range of 700–800 °C.

Lanthanum cobaltite and ferrite cathodes

Cobaltite cathodes

LaCoO3 exhibits a rather high electronic density of states near the Fermi level (E F) [90]. The marked catalytic properties of LaCoO3 are associated with electron occupation of the crystal field d state near E F and with the buildup of surface charge so as to enhance the electron transfer between a surface cation and a potentially catalyzed species.

\( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Co}}{{\text{O}}_{3 - }}_\delta \) exhibits a complex behavior with regard to the dependence of nonstoichiometry, electrical, and magnetic properties on the strontium content, temperature, and oxygen partial pressure. It is of interest to consider the defect structure of these oxides. Petrov et al. [91] proposed a defect model associated with the defect structure of \( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Co}}{{\text{O}}_{3 - }}_\delta \) in which strontium ions are assumed to occupy the regular La lattice sites, \( {\text{S}}{{\text{r}}_{\text{La}}}^\prime \), leading to predominantly electron holes. In order to maintain electrical neutrality, the substitution of Sr ions must be compensated by the formation of equivalent positive charges, comprising \( {\text{Co}}_{{\text{Co}}}^ \cdot \) and oxygen vacancies \( \left[ {{\text{V}}_O^{ \cdot \cdot }} \right] \). The overall electroneutrality condition is then given by

$${\text{Sr}}^{ \times }_{{{\text{Sr}}}} + {\text{Co}}^{ \cdot }_{{{\text{Co}}}} + 2{\text{V}}^{{ \cdot \cdot }}_{{\text{O}}} = {\text{S}}{{\text{r}}_{\text{La}}}^\prime + {\text{Co}}^{ \times }_{{{\text{Co}}}} ,$$
$$2{\text{Co}}^{ \cdot }_{{{\text{Co}}}} + {\text{O}}^{{\text{ $ \times $ }}}_{{\text{O}}} \to 2{\text{Co}}^{{\text{ $ \times $ }}}_{{{\text{Co}}}} + {\text{V}}^{{ \cdot \cdot }}_{{\text{O}}} + \frac{1}{2}{\text{O}}_{2} {\left( {\text{g}} \right)},$$


$${\left[ {{\text{V}}^{{ \cdot \cdot }}_{{\text{O}}} } \right]}{\left[ {{\text{Co}}^{{\text{ $ \times $ }}}_{{{\text{Co}}}} } \right]}^{2} = {\text{K}}_{{{\text{V}}^{{ \cdot \cdot }}_{{\text{O}}} }} {\left[ {{\text{Co}}^{ \cdot }_{{{\text{Co}}}} } \right]}^{2} {\left[ {{\text{O}}^{{\text{ $ \times $ }}}_{{\text{O}}} } \right]}{{P}}^{{ - 1 \mathord{\left/ {\vphantom {1 2}} \right. \kern-\nulldelimiterspace} 2}}_{{{\text{O}}_{2} }} ,$$

where \( {\text{Co}}_{\text{Co}}^\cdot \) and \( {\text{Co}}_{\text{Co}}^\times \) denote Co4+ and Co3+ ions at a normal Co site, respectively, and \( {K_{{\text{V}}_{\text{O}}^{ \cdot \cdot }}} \) is the equilibrium constant.

Usually, cobalt-based materials display higher ionic and electronic conductivities than other cathode materials. Therefore, the use of cobalt-containing cathode materials should result in a decreased cathode polarization resistance. \( {\text{L}}{{\text{a}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{Co}}{{\text{O}}_{3 - }}_\delta \) has a marked electrode activity due to high oxygen diffusivity and high dissociation ability of oxygen molecules [92]. However, a large amount of cobalt results in an increased thermal expansion coefficient, which may result in a delamination at the cathode/electrolyte interface or cracking of the electrolyte [93]. By substituting the La cation with an alternative cation, such as Gd or Pr, a decrease in the ASR and TEC can be expected to occur. The TECs of \( {\text{L}}{{\text{n}}_{0.6}}{\text{S}}{{\text{r}}_{0.4}}{\text{Co}}{{\text{O}}_{3 - }}_\delta \left( {{\text{Ln}} = {\text{La,Pr,Nd,Sm,and Gd}}} \right) \) decrease from La to Gd. A similar trend has also been observed in the analogous Ln1−x Sr x MnO3 and Ln0.8Sr0.2Fe0.8Co0.2O3 systems [93, 94]. The large TEC of the cobaltites (~20 × 10−6 K−1) compared with that of the manganites (~11 × 10−6 K−1) is due to the formation of oxygen vacancies, spin-state transitions associated with Co3+ [95], and the relatively weaker Co–O bond compared with the Mn–O bond [96].

Doping Cu into the Co site can further enhance the ionic conductivity and catalytic activity of \( \left( {{\text{La,Sr}}} \right){\text{Co}}{{\text{O}}_{3 - \delta }} \), although the electrical conductivity of \( \left( {{\text{La,Sr}}} \right)\left( {{\text{Co,Cu}}} \right){{\text{O}}_{3 - }}_\delta \left( {\text{LSCC}} \right) \) was lower [97]. Among perovskites with different rare earth cations at the A-site, those incorporating Pr3+ exhibited the highest electrical conductivity and the lowest overpotential values due to the additional contribution of Pr3+/Pr4+ valence change [98]. A combination of Pr, Sr for the A-site, and Co, Cu for the B-site material \( {\text{P}}{{\text{r}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{C}}{{\text{o}}_{1 - y}}{\text{C}}{{\text{u}}_y}{{\text{O}}_{3 - \delta }} \) was studied by Zhu et al. [99]. Their results indicated that \( {\text{P}}{{\text{r}}_{0.7}}{\text{S}}{{\text{r}}_{0.3}}{\text{C}}{{\text{o}}_{1 - y}}{\text{C}}{{\text{u}}_y}{{\text{O}}_{3 - \delta }}\,{\text{materials}}\,\left( {y = 0.05 - 0.3} \right) \) had lower overpotential compared with those of \( {\Pr_{0.7}}{\text{S}}{{\text{r}}_{0.3}}{\text{Co}}{{\text{O}}_{3 - \delta }} \) and especially low overpotentials for compounds with the \( {\text{P}}{{\text{r}}_{0.7}}{\text{S}}{{\text{r}}_{0.3}}{\text{C}}{{\text{o}}_{0.9}}{\text{C}}{{\text{u}}_{0.1}}{{\text{O}}_{3 - }}_\delta \,{\text{composition}} \). The maximum power density of an anode-supported cell consisting of \( {\text{P}}{{\text{r}}_{0.7}}{\text{S}}{{\text{r}}_{0.3}}{\text{C}}{{\text{o}}_{0.9}}{\text{C}}{{\text{u}}_{0.1}}{{\text{O}}_{3 - }}_\delta \,{\text{cathode}} \), SDC electrolyte, and \( {\text{NiO/C}}{{\text{e}}_{0.8}}{\text{S}}{{\text{m}}_{0.2}}{{\text{O}}_{1.9}}{\text{anode reached }}481{\text{ mW c}}{{\text{m}}^{ - 2}}{\text{at }}750^\circ {\text{C}} \).

For Mn doping at the B-site, Chen et al. [100] observed that \( {\text{L}}{{\text{a}}_{0.6}}{\text{S}}{{\text{r}}_{0.4}}{\text{C}}{{\text{o}}_{0.8}}{\text{M}}{{\text{n}}_{0.2}}{{\text{O}}_{3 - }}_\delta \) has the lowest cathodic overpotential among the tested \( {\text{L}}{{\text{n}}_{0.6}}{\text{S}}{{\text{r}}_{0.4}}{\text{C}}{{\text{o}}_{0.8}}{\text{M}}{{\text{n}}_{0.2}}{{\text{O}}_{3 - }}_\delta \left( {{\text{Ln = La,Gd,Sm,or Nd}}} \right) \) compounds at 500–800 °C. The electrical conductivity of \( {\text{L}}{{\text{a}}_{0.6}}{\text{S}}{{\text{r}}_{0.4}}{\text{C}}{{\text{o}}_{0.8}}{\text{M}}{{\text{n}}_{0.2}}{{\text{O}}_{3 - }}_\delta \) reaches about 1,400 S cm−1 at 500 °C. However, these oxides also have large thermal expansion coefficients greater than 18 × 10−6 K−1.

