Progress in proton‐conducting oxides as electrolytes for low‐temperature solid oxide fuel cells: From materials to devices

Among various types of alternative energy devices, solid oxide fuel cells (SOFCs) operating at low temperatures (300‐600°C) show the advantages for both stationary and mobile electricity production. Proton‐conducting oxides as electrolyte materials play a critical role in the low‐temperature SOFCs (LT‐SOFCs). This review summarizes progress in proton‐conducting solid oxide electrolytes for LT‐SOFCs from materials to devices, with emphases on (1) strategies that have been proposed to tune the structures and properties of proton‐conducting oxides and ceramics, (2) techniques that have been employed for improving the performance of the protonic ceramic‐based SOFCs (known as PCFCs), and (3) challenges and opportunities in the development of proton‐conducting electrolyte‐based PCFCs.


| INTRODUCTION
The extraordinary concerns regarding the depletion of fossil fuel resources and the emission of greenhouse gases have promoted considerable research activities on clean, sustainable, and highly efficient energy conversion and storage technologies. Fuel cells can directly convert the chemical energy in a fuel (eg, hydrogen and hydrocarbons) into electricity with high conversion efficiencies and low emissions. Based on types of electrolytes, fuel cells have been classified into alkaline fuel cell (AFC), proton exchange membrane fuel cell (PEMFC), phosphoric acid fuel cell (PAFC), molten carbonate fuel cell (MCFC), and solid oxide fuel cell (SOFC). 1 AFCs, PEMFCs, and PAFCs, which are operated at low temperatures (<250°C), require pure hydrogen fuel to maintain stable performance, because hydrogen must be activated over noble metal catalysts (such as Pt) at low temperatures and these catalysts are easily poisoned by impurities (such as CO). [2][3][4] Therefore, hydrocarbons (such as natural gas) may not be directly used for these fuel cells. In contrast, MCFCs and SOFCs are operated at high temperatures (650-1000°C). 5,6a,6b Many catalysts are very active for hydrocarbons at high temperatures. Furthermore, catalysts have high tolerance for poisons at high temperatures. 7,8 MCFCs use the molten salt of lithium potassium carbonate as an electrolyte at a high temperature, which can in situ convert fossil fuel to hydrogen-rich gases in the anode, eliminating the external production of hydrogen. However, carbonate electrolytes can easily corrode the anode and cathode, accelerating the degradation of MCFC components and decreasing the durability and cell life. 9 Therefore, in the area of high-temperature fuel cells, attention has been focused on SOFCs with metal oxide electrolytes. 1,10 The all-solid structure of SOFCs reduces the possibility of erosion, enabling the assembly of SOFC stacks/modules. Because SOFCs are made entirely of solid materials, they are not limited to the flat plane configuration and are also designed as rolled tubes. 11

| 985
ZHANG ANd HANG HU The high operating temperature of SOFCs provides several advantages. 6a,6b,7,12 SOFCs can be run on a variety of fuels including natural gas. Noble metal catalysts (eg, Pt) are not required. Furthermore, waste heat from SOFC systems may be captured and reused, increasing the theoretical overall efficiency to as high as 70%-75%. Nevertheless, high operating temperatures also generate many challenges for SOFC systems. 13−15 One of them is the carbon dust deposited on the anode, which slows down the internal reforming process. It is also very difficult to well connect different components in SOFCs at high temperatures. Furthermore, high operating temperatures require the excellent thermal stability of materials, increasing the cost of the system. Therefore, decreasing the operating temperatures of SOFCs has become an important topic. 16 It is well-known that the most common metal oxide electrolyte in SOFCs is a ceramic material called yttria-stabilized zirconia (YSZ), a typical oxide-ion conductor. YSZ and other oxide-ion conductors require a high temperature (750°C or above) to achieve desired ion conductivity. 1 In contrast, proton conductors exhibit higher ion conductivity at much lower temperatures (350-600°C) due to relatively low activation energy for proton migration in solid oxides. 17 Therefore, it is very attractive to develop SOFCs using a protonconducting ceramic oxide as the electrolyte, which are also called as protonic ceramic fuel cells (PCFCs). However, only few reports show the peak power densities of PCFCs above 400 mW cm −2 at 500°C, 18−22 which is a benchmark for traditional low-temperature SOFCs (LT-SOFCs). 23 −25 Furthermore, some protonic ceramic electrolytes may react with CO 2 , challenging their utilization at a lower temperature with carbon-containing fuels. 26 This has stimulated intensive efforts to develop more efficient PCFCs, leading to the rapidly increased number of publications ( Figure 1), particularly since 2000.
In recent years, several interesting review articles were published on the development of proton-conducting electrolytes for PCFCs. 27−33 However, each of them has focused on a specific material system or fabrication technique. Li et al reviewed the effects of sintering aids on the electrical conductivity of proton-conducting oxides without description about fuel cell (device) performances. 33 Kim et al summarized the applications of proton-conducting oxides in fuel cells, electrolysis cells, hydrogen purification, and ammonia synthesis with limited discussions about electrolyte materials. 30 So far, there has not been a comprehensive review to summarize recent advances in proton-conducting electrolytes and PCFCs at the level from materials to devices. This has encouraged us to write this review article: (1) Various types of proton conductors will be introduced, along with the principle and mechanism of proton conduction in crystal structures; (2) Strategies to tune the proton conductivity and chemical stability of proton-conducting oxides will be summarized, followed by solutions of sinterability and electron leakage issues in proton ceramics; (3) Strategies of improving PCFC device performance from traditional methods to innovative concepts will be highlighted; (4) The challenges and opportunities in the development of proton-conducting electrolytebased PCFCs will be discussed in the last section.

| T YPES OF PROTON CONDUCTORS
Proton conductors can be classified based on their working temperatures. Here, we summarize and discuss advances in high-temperature solid oxide proton conductors, which are generally employed as proton-conducting electrolytes in SOFCs. These materials are also known as high-temperature proton conductors (HTPCs) to distinguish them from protonconducting polymer materials used in PEMFCs that operate at temperatures of ˂100°C.

| A 3+ B 3+ O 3 -type perovskite
The A 3+ B 3+ O 3 -type perovskite materials were also discovered as promising proton conductors, for example, doped LaYO 3 , LaErO 3 , and LaScO 3 . 56−59 Consequently, enhanced proton conductivity in Sr-doped perovskite La 0.9 Sr 0.1 MO 3-δ (M = Sc 3+ , In 3+ , and Lu 3+ ) was reported by Nomura and coworkers. 60 The substitution of La by Sr in the perovskite is favorable for generating oxygen vacancies. The produced vacancies can trap water, and then, protons can be easily introduced into these oxides (as described by Equation  1). Among these materials, the compound with Sc 3+ at the B site shows the highest proton conductivity of 4 × 10 −3 S cm −1 due to the decreased size of the B-site cation from Lu 3+ (0.861 Å) to Sc 3+ (0.745 Å). In general, if the size of the B-site cation increases, oxygen sites would split gradually, and the dimensionality of proton conduction pathway would reduce from 3-D to 1-D, resulting in a decrease in proton concentration and mobility. 61 The conductivity could be further enhanced with more Sr-dopant (La 0.8 Sr 0.2 ScO 3-δ ). 62 At 600°C, the compound with x = 0.2 (ie, ratio of Sr to La) shows a proton conductivity of 6 × 10 −3 S cm −1 , which is 1.5 times of that with x = 0.1. The proton conductivity of the optimized La 0.8 Sr 0.2 ScO 3-δ material is significantly higher than that of A 2+ B 4+ O 3 -type SrCe 0.95 Yb 0.05 O 3-δ and very close to the value of BaCe 0.95 Y 0.05 O 3-δ at the same temperature. 63 Moreover, the tolerance factor (0.91) of LaScO 3 is higher than that (0.89) of A 2+ B 4+ O 3 -type SrCeO 3 , showing a higher structural stability. Therefore, the LaScO 3 -type perovskite could be an alternative to cerate-based materials for the stable operation of SOFCs.