Sr-doped samarium cobaltite (Sm1−x Sr x CoO3, SSC) is another widely studied cathode material for IT-SOFCs showing a higher electrical conductivity. The electrical conductivity increases with Sr doping and a maximum conductivity is usually observed at x = 0.5. The electrical conductivity of this composition shows metallic-like behavior and reaches up to 103 S cm−1 in the temperature range of 800–1100 °C [101, 102]. Although SSC has been studied as a cathode material for SOFCs with yttria-stabilized zirconia [103], La1−x Sr x Ga1−y Mg y O3 [102] and doped ceria electrolyte [104], its thermal expansion coefficient is greater than 20 × 10−6 K−1, which could produce stresses at the interface during thermal cycling. In addition, reactivity tests of Sm1−x Sr x CoO3 with YSZ show that Sm1−x Sr x CoO3 is stable below 800 °C because SrZrO3 is formed only above 900 °C [103].

Besides the disadvantage of high thermal expansion coefficient, cobaltite cathodes react readily with YSZ to form insulating compounds during the sintering process, which degrades their performance. Therefore, a diffusion barrier layer is necessary for the cells consisting of cobaltite-based cathodes and the YSZ electrolyte [105, 106].

Ferrite cathodes

Lanthanum ferrite (LaFeO3) is expected to be more stable than cobaltite perovskites because the Fe3+ ion has a stable electronic configuration 3d5. Sr-doped LaFeO3 (LSF) cathodes have shown promising performance with respect to the power density and stability at 750 °C [107109]. In iron-based cathodes, reactivity with YSZ electrolyte is significantly reduced. In addition, TECs of the ferrite–perovskite are relatively close to those of the YSZ and CGO electrolyte.

The addition of Sr to LaFeO3 creates \( {\text{S}}{{\text{r}}_{\text{La}}}^\prime \), thus causing a charge unbalance. The charge neutrality is then maintained by the formation of Fe4+ ions or oxygen vacancies \( \left( {{{\text{V}}_{\ddot o}}} \right) \). The highest theoretical conductivity expected by Sr addition would be at x = 0.5 creating a maximum Fe4+/Fe3+ ratio of 1:1 [110]. At high temperatures, LSF will lose oxygen to form oxygen vacancies at the cost of decreasing hole concentration. This reaction can be represented in Kröger-Vink notation:

$$ 2{\text{Fe}}_{{\text{Fe}}}^ \cdot + {\text{O}}_{\text{O}}^ \times \to \frac{1}{2}{{\text{O}}_2} + 2{\text{Fe}}_{{\text{Fe}}}^ \times + {\text{V}}_{\text{O}}^{ \cdot \cdot } $$

where \( {\text{Fe}}_{\text{Fe}}^\cdot \) represents Fe ions in the 4+ valence state (similar to holes), \( {\text{Fe}}_{\text{Fe}}^\times \) denotes Fe cations in 3+ valence state, and \( {{\text{V}}_{\ddot o}} \) represents the oxygen vacancy. Under relatively high oxygen partial pressures (e.g., air or O2), holes are the predominant charge carriers while at lower oxygen partial pressures \( {\text{S}}{{\text{r}}_{\text{La}}}^\prime \) defects are compensated by oxygen vacancies [111]. Because of the much higher mobility of the electrons or holes than that of oxygen ions, the total conductivity in ferrites is dominated by holes rather than oxygen ions. Therefore, the total conductivity in ferrites is dominated by a hole-conduction mechanism [110].

By incorporating La deficiency in La0.8Sr0.2FeO3, Ralph et al. [107] found that the ASR of this cathode was significantly reduced and reached 0.1 Ω cm2 at 800 °C. The thermal expansion coefficient of the La0.7Sr0.25FeO3 cathode closely matches with those of CGO and YSZ. This cathode has demonstrated no degradation over a 500-h operation indicating a promising material for lower-temperature SOFCs.

Simner et al. [108] found that an anode-supported cell consisting of La0.8Sr0.2FeO3 cathode, YSZ electrolyte, and NiO–YSZ anode exhibited a powder density as high as about 0.9–0.95 W cm−2 at 750 °C and 0.7 V and exhibited excellent stability over a 300-h test period. In addition, they found that Sr-doped LaFeO3 showed significant variability in performance depending on the type of current collectors used [112]. The use of platinum and silver current collectors is accompanied by the improved cell performance and stability while the cell utilizing gold cathode current collectors shows a severe degradation of performance. Similar effects caused by different current collectors on other cathode materials have not yet been reported.

Cu-doped lanthanum strontium ferrite showed superior kinetics for the electrochemical reduction of oxygen [113]. The current densities of cells with La0.7Sr0.3Cu0.2Fe0.8O3 as a cathode are approximately two to three times higher than that with La0.7Sr0.3FeO3 and ten times higher than that with La0.8Sr0.2FeO3 as a cathode. Copper doping significantly reduces the allowable sintering temperatures during the electrode fabrication. An enhanced low-temperature sintering of a Sr-doped lanthanum ferrite composition \( {\left( {{\text{L}}{{\text{a}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}} \right)_{0.{\text{98}}}}{\text{F}}{{\text{e}}_{0.{\text{98}}}}{\text{C}}{{\text{u}}_{0.0{\text{2}}}}{{\text{O}}_{\text{3}}} \) (with 2-mol% Cu-doping and 2-mol% A-site deficiency) has also been observed below 1,000 °C by Simmer et al. [114]. The cathode can be sintered onto YSZ without significant Zr4+ diffusion into the perovskite. However, the Cu-doped composition does exhibit chemical interaction with YSZ (typically above 950 °C), resulting in the precipitation of monoclinic zirconia. Initial single-cell studies utilizing a \( {\left( {{\text{L}}{{\text{a}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}} \right)_{0.{\text{98}}}}{\text{F}}{{\text{e}}_{0.{\text{98}}}}{\text{C}}{{\text{u}}_{0.0{\text{2}}}}{{\text{O}}_{\text{3}}} \) cathode on an anode-supported YSZ cell have indicated power densities in the range of 1.35 ~ 1.75 W cm−2 at 750 °C and 0.7 V [113].

Although pure LSF is thermodynamically more stable than that of cobalt-containing perovskites [115], it was observed that its electrochemical performance was inferior to LSCF with a similar A-site stoichiometry [115117]. A potential degradation mechanism for LSF cathodes is the interdiffusion of Zr and La [112, 118120]. Anderson et al. [120] found that there are reactions of YSZ with \( {\text{L}}{{\text{n}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}{\text{Fe}}{{\text{O}}_{\text{3}}}\left( {{\text{Ln}} = {\text{La}},{\text{Sr}},{ \Pr },{\text{or Nd}}} \right){\text{and L}}{{\text{a}}_{0.{\text{8}}}}{\text{B}}{{\text{a}}_{0.{\text{2}}}}{\text{Fe}}{{\text{O}}_{\text{3}}} \) with the sole exception being La0.8Ca0.2FeO3 (LCF). The reactivity either results in secondary phase formation or incorporation of Zr into the perovskite B-site. For the case of Ca, it may be a consequence of a lower limit of Zr dissolution into the LCF compound due to its smaller unit cell volume. Therefore, protective ceria-based barrier layers are usually used to prevent the reaction between LSF and YSZ during the sintering step of cathodes [109, 120122].