(1)  Figure 3 gives the crystal structures of the Ba 3 Ca 1+x Nb 2-x O 9-δ based on different structural models (Fm ‼ 3 m, R ‼ 3m, and P ‼ 3m1), showing the possible conduction pathway for protons. 65 These kind of materials were demonstrated to be good HTPCs on exposure to H 2 O-containing atmospheres. The conductivities of them are expected to achieve the same range as those for doped SrCeO 3 and BaCeO 3 . Additionally, in contrast with cerate-based perovskites, these new structured materials do not show electronic conduction after treatment in highly reducing atmospheres and thus deserve more considerations as alternative electrolyte materials to the ABO 3 -type perovskite for enhancing the electrochemical performance of fuel cell devices.
Some other materials are characterized by a disordered oxygen sublattice, in which the oxygen sites are only partially occupied. Typical representative of these perovskite-related oxides are Sr 6-2x Nb 2+2x O 11+3x , Sr 6-2x Ta 2+2x O 11+3x , and (Ba 1y Ca y ) 6 Nb 2 O 11 . 66,67 They are able to incorporate up to one H 2 O molecule per formula unit. More importantly, their conductivities are purely protonic at temperatures up to 600°C. However, the shortcoming of these materials is relatively low conductivity (less than 10 −3 S cm −1 ), which is insufficient for efficient fuel cells. Another perovskite-related protonconducting material is Ba 2 In 2 O 5 . 68−72 Its brownmillerite structure is derived from a perovskite-type crystal, where one sixth of the oxide ions are missing. However, its oxide-ion conductivity increases significantly when the temperature is higher than 900°C. 73,74 Sufficient proton conductivity may be obtained and stabilized under lower temperatures by partially replacing Ba and/or In, for example, Ba 2 (In 1-x Ti x ) 2 O 5+x and (Ba 1-x La x ) 2 In 2 O 5+x systems. 75,76 The further research on these perovskite-related systems will offer more opportunities for designing new proton conductors.

| Proton conductors with other crystal structures
Proton conduction has been observed in some materials without structural protons, for example, (Li/Na/K) 3 PO 4 and other phosphate-based salts. 77−79 Because the larger La 3+ ion is likely to be beneficial for proton mobility, LaPO 4 -based materials are considered as potential proton-conducting electrolytes, which possess excellent ability to dissolve and transport protons with stable crystal structures. Furthermore, Norby and Christiansen observed decent proton conductivity in Ca-and Sr-substituted LaPO 4 . 80 By introducing acceptor substituents with suitable ionic radii (eg, Ca and Sr), a high concentration of charge-compensating protons dissolved from water vapor was obtained. The proton conductivities are 6 × 10 −5 and 3 × 10 −4 S cm −1 at 800°C for Ca-and Sr-substituted LaPO 4 , respectively. These findings encourage further studies of proton conduction in phosphates. However, the doped LaPO 4 has limited solubilities for dopant ions. Excess doping would lead to the formation of secondary phases and strong segregation. In addition to LaPO 4 -based materials, other types of phosphates, for example, lanthanum polyphosphate (LaP 3 O 9 ) and lanthanum oxophosphate (La 7 P 3 O 18 ) received more attentions. 81,82 Both of them have considerable proton conductivity in wet atmospheres, achieving ~3 × 10 −4 S cm −1 at 700°C with Sr as the dopant, which is comparable to some perovskite-type materials like In-doped CaZrO 3 . Slater and coworkers demonstrated high-temperature proton conduction in a new phase of La 1-x Ba 1+x GaO 4-x/2 that contains tetrahedral Ga, with significantly higher conductivity than that of other phases containing tetrahedral ions, for example, the doped LaPO 4 . 83 Proton migration pathway around a GaO 4 tetrahedron is shown in Figure 4, where the proton follows a curved path between two intratetrahedral oxide ions under repulsive interactions with adjacent cations. 84 Norby's group subsequently observed proton conduction in a series of acceptor-doped rare-earth ortho-niobates and ortho-tantalates, with a general formula of RE 1-x A x MO 4 (RE = La, Nd, Gd, Tb, Er, or Y; A = Ca, Sr, Ba; x = 0.01-0.05; M = Nb, Ta). 85 These materials have the mixed protonic, native ionic and electronic conduction properties depending on different conditions. The proton conductivity is dominant under humidified conditions below 800°C, and the highest proton conductivity of ~10 −3 S cm −1 was observed for Ca-doped LaNbO 4 , showing potential as an electrolyte for PCFCs. More importantly, these materials possess relatively high proton conductivity among oxides that do not contain Ba or Sr, and thus, they are very suitable for fuel cells operating in CO 2 -containing atmospheres. Notably, the major drawback of these materials is relatively low conductivity compared to perovskite-type cerate-based materials. Future efforts are necessary to solve the issue for practical PCFC applications.

| Enhancement in proton conductivity
There are various compounds that exhibit remarkable proton conductivity, with an acceptor-dopant concentration of 10 mol% in most cases. 86 Using the ABO 3 -type perovskite as an example, the level of conductivity is dependent on the nature of atoms in the perovskite: the conductivity generally increases with decreasing electronegativity of A and B elements. On the other hand, Kreuer showed that the formation enthalpy of proton defects decreases with decreasing electronegativity of elements, for example, from Ca to Ba (at A site) and from Ti to Ce (at B site). 86 According to these principles, the most developed proton-conducting systems are barium cerate (BaCeO 3 ) and barium zirconate (BaZrO 3 ) doped by divalent or trivalent elements. 86 3.1.1 | Single-doping strategy It is possible to obtain sufficient defects in the sublattices of cerates and zirconates by a single doping. For example, by introducing an alkaline-earth element as dopant into the zirconate, the formation of oxygen vacancies can be represented by the Kröger-Vink notation 87 : where M represents a divalent cation, V ⋅ ⋅ o a compensating oxygen vacancy, and R a trivalent cation. These equations indicate that the oxygen vacancies are formed by a compensating force of negatively charged defects.
All the elements of the lanthanide (Ln) series have been explored as dopant for the barium cerate: Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu. 88−99 The introduction of the Ln dopant can improve the conductivity of BaCeO 3based materials. Namely, the conductivity of BaCe 1-x Ln x O 3-δ increases with increasing x, achieving the maximum at x = 0.1-0.25. This is a typical trend that has also been observed for other doped oxides. 100 However, the dissolution of Ln 2 O 3 into the perovskite structure forms free oxygen vacancies only at relatively low dopant concentrations, thus the decrease of conductivity with a further increased loading of Ln 2 O 3 is observed due to decreased oxygen vacancies.
The conductivity of BaCe 0.9 Ln 0.1 O 3-δ (Ln = La, Nd, Sm, Gd, Yb, Tb, or Y) is also dependent on structural parameters. It is generally accepted that lower distortion of the lattice and larger free volume can promote the ionic conductivity in perovskites. The degree of distortion is given by (and inversely proportional to) the Goldschmidt tolerance factor (t):

ZHANG ANd HANG HU
where r A and r B are the ionic radii of cations located at A and B sites, respectively; r O the ionic radius of oxygen. It was found that with increasing ionic radius of the dopant, the tolerance factor decreases. 101 On the other hand, the free volume, which is the difference between the volume of the unit cell and the volume of the constituent ions, increases with increasing ionic radius of the dopant. Thus, dopants with a smaller ionic radius generally exhibit a lower free volume and thus a decreased ionic mobility, whereas the smaller ionic radius tends to induce a larger tolerance factor and a higher ionic mobility. Therefore, Gd, Sm, and Y dopants, which possess the intermediate ionic radius (see Figure 5), are expected as the best candidates to optimize the ionic conductivity of barium cerates, as revealed by Amsif (20 μm). 103 This indicates the better proton conductivity of the In and Y codoped electrolyte than the single-doped one. However, Zhang et al found that the proton conductivity of Ln (Y, Nd, Gd, and Yb) and In co-doped samples (BaCe 0.8-x Ln 0.2 In x O 3-δ ) decreased with increasing In content. 