Ferro-cobaltite cathodes

La1−x Sr x Fe1−y Co y O3 shows good electrical conductivity, a high oxygen surface exchange coefficient, and a good oxygen self-diffusion coefficient between 600 and 800 °C. The oxygen self-diffusion coefficient of LSCF is 2.6 × 10−9 cm2 s−1 at 500 °C, which is superior in performance to that of La0.8Sr0.2MnO3 which has a oxygen self-diffusion coefficient of 10−12 cm2 s−1 at 1,000 °C [123, 124]. In addition, LSCF does not react with ceria-based electrolytes [125, 126]. Although the electrical conductivity was found to decrease slightly as the Sr content decreased, the TECs were found to be the lowest for the composition with the highest A-site deficiency, that is \( {\text{13}}.{\text{8}} \times {\text{1}}{0^{ - {\text{6}}}}{K^{ - {\text{1}}}}\,\,{\text{for }}\,{\text{L}}{{\text{a}}_{0.6}}{\text{S}}{{\text{r}}_{0.2}}{\text{C}}{{\text{o}}_{0.2}}{\text{F}}{{\text{e}}_{0.8}}{{\text{O}}_{3 - }}_\delta \,{\text{at }}\,700^\circ {\text{C}} \). This TEC value matches commonly used electrolytes [127].

Usually, the increase of ionic conductivities is more influenced by Sr concentration at the A-site while the increase of the electronic conductivities is more influenced by Fe and Co concentration at the B-site. The A-site deficiency has only a small effect on the TEC while a higher Sr content results in a higher TEC due to higher oxygen vacancy concentrations. Mai et al. [117] recently observed that a small A-site deficiency and a high strontium content had a particularly positive effect on the cell performance using \( {\text{L}}{{\text{a}}_{{\text{1}} - x - y}}{\text{S}}{{\text{r}}_x}{\text{C}}{{\text{o}}_{0.{\text{2}}}}{\text{F}}{{\text{e}}_{0.{\text{8}}}}{{\text{O}}_{{\text{3}} - }}_\delta \) as a cathode material. The measured current densities of cells consisting of \( {\text{L}}{{\text{a}}_{{\text{1}} - x - y}}{\text{S}}{{\text{r}}_x}{\text{C}}{{\text{o}}_{0.{\text{2}}}}{\text{F}}{{\text{e}}_{0.{\text{8}}}}{{\text{O}}_{{\text{3}} - }}_\delta \,{\text{cathode}} \), CGO electrolyte, and a NiO–YSZ anode were as high as 1.76 A cm−2 at 800 °C and 0.7 V, which is about twice the current density of cells with LSM/YSZ cathodes.

At lower operation temperatures, LSCF-based cathodes are superior to LSM-type cathodes due to a lower area-specific resistance. However, LSCF-type perovskites are generally incompatible with YSZ electrolytes due to undesirable interface reactions. Therefore, a CGO diffusion barrier layer is used to prevent the formation of low conductive compounds without negatively affecting the electrochemical performance [128, 129]. Another cause of higher degradation rates for cells with LSCF cathodes lies in the diffusion of strontium out of LSCF, which leads to a strontium depletion in the cathode and significantly lower performance. It has been observed that the cells with slightly less Sr in the \( {\text{cathode }}\left( {{\text{L}}{{\text{a}}_{0.{\text{58}}}}{\text{S}}{{\text{r}}_{0.{\text{38}}}}{\text{C}}{{\text{o}}_{0.{\text{2}}}}{\text{F}}{{\text{e}}_{0.{\text{8}}}}{{\text{O}}_{{\text{3}} - }}_\delta } \right) \) showed lower performance than that of the \( {\text{L}}{{\text{a}}_{0.{\text{58}}}}{\text{S}}{{\text{r}}_{0.{\text{4}}}}{\text{C}}{{\text{o}}_{0.{\text{2}}}}{\text{F}}{{\text{e}}_{0.{\text{8}}}}{{\text{O}}_{{\text{3}} - }}_\delta \) [130, 131]. It has been found that partial decomposition of the perovskite usually has greater influence on the cell degradation than interface reactions during the operation [132, 133].

Although the magnitudes of the electrical conductivity and thermal expansion of the \( {\text{L}}{{\text{a}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}\left( {{\text{Mn}},{\text{Fe}},{\text{Co}}} \right){{\text{O}}_{{\text{3}} - \delta }} \) perovskite are mainly dependent on the percentage of Co in the compositions [134], substitution of the La cation at A-sites by Pr cation with multiple valence states can further improve the cathode performance. Meng et al. [135] investigated \( {\text{P}}{{\text{r}}_{1 - x}}{\text{S}}{{\text{r}}_x}{\text{C}}{{\text{o}}_{0.8}}{\text{F}}{{\text{e}}_{0.2}}{{\text{O}}_{3 - \delta }}\left( {\text{PSCF}} \right)\left( {x = 0.2\sim 0.6} \right) \) as cathode materials for IT-SOFCs. Their results showed that the material with a composition of \( {\text{P}}{{\text{r}}_{0.{\text{6}}}}{\text{S}}{{\text{r}}_{0.{\text{4}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{F}}{{\text{e}}_{0.{\text{2}}}}{{\text{O}}_{{\text{3}} - }}_\delta \) has the highest conductivity, 1,040 S cm−1, at 300 °C. The ASR of the PSCF–GDC (50:50 by weight) cathode is as low as 0.046 Ω cm2 at 800 °C. The power densities of the electrolyte-supported cell consisting of PSCF-GDC cathodes, a GDC electrolyte, and a NiO–GDC anode reach 520 mW cm−2 at 800 °C and 303 mWcm−2 at 700 °C. All these results indicate that PSCF-GDC (50:50 by weight) is a promising cathode material for IT-SOFCs. Colomer et al. [30] studied the \( {\text{S}}{{\text{r}}_{0.{\text{8}}}}{\text{C}}{{\text{e}}_{0.{\text{1}}}}{\text{F}}{{\text{e}}_{0.{\text{7}}}}{\text{C}}{{\text{o}}_{0.{\text{3}}}}{{\text{O}}_{{\text{3}} - \delta }}\left( {\text{SCFC}} \right) \) cathode material for IT-SOFCs. The area-specific resistivity of SCFC electrodes is 0.86 Ω cm2 at 700 °C. They suggested that the enhanced oxygen ion diffusion in SCFC provides superior cathodic performances compared with that of LSCF at 700 °C.

\( {\text{B}}{{\text{a}}_{0.{\text{5}}}}{\text{S}}{{\text{r}}_{0.{\text{5}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{F}}{{\text{e}}_{0.{\text{2}}}}{{\text{O}}_{{\text{3}} - }}_\delta \left( {\text{BSCF}} \right) \) has been studied as an excellent oxygen permeation membrane material in the last decade. Haile and Shao [136] first employed BSCF as a cathode material in a solid oxide fuel cell using hydrogen as the fuel. They found that the peak power densities of an anode-supported cell consisting of a NiO-SDC anode, an SDC electrolyte, and a BSCF cathode reach 1,010 and 402 mW cm−2 at 600 °C and 500 °C, respectively. The area-specific resistances of this material are 0.055–0.071 and 0.51–0.61 Ω cm2 at 600 and 500 °C, respectively, which are considerably lower than that of other perovskite type cathode materials under similar operation conditions. Although this potential cathode material shows excellent electrochemical performance, the obvious disadvantage of BSCF is the high TEC, with a value of 20 × 10−6 K−1 between 50 °C and 1,000 °C [137], as is common with Co-containing perovskites. This indicates that thermal cycling could be problematic for cells using this cathode material due to a mismatched TEC with those of other cell components. In order to solve the issue of thermal expansion, one possibility is to make composite cathode materials with BSCF. However, the introduction of the composite may lessen the BSCF performance.