104 This happened because the increase of In content decreases the lattice volume and proton concentration, hampering the proton diffusion. The Y and Nd co-doped barium cerate at Ce site (BaCe 0.8 Y x Nd 0.2-x O 3-δ ) was investigated by Su et al. 105 As x increases in BaCe 0.8 Y x Nd 0.2-x O 3-δ , its conductivity increases to the maximum at x = 0.15 and then decreases. Furthermore, Lee et al observed better phase stability of co-doped BaCe 0.8 Y x Nd 0.2-x O 3-δ than that of single-doped BaCe 0.8 Y 0.2 O 3-δ , but did not see the increase in conductivity with increasing x values. 106 This happened probably because the BaCe 0.8 Y x Nd 0.2-x O 3-δ samples synthesized by these two groups possess different microstructures. Fu et al explored BaCe 0.85 Y 0.1 Nd 0.05 O 3-δ as an electrolyte for PCFCs. 107 A power density of 237 mW cm −2 was obtained at 700°C based on a 50-μm electrolyte with wet ethane fuel. Additionally, Zhang and Zhao reported Sr and Nd co-doped BaCeO 3 at both A and B sites (Ba 1-x Sr x Ce 0.9 Nd 0.1 O 3-δ ). 108 When Srcontent increases from x = 0 to x = 0.2, the structural distortion of the perovskite also increases, leading to a significant decrease in oxide-ion conductivity, but increasing the proton conductivity contribution to the total conductivity in both hydrogen-rich and oxygen-rich atmospheres. Therefore, Sr and Nd co-substituted BaCeO 3 materials are considered as promising proton conductors for PCFCs if the composition can be further optimized.

| Increase in chemical stability
The application of BaCeO 3 -based electrolytes in intermediatetemperature electrochemical devices is limited by poor stability in CO 2 and water vapor atmospheres. 109,110 This promoted attempts to improve their stability. In this regard, the codoping strategy plays a crucial role. Radojkovic´ et al examined the structure change of Y and Nb co-doped BaCeO 3 . 111 They prepared BaCe 0.9-x Nb x Y 0.1 O 3-δ (x = 0-0.05) materials by solid-state reaction method at 1000°C, generating a stable single-phase crystal at any Nb concentrations. Further sintering of the as-constructed ceramic tablet at 1550°C ensured a stable microstructure with a relative density of 92%-96%. The co-doped samples showed improved stability in pure CO 2 at 700°C. However, the grain size of the samples decreased from 1.6 to 1.1 μm as the Nb content increased, which caused the decrease of the conductivity from 1.0 × 10 −2 to 0.3 × 10 −2 S cm −1 in wet hydrogen at 600°C. This is because more grain boundaries were produced, acting as a barrier of charges. In the sintered ceramics, both the grain interior and the grain boundaries contribute to the overall electrical performance. This means even though the grain interior is tuned to achieve enhanced proton conduction through the bulk, a high grain F I G U R E 5 Free volume and tolerance factor in various BaCe 0.9 Ln 0.1 O 3-δ (Ln = La, Nd, Sm, Gd, Y, and Yb) samples as a function of ionic radius. Reprinted with permission from Ref. 101 Copyright 2011 Elsevier boundary resistance may limit the practical use of the material. Increasing the sintering temperature is an effective strategy to obtain larger grains with fewer boundaries. However, the chemical composition of oxides could be changed under high-temperature conditions, as we discussed in section 3.3. Although a high Nb content negatively influences the conductivity, proton conduction is dominated rather than oxideion conduction because the enthalpy for vacancy hydration is less exothermal with the increasing Nb content.