For most of the perovskite oxides containing alkaline-earth elements, one stability concern is that they could react with gas species such as CO2, SO2, or water vapor, especially with carbon dioxide from the air, at a significant rate at temperatures less than 800 °C [138143]. Since the presence of CO2 and water vapor is a characteristic of atmospheric air, these materials for SOFCs are preferably operated at temperatures above 800 °C. Otherwise, a prepurification process for air is required to remove carbon dioxide when using strontium cobaltite- or strontium manganite-based cathodes at temperatures below 800 °C [144].

Yan et al. [145] investigated the effect of CO2 on the performance of a BSCF cathode. They found that the BSCF cathode is susceptible to CO2 even in relatively small quantities at temperature ranges of 450–750 °C and especially at lower temperatures. The poisoning effect of CO2 on the performance of BSCF is further aggravated when the CO2 content increases and the temperature decreases. The deterioration of BSCF performance by carbon dioxide is reversible up to 550 °C. A further decrease of the operation temperature irreversibly reduced the single cell performance. It was found that the presence of CO2 inhibits the oxygen reduction process over the BSCF cathode surface and consequently decreases the cell performance. In the case of single chamber systems fed with hydrocarbons, a high concentration CO2 is produced. Therefore, the BSCF cathode is not suitable for single chamber fuel cells using hydrocarbons as fuels especially at low temperatures.

Nickelate cathodes

At atmospheric oxygen pressure, perovskite type nickelate LaNiO3−δ is stable only below 1230–1270 K; further heating leads to decomposition into \( {\text{L}}{{\text{a}}_{\text{2}}}{\text{Ni}}{{\text{O}}_{{\text{4}} + }}_\delta \) with a K2NiF4-type structure and NiO, via the separation of Ruddlesden–Popper (RP) type \( {\text{L}}{{\text{a}}_{\text{4}}}{\text{N}}{{\text{i}}_{\text{3}}}{{\text{O}}_{{\text{1}}0 - }}_\delta \,{\text{and L}}{{\text{a}}_{\text{3}}}{\text{N}}{{\text{i}}_{\text{2}}}{{\text{O}}_{{\text{7}} - }}_\delta \) phases at intermediate stages [146148]. These transformations limit the practical use of LaNiO3−δ -based materials, and they can be partly suppressed by a doping into the nickel sublattice [147].

Iron-doped nickelate \( {\text{LaF}}{{\text{e}}_{{\text{1}} - x}}{\text{N}}{{\text{i}}_x}{{\text{O}}_{{\text{3}} - }}_\delta \) exhibits an attractive combination of electronic transport properties, a moderate thermal expansion, and phase stability in air. Researchers at NTT Corporation in Japan have been investigating LaNi0.6Fe0.4O3 (LNF) as a cathode material for SOFCs operated at intermediate temperatures [148150]. LNF has many desirable characteristics, including a high electrical conductivity, less mismatch of thermal expansion coefficient with a zirconia electrolyte, high catalytic activity, and high durability against the cathode poisoning by volatile chromium species from metallic components [151, 152]. The electronic conductivity of LNF is 580 S cm−1 at 800 °C [148]. The average thermal expansion coefficient of LNF is lower (11.4 × 10−6 K−1) than that of \( {\text{L}}{{\text{a}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}{\text{Mn}}{{\text{O}}_{\text{3}}}\left( {{\text{12}}.0 \times {\text{1}}{0^{ - {\text{6}}}}{{\text{K}}^{ - {\text{1}}}}} \right) \) and closer to the TEC value of 8YSZ (10.5 × 10−6 K−1). Unfortunately, LNF is more reactive toward ZrO2-based electrolytes than conventional La(Sr)MnO3 at temperatures used for sintering [150]. The amount of La2Zr2O7 formation rapidly increases when the sintering temperature of the LNF cathode exceeded 1,000 °C. A preloading treatment at a very high current density (>1.0 A cm−2) causes current-voltage characteristics of cells using LNF as the cathode to be improved significantly. An anode-supported cell consisting of 10 mol% Sc2O3-1 mol% Al2O3-stabilized ZrO2 (SASZ) electrolyte, a NiO–SASZ anode, and an LNF cathode sintered at 1,000 °C exhibited a maximum power density of 1.56 W cm−2 at 800 °C.

Recently, an improvement of the initial performance of the LNF cathode was observed by placing an active layer consisting of Ce0.8Sm0.2O1.9 (SDC) or Ce0.9Gd0.1O1.95 (GDC) and LNF particles between the LNF cathode and a scandia alumina-doped zirconia electrolyte [153]. It was found that the LNF tends to react with SDC when the sintering temperature is 1,100 °C or higher. However, this reaction does not seem to affect the performance of LNF cathodes containing a LNF–SDC active layer when the sintering temperature is below 1,250 °C.

For the Sr-doped LNF (LSNF), the electronic conductivity reached a peak value when the total concentration of Ni and Sr is about 0.6 in the temperature range of 600–800 °C [149]. The conductivity of LSNF exhibits temperature dependence similar to that of metals when the concentrations of Ni and Sr are high. However, the conductivities drop sharply when the temperature exceeds about 600 °C, which is probably caused by an increase in the oxygen vacancy concentration, rather than by the degradation of the structure symmetry. An oxygen vacancy increase is expected to decrease the electronic conduction of perovskite oxides.

As a reference for choice of cathode materials in SOFC systems, Table 2 summarizes the thermal expansion coefficients and electronic and ionic conductivities of a number of perovskite-type cathode materials.

Table 2 Perovskite-type oxide materials: thermal expansion coefficient (TEC), electronic (σ e ), and ionic conductivities (σ i ) in air

K2NiF4 type structure cathode materials

\( {\text{L}}{{\text{a}}_{\text{2}}}{\text{Ni}}{{\text{O}}_{{\text{4}} + }}_\delta \)-based compounds have a K2NiF4-type structure and are usually formulated as \( {{\text{A}}_{\text{2}}}{\text{B}}{{\text{O}}_{{\text{4}} + }}_\delta \), which is described as a stacking of perovskite ABO3 layers alternating with rock salt AO layers along the c-direction [164] as shown in Fig. 5 [165]. Such oxides are of interest as an oxygen electrode material for applications in various solid electrolyte devices. This is primarily associated with the high electronic conductivity of \( {{\text{A}}_{\text{2}}}{\text{B}}{{\text{O}}_{{\text{4}} + }}_\delta \left( {{\text{B}} = {\text{Co}},{\text{Ni}},{\text{Cu}}} \right) \) compounds and solid solutions based on them due to the mixed valence of the B-site metal. In addition, a high concentration of oxygen interstitials offers the possibility of rapid oxygen transport through the ceramic material and thus provides a new type of mixed ionic electronic conductor cathode materials.

Fig. 5
figure 5

K2NiF4 type structure showing alternating AO and ABO3 layers [165]. Copyright © 2005 Elsevier B.V.