The effects of Nb and Sm co-doping on the stability of BaCeO 3 -based ceramics were also explored. Xie et al synthesized BaCe 0.8-x Nb x Sm 0.2 O 3-δ system (x = 0-0.1) by solid-state reaction method. 112 The adding of Nb into BaCe 0.8 Sm 0.2 O 3-δ led to great stability in CO 2 and H 2 O-containing atmospheres, but decreased its conductivity due to the decline of free OH• vacancies. Similarly, Nb dopant can enhance the stability of BaCe 0.8 Gd 0.2 O 3-δ in the presence of CO 2 and H 2 O, with a slightly lower conductivity as well. 113 The Sn and Y co-doped BaCe 0.8-x Sn x Y 0.2 O 3-δ (x = 0-0.25) are very stable under fuel cell operating conditions and have acceptable proton conductivity in the temperature range of 600-1000°C. 114 More studies on the Sn-containing samples revealed that these materials can achieve higher relative densities of above 96% compared with single-doped BaCe 0.8 Y 0.2 O 3-δ (88.3%). 114 The chemical stability of barium cerate materials can also be improved by partial substitution of Ce with Ti, In, and Zr. 115,116 For example, Ti 4+ was doped into B site of BaCe 0.8 Y 0.2 O 3-δ to modify its chemical stability. 115 Phase identification demonstrated that the resultant BaCe 0.6 Ti 0.2 Y 0.2 O 3-δ remained the pervoskite structure at a Ti-doping level of 20%. The exposure in 94% N 2 + 3% CO 2 + 3% H 2 O at 700°C for 10 hours did not change its structure. In contrast, significant decomposition was observed in single-doped BaCe 0.8 Y 0.2 O 3-δ . The conductivity of the Tidoped material reached 7.2 × 10 −3 S cm −1 at 700°C in wet H 2 condition, only slightly lower than the value of single-doped BaCe 0.8 Y 0.2 O 3-δ (8.5 × 10 −3 S cm −1 ). Electrochemical performance of a fuel cell based on a 20-μm BaCe 0.6 Ti 0.2 Y 0.2 O 3-δ electrolyte showed a peak power output of 244 mW cm −2 at 700°C. Imperfectly, the open-circuit voltage (OCV) only reached 0.92 V, which is attributed to the partial reduction of Ce and Ti at the anode side, generating electronic defects. Besides, BaCe 0.7 A 0.1 Gd 0.2 O 3-δ (A = In or Zr) system was also investigated. 116 All of the samples showed pure, stable perovskite structure. Particularly, the In-doped sample reached higher conductivity than the Zr-doped one in the range of 300-800°C because of better sintering densification.

| Improvement in sinterability
Improving the sinterability of proton-conducting ceramics is important for achieving high-performance electrolytes.
For example, the chemically stable yttrium-doped barium zirconate is refractory, which requires a very high sintering temperature (above 1700°C) and a long time to obtain both a single phase and a dense microstructure. Such a high temperature may accelerate the evaporation of barium from the oxide, leading to a decreased conductivity. It is also difficult to select ideal porous anode or cathode that can be combined with the refractory electrolyte for co-sintering at very high temperatures. For practical use, improvement of the sintering behavior of proton ceramics at reduced temperatures to maintain its high conductivity and chemical stability is critical. However, in a conventional solid-state reaction, solid precursors need to be fired at a high temperature (such as 1600°C) for more than 10 hours. 117,118 The required sintering temperature may be decreased using nanosized powders. Additionally, many sintering aids were successfully developed to improve the sinterability of proton ceramics. 119−122 However, some additives, for example, TiO 2 , MgO, MoO 3 , Al 2 O 3 , and Bi 2 O 3 have negative effects on the conductivity of the sintered ceramics.