Taking \( {\text{L}}{{\text{a}}_{{\text{2}} - x}}{\text{S}}{{\text{r}}_x}{\text{Co}}{{\text{O}}_{{\text{4}} + }}_\delta \,{\text{oxides}} \) as an example, the following defect model was proposed for K2NiF4-type structure oxides by Vashook et al. [166]. In the \( {\text{L}}{{\text{a}}_{{\text{2}} - x}}{\text{S}}{{\text{r}}_x}{\text{Co}}{{\text{O}}_{{\text{4}} + }}_\delta \,{\text{oxide}} \), \( {\text{S}}{{\text{r}}_{\text{La}}}^\prime \) and \( {\text{V}}_{\text{O}}^{ \cdot \cdot } \) are assumed to exist as defects. The reaction can be represented in Kröger-Vink notation:

$$ 2{\text{SrO}} + 2{\text{La}}_{{\text{La}}}^ \times + {\text{O}}_{\text{O}}^ \times \Leftrightarrow {\text{2S}}{{\text{r}}_{{\text{La}}}}^\prime + {\text{L}}{{\text{a}}_2}{{\text{O}}_3} + {\text{V}}_{\text{O}}^{ \cdot \cdot } $$

Co3+ cations can disproportionate into \( {\text{C}}{{\text{o}}^{2 + }}\left( {{\text{Co}}_{\text{Co}}^\prime} \right)\,{\text{and}}\,{\text{C}}{{\text{o}}^{4 + }}\left( {{\text{Co}}_{\text{Co}}^\cdot } \right) \). An increase of conductivity of \( {\text{Sr - doped LaCo}}{{\text{O}}_{4 + }}_\delta \) with oxygen partial pressure could be understood as the result of an increasing concentration of holes according to the following reaction

$$ 2{\text{ Co}}_{{\text{Co}}}^ \times + 0.5{{\text{O}}_2} + {\text{V}}_{\text{O}}^{ \cdot \cdot } \Leftrightarrow 2{\text{ Co}}_{{\text{Co}}}^\cdot + {\text{O}}_{\text{O}}^ \times $$

The electrical conductivities of the \( {{\text{A}}_{\text{2}}}{\text{B}}{{\text{O}}_{{\text{4}} + }}_\delta \)-type nickelates and cobaltites are of a p-type conduction mechanism. These compounds reach nearly 100 S cm−1 at 800 °C under high oxygen partial pressures. At lower oxygen partial pressures, their conductivities decrease slower than that of some perovskite oxides [167].

The oxygen diffusivity of \( {\text{L}}{{\text{a}}_{{\text{2}} - x}}{\text{S}}{{\text{r}}_x}{\text{Ni}}{{\text{O}}_{{\text{4}} + }}_\delta \) is higher than that of LSCF, particularly at lower temperatures, but lower than that of LSC [168]. In addition, \( {\text{L}}{{\text{a}}_{{\text{2}} - x}}{\text{S}}{{\text{r}}_x}{\text{Ni}}{{\text{O}}_{{\text{4}} + }}_\delta \) appears to be more stable than either of these two materials in terms of thermal behavior at high temperatures. The thermal expansion coefficient of \( {\text{L}}{{\text{a}}_2}{\text{Ni}}{{\text{O}}_{4 + }}_\delta \,{\text{is }}13.7 \times {10^{ - 6}}{{\text{K}}^{ - 1}} \), which matches well with those of possible electrolyte materials such as YSZ or CGO. Recently, Zhao et al. [169] found that the interfacial polarization resistance of \( {\text{L}}{{\text{a}}_{{\text{1}}.{\text{2}}}}{\text{S}}{{\text{r}}_{0.{\text{8}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{N}}{{\text{i}}_{0.{\text{2}}}}{{\text{O}}_{{\text{4}} + }}_\delta \)-based electrodes was 1.36 Ω cm2 at 600 °C when the electrode was not activated. Current treatment performed at 200 mA cm−2 caused an activation increase within 30 min. The power density of an anode-supported single cell consisting of a \( {\text{L}}{{\text{a}}_{{\text{1}}.{\text{2}}}}{\text{S}}{{\text{r}}_{0.{\text{8}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{N}}{{\text{i}}_{0.{\text{2}}}}{{\text{O}}_{{\text{4}} + }}_\delta - {\text{GDC cathode}} \), a GDC electrolyte, and a NiO–GDC anode reached 350 mW cm−2 at 600 °C. This composite cathode shows potential application in SOFCs, although the long term-stability remains unproven.

Doping at the La-site with alkaline-earths (Sr, Ba, Ca) and other rare earths (Nd or Pr), or doping at the Ni-site with other transition metals (typically Cu or Co) may lead to dramatic changes of the structural and physical properties. \( {\text{L}}{{\text{a}}_2}{\text{N}}{{\text{i}}_{1 - x}}{\text{C}}{{\text{u}}_x}{{\text{O}}_{4 + }}_\delta \left( {0 \leqslant x \leqslant 1} \right) \) has been evaluated as a possible cathode material for SOFCs [170]. The thermal expansion coefficients are in the range of \( {\text{1}}0.{\text{8}}\sim {\text{13}}.0 \times {10^{ - {\text{6}}}}{{\text{K}}^{ - {\text{1}}}} \). Lower ASR values are obtained with a LSGM electrolyte compared with those with 8YSZ as an electrolyte material. For the \( {\text{L}}{{\text{a}}_{\text{2}}}{\text{N}}{{\text{i}}_{0.{\text{6}}}}{\text{C}}{{\text{u}}_{0.{\text{4}}}}{{\text{O}}_{{\text{4}} + }}_\delta \) composition, the total conductivity reached 87 S cm−1 at 580 °C.

The \( {\text{B}}{{\text{a}}_{1.2}}{\text{S}}{{\text{r}}_{0.8}}{\text{Co}}{{\text{O}}_{4 + \delta - }}\;{\text{GDC}} \) (70:30 by weight) composite cathode showed a low polarization, 0.17 Ω cm2, on a GDC electrolyte [171]. However, similar to that of Ba0.5Sr0.5Co0.8Fe0.2O3, this material may also be susceptible to reactions with CO2 in air.

Effect of addition of noble metal phases

As the operating temperature is reduced, electrode polarization losses increase. The cathode thus becomes the limiting component for further progress. One way to overcome this limitation may be adding a certain amount of noble metal, such as, palladium, silver, or platinum to the cathode. The addition of noble metal phases to an active cathode layer is mainly used to enhance the oxygen reduction reaction.

The melting point of silver is 961 °C. Metallic silver is a potential component for the cathode in SOFCs operated at less than 800 °C because of its good catalytic activity, high electrical conductivity, and its relatively low cost. A composite cathode consisting of high ionic conductivity copper-subtituted bismuth vanadates (Bi2V0.9Cu0.1O5.35) and silver was used for low-temperature SOFCs by Xia et al. [172]. This composite cathode exhibits a remarkable catalytic activity for oxygen reduction at 500–550 °C and greatly reduces the cathode–electrolyte interfacial polarization resistances down to about 0.53 Ω cm2 at 500 °C. The power density of an anode-supported cell consisting of a NiO–GDC anode, a GDC electrolyte, and a Ag-Bi2V0.9Cu0.1O5.35 composite cathode reaches 0.231 W cm−2 at 500 °C.

Different methods have been used to prepare Ag-based composite cathode. Wang et al. [173] reported that the Ag coating improved the oxygen exchange reaction activity of \( {{\text{L}}_{0.{\text{6}}}}{\text{S}}{{\text{r}}_{0.{\text{4}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{F}}{{\text{e}}_{0.{\text{2}}}}{{\text{O}}_{\text{3}}} - {\text{CGO}} - {\text{Ag}} \) composite cathode and interface conductivities at 600 °C. Ag nanoparticle-infiltration is another effective and low-cost approach to improve the cathode performance [174, 175]. Recently, Sakito et al. [175] observed that the infiltration of ~18 wt.% Ag fine particles into LSCF results in ~50% enhancement of power density. The Ni–GDC/GDC/LSCF–Ag cell shows a maximum power density of 0.415 W cm−2 at 530 °C, which is very attractive for IT-SOFCs. Ag was also co-sputtered with LSC by a dc magnetron sputtering method [176]. The interfacial resistance of \( {\text{L}}{{\text{a}}_{0.{\text{7}}}}{\text{C}}{{\text{o}}_{0.{\text{3}}}}{{\text{O}}_{\text{3}}}-{\left( {{{\text{Y}}_{\text{2}}}{{\text{O}}_{\text{3}}}} \right)_{0.{\text{25}}}}{\left( {{\text{B}}{{\text{i}}_{\text{2}}}{{\text{O}}_{\text{3}}}} \right)_{0.{\text{75}}}} \) decreases from 0.4 Ω cm2 at 750 °C to 0.3 Ω cm2 by the addition of 30 vol.% Ag. The Ag–LSC electrodes showed good stability during 200 h annealing at 750 °C. Although silver-based cathodes show good electrochemical properties, it should be pointed out here that oxygen is easily dissolved into silver and forms an Ag/Ag2O solid solution. In addition, even far below its melting point, silver is relatively mobile. Therefore, these concerns should be addressed prior to a long-term application of silver-based cathodes in SOFCs.

Pd is another noble metal widely used to improve the cathode performance in SOFCs. Sahibzada et al. [177] investigated the effect of the addition of Pd to a LSCF cathode and found the overall cell resistances decreased by 15.5% at 650 °C and 40% at 550 °C indicating a promotion effect of Pd. Simmer et al. [109] also observed that Pd-facilitated initial performance improvements in the \( {\text{L}}{{\text{a}}_{0.{\text{8}}}}{\text{S}}{{\text{r}}_{0.{\text{2}}}}{\text{Fe}}{{\text{O}}_{{\text{3}} - }}_\delta - {\text{C}}{{\text{e}}_{0.{\text{8}}}}{\text{S}}{{\text{m}}_{0.{\text{2}}}}{{\text{O}}_{{\text{1}}.{\text{9}}}}\,{\text{cathode}} \) with respect to the catalytic activity for O2 reduction. However, this enhancement is not sustainable and is followed by a rapid degradation of power density in less than 24 h compared with reference cells fabricated without Pd additions. The alloying of Pd particles with Pt that has presumably migrated from the cathode current collector may be responsible for the observed degradation phenomenon, which is confirmed by SEM/EDX examinations. This is also indirectly confirmed by Haanappel's results [178] that the addition of Pt to cathode materials does not positively affect the electrochemical performance of cells. Haanappel et al. [178] systematically studied the effects of Pd prepared by different methods on the catalytic activity of cathodes for the oxygen reduction. They found that neither infiltration of the cathode with Pd solution nor mixing with Pd black results in a positive effect. The catalytic effect was only observed with Pd loading on activated carbon. The well-dispersed Pd particles on an activated carbon support may be the reason for the increased catalytic effect. This also explains why their results are different from Simmer's work [109]. Usually, the particle size, the distribution, and the crystallinity of noble metals in the cathode play an important role on their catalytic activity. Therefore, a similar composition of materials prepared by different methods may show different results. As for the application of Pd on activated carbon, the stability and lifetime under SOFC operating conditions are potential concerns. It should be pointed out here that the addition of noble metals, in particular Pt, does not seem a good solution to improve the performances of cells since one target to commercialization of SOFC is to decrease price.

Microstructure optimization of cathode materials

The length of the three-phase boundary correlates well with the interfacial resistances to electrochemical oxidation of fuels at the anode [179181] and reduction of oxygen at the cathode [31, 182185]. Therefore, an extension of the TPB and an increase of the number of active reaction sites become necessary to improve electrode performances. This can be achieved by developing electrode materials with a higher mixed ionic-electronic conductivity and optimizing the microstructure of electrodes.

Nanostructured electrode materials with significantly higher surface areas have demonstrated superior electrochemical properties [186]. Liu et al. [183] reported the fabrication of nanostructured and functionally graded composite cathodes, which are graded in microstructure as well as in composition, using a combustion CVD process. Their results showed that nanostructured electrodes dramatically reduce the cathode/electrolyte interfacial polarization resistances (0.17 Ω cm2). The power density of an anode-supported cell consisting of a SSC–SDC cathode, a GDC electrolyte, and a Ni–SDC anode reached 0.375 W cm−2 at 600 °C while the long-term stability of these electrodes was not reported.

In recent years, the wet impregnation method has showed promise in the development of nanostructured electrodes with high performance. Many examples can be found in two recent review papers [40, 187]. Huang et al. [188] found that adding LSCo to an LSM-YSZ cathode by the infiltration method had a dramatic effect on performance but forming a mixed Sr-doped LaCo0.2Mn0.8O3 did not. Mixed conductivity at the three-phase boundary may explain the improved performance. The impregnation of nanoscale particles, forming connected networks, into solid oxide fuel cell electrodes has also been shown to lead to considerable improvement in performance [189]. Impedance analysis of the impregnated electrode suggests that the particular microstructure of the nanoparticulate pore-wall coating minimizes the rate-limiting step in the oxygen reduction reaction within LSM-YSZ cathodes.

Although the wet impregnation method has demonstrated to be an effective technique to achieve advantageous microstructures and a high performance, the durability of these materials is a big concern over long periods at high temperatures. Generally, the nanosized particles tend to aggregate at a high temperature, and the catalytic activity decreases with the operation time.

In addition, it should be noted that materials synthesized by different methods usually have very different microstructures and properties. For cathode materials with the same composition, the materials prepared by a sol–gel or glycine–nitrate combustion method usually have much smaller grain size and higher surface area compared with those prepared by a conventional solid-phase reaction, thus leading to better cell performances. However, the feasibility of scale-up and cost of these processes must be taken into account when considering incorporation into practical applications.

Chromium-resistant cathode materials for IT-SOFCs

In recent years, IT-SOFCs have received much attention as potential power generation systems with the potential for high efficiency operation. An obvious advantage of IT-SOFCs is that metallic alloys can be used as interconnect materials, which significantly reduces the system cost. In particular, chromium-containing ferritic stainless steels are the most promising and widely used oxidation-resistant alloys due to their appropriate thermal expansion behaviors and low cost. However, the application of chromium-containing alloys still poses many challenges even at reduced temperatures. For instance, the oxide scale formed on the surface of the alloys after a long exposure in the SOFC environment results in a high electrical resistance that eventually degrades cell performance [190, 191]. In addition, under high temperatures and oxidizing atmospheres, volatile chromium (Cr) species such as CrO3 and CrO2(OH)2 are generated [192, 193]. These species can cause rapid performance deterioration in SOFCs due to the poisoning of cathodes such as (La,Sr)MnO3 and (La,Sr) (Co,Fe)O3 toward the oxygen reduction reaction [194199]. The poisoning is largely due to the deposition of Cr species at the electrode/electrolyte interface regions or on the electrolyte surface.

Figure 6 shows the influence of steel samples with and without coatings on the performance of anode-supported single cells consisting of a YSZ electrolyte, a Ni/YSZ cermet anode, and a (La,Sr) (Co,Fe)O3 cathode, operated at 700–800 °C with a hydrogen fuel [200, 201]. Without the presence of the steel sample, no degradation was observed. Introduction of the Crofer22 steel coupon into the chamber of the tested cell resulted in a rapid degradation of cell voltage although the current flow was not passed through the steel. Samples treated with a protective (Cr,Mn)3O4 spinel layer on the surface of Crofer22 did not show this effect, even when examined over a nearly 2,000-h operation. Further tests are needed to confirm this behavior over a time period ranging from 3,000 to 8,000 h. Preliminary evaluation of chromium content in the cathode material shows a distinct correlation with the amount of chromium present in the air supplied to the cathode.

Fig. 6
figure 6

Influence of steel samples with and without coating on the performance of single cells [200]. Copyright © 2006 Wiley-Blackwell

Possible mechanism of chromium deposition at cathodes

Taniguchi et al. [202, 203] made extensive investigations on chromium poisoning of lanthanum strontium manganite in terms of the overpotential of cathodes, thermodynamic activity of Cr2O3, and water vapor content in the cathode compartment. These observations provide a fundamental understanding of the degradation mechanisms presented in cathode reactions.

The vapor pressure of chromium-containing species over Cr2O3 or other chromium-containing oxides has been thoroughly investigated and is summarized as follows [187, 204]:

  1. (1)

    Effect of oxygen partial pressure: CrO3(g), CrO2(g), and CrO(g) have different vapor pressures over Cr2O3 and vary as a function of the oxygen potential. With increasing oxygen partial pressure, the equilibrium vapor pressure of each species increases. Under dry air, CrO3 is the predominately present species.

  2. (2)

    Effect of water vapor: Since CrO2(OH)2(g) and other species containing O and H are relatively stable, the H2O vapor pressure is also an important factor. Under humid air, the predominant volatile species is CrO2(OH)2.

  3. (3)

    Effect of temperature: The vapor pressure of CrO3(g) exhibits large temperature dependence, while the vapor pressure of CrO2(OH)2(g) is small.

Yokokawa et al. [205] considered chromium poisoning from the viewpoint of thermodynamic stability of the perovskite cathodes. The Cr2O3 deposition or spinel CrMn2O4 formation on the TPBs is a direct cause of chromium poisoning.

The interaction between alloy interconnects and cathodes in SOFCs have been extensively investigated. In the earlier studies, Badwal et al. [195] and Taniguchi et al. [202] considered that the deposition of Cr species is closely related to the oxygen activity at the electrode/electrolyte interface. The deposition of Cr species is believed mainly by the electrochemical reduction of gaseous Cr species to solid Cr2O3 in competition with the oxygen reduction reaction. However, Konysheva et al. [194] recently reported that the degradation of cells by chromium poisoning could have two causes. On one hand, the Cr species may react with some cathode materials to form poorly conductive phases, such as SrCrO4; on the other hand, the deposited solid chromium oxide species may block the gaseous oxygen diffusion and transportation as well as the active TPBs. In recent years, Jiang et al. [206210] proposed that the Cr deposition process at the SOFC cathode is, in principle, a non-electrochemical process. The deposition process is kinetically limited by nucleation reactions between the gaseous Cr species and nucleation agents. The reactions can be generally written as follows [210, 211]:

$$ {\text{1}}/{\text{2C}}{{\text{r}}_{\text{2}}}{{\text{O}}_{\text{3}}}\left( {\text{oxide scale}} \right) + {\text{3}}/{\text{4}}{{\text{O}}_{\text{2}}}_{\left( {\text{g}} \right)} \to {\text{Cr}}{{\text{O}}_{{\text{3}}\left( {\text{g}} \right)}} $$
$$ {\text{Cr}}{{\text{O}}_{\text{3}}} + {\text{N}} \to {\text{Cr}} - {\text{N}} - {{\text{O}}_x}\left( {\text{nuclei}} \right) + {{\text{O}}_{\text{2}}} $$
$$ {\text{Cr}} - {\text{N}} - {{\text{O}}_x}\left( {\text{nuclei}} \right) + {\text{Cr}}{{\text{O}}_{\text{3}}} \to {\text{C}}{{\text{r}}_{\text{2}}}{{\text{O}}_{\text{3}}} $$

where N denotes the nucleation agent. Similarly, Cr deposition steps can also be written for other gaseous Cr species. In the case of the LSM electrode, the nucleation agent has been identified to be the manganese species (i.e., Mn2+) generated under cathodic polarization conditions [207, 208] and for the LSCF electrode, it is the SrO species enriched originally or segregated at the electrode surface [197, 206]. Further reactions between the nuclei, gaseous Cr species, and nucleation agents lead to the deposition and formation of (Cr,Mn)3O4 and SrCrO4 [206209].

$$ {\text{Cr}} - {\text{N}} - {{\text{O}}_x}\left( {\text{nuclei}} \right) + {\text{Cr}}{{\text{O}}_{\text{3}}} + {\text{M}}{{\text{n}}^{{\text{2}} + }} \to {\left( {{\text{Cr}},{\text{Mn}}} \right)_{\text{3}}}{{\text{O}}_{\text{4}}} $$
$$ {\text{Cr}} - {\text{N}} - {{\text{O}}_x}\left( {\text{nuclei}} \right) + {\text{Cr}}{{\text{O}}_{\text{3}}} + {\text{SrO}} \to {\text{SrCr}}{{\text{O}}_{\text{4}}} $$

The mechanism of the Cr deposition process proposed above by Jiang et al. [212] seems more reasonable. It implies that cathode materials, which are free of nucleation agents (such as Mn- and Sr-containing species), would be more tolerant toward the Cr deposition.

Figure 7 schematically illustrates the possible poisoning mechanism of the cathode with chromium species. When Cr2O3(s) is exposed to dry air at temperatures higher than approximately 873 K, CrO3 vapor is easily produced from the surface of Cr2O3(s). Furthermore, the presence of a little amount of water vapor in air significantly increases the vapor pressure of Cr species. The formation of the gaseous chromium oxide and oxyhydroxide species can be expressed as Eqs. 16, 17, 18, 19 [193]. The Cr content is also correlated with oxygen ion vacancy concentration in the cathode materials.

$$ {\text{2Cr}} + {\text{1}}.{\text{5 }}{{\text{O}}_{\text{2}}}_{\left( {\text{s}} \right)} \to {\text{C}}{{\text{r}}_{\text{2}}}{{\text{O}}_{{\text{3}}\left( {\text{s}} \right)}} $$
$$ \frac{1}{2}{\text{C}}{{\text{r}}_{\text{2}}}{{\text{O}}_{{\text{3}}\left( {\text{s}} \right)}} + \frac{3}{4}{{\text{O}}_{\text{2}}}_{\left( {\text{g}} \right)} \to {\text{Cr}}{{\text{O}}_{{\text{3}}\left( {\text{g}} \right)}} $$
$$ {\text{C}}{{\text{r}}_{\text{2}}}{{\text{O}}_{\text{3}}} + {{\text{O}}_{\text{2}}}_{\left( {\text{g}} \right)} + {{\text{H}}_{\text{2}}}{{\text{O}}_{\left( {\text{g}} \right)}} \to {\text{2 Cr}}{{\text{O}}_{\text{2}}}{\left( {\text{OH}} \right)_{\text{2}}}_{\left( {\text{g}} \right)} $$
$$ 2{\text{ Cr}}{{\text{O}}_2}{\left( {\text{OH}} \right)_2}_{\left( {\text{g}} \right)} + 6{\text{ }}{{\text{e}}^{-} } + 3{{\text{V}}_{\ddot o}} \to {\text{C}}{{\text{r}}_2}{{\text{O}}_3} + {\text{ }}3{\text{ O}}_{\text{O}}^\times + 2{\text{ }}{{\text{H}}_2}{\text{O}} $$
Fig. 7
figure 7

Possible poisoning mechanism of the cathode with chromium species

Development of chromium-resistant materials and processes for forming protective coatings

Zhen et al. [210] found that the Cr deposition processes at LSM and LSCF cathodes of SOFCs are kinetically limited by the nucleation reaction between gaseous Cr species and nucleation agents. In the case of the LSM electrode, it is SrO species segregated at the electrode surface. They also found that LaNi0.6Fe0.4O3 and La0.6Ba0.4Co0.2Fe0.8O3 electrodes are highly tolerant toward Cr deposition and are much more stable than LSM and LSCF cathodes under SOFC operation conditions in the presence of a Fe–Cr metallic interconnect.

Recently, Chen et al. [212] found that chromium deposition depends strongly on the Ba composition in \( \left( {{\text{L}}{{\text{a}}_{0.6}}{\text{S}}{{\text{r}}_{0.4 - x}}{\text{B}}{{\text{a}}_x}} \right)\left( {{\text{C}}{{\text{o}}_{0.2}}{\text{F}}{{\text{e}}_{0.8}}} \right){{\text{O}}_3}\left( {\text{LSBCF}} \right)\left( {0 \leqslant x \leqslant 0.4} \right)\,{\text{cathode}} \). For the case of LSCF (x = 0), the Cr deposition is most significant, completely covering the surface of the electrode after a polarization at 200 mA cm−2 for 1,200 min. As the Sr content in LSBCF decreases (i.e., Ba content increases), the Cr deposition was reduced significantly. BaCrO4 is the main phase of the Cr deposit on the surface of the LSBCF when x reaches 0.4. The LSBCF electrodes show similar electrocatalytic activities toward the O2 reduction reaction in the absence of Fe–Cr interconnects and at early stages in the presence of Fe–Cr interconnects. Their results clearly show that significant differences in the Cr deposition are closely related to the Sr and Ba content and are not related to the electrochemical activities of the LSBCF cathodes.

Komatsu et al. [213] investigated the chemical reactivity of LaNi0.6Fe0.4O3 with Cr2O3 at 1,073 K. Their results indicated that LNF shows much better chemical stability against Cr poisoning. They also investigated the effect of Cr poisoning on various cells consisting of either LaNi0.6Fe0.4O3 or La0.8Sr0.2MnO3 as a cathode and either yttria-stabilized zirconia or alumina-doped scandia stabilized zirconia (SASZ) as an electrolyte [151]. They found that the cathodic overpotential was almost the same for cells with LNF cathodes, either in the presence or absence of Cr-containing alloy (Inconel 600). LSM cathodes in the presence of Inconel 600 exhibited a very steep increase in overpotential curves after applying a current. These results indicate that the LNF cathode is electrochemically stable for Cr poisoning compared with that of LSM cathodes. For Cr poisoning, the case of LSCF is similar to that of LSM [213], although LSCF is also a promising cathode material for IT-SOFCs especially when the metallic interconnects were coated with a protective barrier to prevent the volatilization of Cr species.

Chromium poisoning effects on Sm1-x Sr x CoO3 and \( {\text{B}}{{\text{a}}_{0.{\text{5}}}}{\text{S}}{{\text{r}}_{0.{\text{5}}}}{\text{C}}{{\text{o}}_{0.{\text{8}}}}{\text{F}}{{\text{e}}_{0.{\text{2}}}}{{\text{O}}_{{\text{3}} - }}_\delta \) have not been reported yet. Considering the results summarized above for LSM and LSCF, SrCrO4 may also form on the surface of SSC and BSCF in the presence of a Cr source thus leading to a performance degradation of cells.

Kim et al. at PNNL [214] investigated the Cr poisoning effects of the Mn-containing ferritic stainless steel on the degradation behavior of various cathodes such as LSCF, LSF, and LNF. Their results show LSCF was the most reactive with Cr species, forming the SrCrO4 phase throughout the entire cathode layer. In the case of LSF, Cr species blocked the “burn-in” behavior, of which the power density increased with time, possibly induced by Pt migration from cathode current collector to the cathode/YSZ interface. Pt was not detected in the cell tested with Crofer22, which showed a fade in power density with time. LNF, which does not contain Sr, revealed the least interaction with Cr among the tested cathodes. However, Cr species were still deposited around the cathode/electrolyte interface leading to Cr poisoning and degradation of cell performance.

It can thus be concluded from the results described previously that cathodes free of nucleation agents (Mn- and Sr-containing species), such as LNF, show the least interaction with Cr species and are the most tolerant toward Cr deposition among the tested cathodes. However, the chromium-poisoning issue cannot be solved only by development of new chromium-tolerant cathode materials. Therefore, protective coatings for metallic interconnectors, especially the spinel coatings of (Co,Mn)3O4 and (Mn,Cr)3O4, are also being explored in order to reduce the Cr volatilization. Spinel coatings may be an effective and feasible approach to solving chromium-poisoning issues during long-term operation. Chen et al. [215] at Lawrence Berkeley National Laboratory showed that Mn–Co-spinels are very promising coating materials with regard to the suppression of Cr2O3 subscale growth and serve to increase electronic conductivities of oxide scales. Kurokawa et al. [216] found La0.65Sr0.3MnO3, La0.6Sr0.4Co0.8Fe0.2O3, and MnCo2O4 coatings, prepared from submicron powders, decreased chromium volatility by a factor of 21–40 at 800 °C.

Various processes have also been used to deposit protective coatings on interconnect alloys in order to protect cathodes from chromium poisoning and improve metallic interconnect stability; Yang et al. [217] at Pacific Northwest National Laboratory fabricated manganese cobaltite spinel protection layers with a nominal composition of Mn1.5Co1.5O4 on ferritic stainless steel by slurry coating technology. The results of thermal, electrical, and electrochemical investigations indicated that the spinel protection layers not only significantly decreased the contact resistance but also effectively acted as a barrier to both the oxygen inward and chromium outward diffusion. A long-term thermal cycling test demonstrated excellent structural and thermomechanical stability of these spinel protection layers. Other developed technologies for depositing various protective coatings on interconnect alloys include large area-filtered arc deposition (LAFAD) and hybrid-filtered arc-assisted electron beam physical vapor deposition (FA-EBPVD) technologies [218, 219] as well as pulsed DC magnetron sputtering technology [220]. Significantly improved thermal stability, low and stable ASR values, and a minimal Cr volatility were observed with coated alloy interconnects compared with the uncoated ones.

Summary and outlook

In summary, state of the art of cathode development in SOFCs has been reviewed in this paper with the main focus being the advantages and existing constraints of various cathode materials. In terms of the composition, oxygen nonstoichiometry and a presence of defects have a great effect on the ionic and electronic transport properties of cathode materials. Sr-doping at the A-site or Sc-substitution at the B-site in perovskite materials enhances the ionic conductivity while Fe and/or Co doping on the B-site increases the electronic conductivity. However, Co doping also has great influence on the TEC of cathode materials. Alternatively, doping with a cation that contains multiple valence states, such as, Pr or Sm, tends to improve the electronic conductivity. Although the electrochemical performance of some alternative cathode materials is significantly higher than that of conventional LSM cathodes, problems persist in the areas of chemical stability and thermal expansion match with other cell components. Usually, a trade-off between electrochemical performance and thermal expansion may be necessary to identify an optimum cathode composition. In addition, a diffusion barrier layer is also necessary to prevent the elemental interdiffusion at electrolyte–cathode interfaces. In terms of the microstructure, TPBs have a great effect on the electrochemical performance of cathode materials, and it has been shown that the method of manufacture has a large impact on the final microstructure. Electrode materials with a porous structure stable at the necessary fabrication and operating conditions are desirable for improving the performance of SOFC cathodes. Furthermore, long-term stability is a significant factor to be considered for practical SOFC stack applications.

When chromium-containing alloys are used as components in IT-SOFCs, the migration of chromium via chromia scale volatilization into SOFC cathodes can lead to a severe degradation of cell performances. Among the well-studied cathode materials such as LSM, LSF, LSC, LSCF, LNF, SSC, and BSCF, it has been found that LNF, free of nucleation agents (e.g., Mn and Sr), is less reactive toward chromium species and has a better tolerance against poisoning than other cathode materials. LNF represents a promising cathode material for IT-SOFCs with resistance to chromium-poisoning. In addition, protective coatings for metallic interconnects are also an important step in order to reduce the Cr volatilization and prevent the oxidation of interconnector materials. Protective coatings may be an effective approach to solving the chromium-poisoning issue during long-term operation.

More fundamental studies on the reaction mechanisms and the kinetics of oxygen reduction in SFOC cathodes are highly desirable and the work should be intensified in the future. By means of in situ spectroscopy technology coupled with impedance spectroscopy measurements and the use of thin film or microelectrodes with well-defined geometry, it is possible to obtain new insights into the electrochemical oxygen reduction reaction mechanisms. This understanding will lead to the achievement of a rational design in new cathode materials and their microstructure. In addition, quantum chemical calculations may be an effective approach to elucidate the interactions between O2 and the cathode since this technique can provide electronic structure, geometrical parameters, and energetics of bulk and adsorbed intermediate species. Improving the performance of the existing cathode materials can also be achieved through optimization of processing, composition, and microstructure. Work on the optimization of interfacial properties and stability is also necessary for stack development.