Alternatively, dopants can affect the sinterability of proton-conducting ceramics. In general, it is difficult to achieve high conductivity, high sinterability, and excellent chemical stability simultaneously by using a single dopant. Using two or more dopants into the B site were found to be an effective strategy for improving the sintering properties and electrical performance of the perovskite-type proton ceramics. 123−127 For example, Liu and co-workers achieved the improvement in sinterability and proton conductivity for Sm and Y co-doped BaCeO 3 . 128 Doping BaZrO 3 with double dopants at the B site was also reported as an effective way to improve the sinterability. 129 Specifically, the performance of BaZr 0.6 M 0.2 Y 0.2 O 3-δ with various M (M = Zr 4+ , Ce 4+ , Pr 3+ , Nd 3+ , Sm 3+ , and Gd 3+ ) in addition to the Y-dopant were comparatively evaluated. The results show that (1) all the studied rare-earth dopants can improve the sinterability of materials; (2) the chemical stability of the compounds increases in CO 2 atmosphere with decreasing the ionic radii of M. Among various compositions, the Nd 3+ doped sample possesses the highest proton conductivity (8.56 × 10 −3 S cm −1 at 600°C) along with excellent sinterability and resistance to CO 2 . Besides, the effects of dopants on the conductivity and sinterability of BaCe 0.85 Ln 0.15 O 3-δ (Ln = Gd, Y, Yb) ceramics has been evaluated. 130 Y-doped sample shows the highest conductivity, while the sinterability depends on the ionic radius of the selected dopant. The obtained grain size of the ceramics decreases with decreasing ionic radius of the dopant. Furthermore, the sinterability and chemical stability of Y and In co-doped barium cerates (BaCe 0.7 In x Y 0.3-x O 3-δ ) can be improved by increasing In concentration, but the electrical conductivity decreases. 131 This indicates that an intermediate composition is better for keeping a relatively high conductivity and an acceptable pellet density. More exploration of the co-doping strategies is expected to achieve larger grain size and improve the relative density of the proton ceramics.
In summary, the following factors are generally beneficial to the proton conductivity of the sintered ceramics: (1) increase in the dopant concentration (eg, 10-20 mol% in most cases) of the oxide; (2) increase in the average grain size at optimized sintering temperatures; (3) increase in the relative density of the ceramic pellet; and (4) measurement atmospheres under highly reducing and humidified conditions. It is recognized that the sinterability is crucial to achieve the desired electrical performance among these factors. However, the conductivity can also be influenced by the synthesis route and synthesis temperature of the powders, as summarized in Table 1. This means that the conductivity is not always proportional to the relative density or grain size. 132,133

| Minimization of electron leakage
The electron leakage issue also deserves more attentions. BaCe 0.2 Zr 0.7 Y 0.1 O 3-δ is a well-developed solid solution as the proton ceramic system, but it is not a pure proton conductor. The material possesses four kinds of charge carriers: protonic defects (OH Besides, under reducing atmospheres, significant electronic conduction in high Ce-containing samples can be observed, 141 where Ce ions are reduced to a lower oxidation state (ie, Ce 3+ ) and thus electrons can be produced: The generated electronic conductivity can significantly impair device performance, leading to reduced cell voltage and overall efficiencies.
From the perspective of energy efficiency, Nakamura and co-workers proposed that reducing the electrolyte thickness is not always an ideal approach to achieve high-performance PCFCs, since the energy loss becomes more significant in a thinner electrolyte due to internal current leakage. 142 Several strategies have been developed to mitigate the electron leakage in the electrolyte as much as possible. (1) Using an additional electron-blocking layer between the electrode and the electrolyte has been demonstrated as an effective way. For instance, a core-shell structured Ce 0.8 Sm 0.2 O 2-δ (SDC)@ Ba(Ce, Zr) 1−x (Sm,Y) x O 3-δ electron-blocking layer was in situ formed between Ni-BZY anode and SDC electrolyte. 143 The Ba-containing shell effectively protected SDC grains from reduction under fuel cell operating conditions, thus electronic conduction through the electrolyte was totally eliminated. However, this method may cause a larger interface resistance.
(2) The double-layer electrolyte strategy can also be considered, as demonstrated in the bilayered BZCY/SDC structure for SOFCs with high OCVs. 144 These techniques of SOCFs are highly expected to be applicable for PCFC devices. (3) Tuning the composition of proton ceramics is another strategy. Dippon et al evaluated the effects of Ce content on the magnitude of electronic leakage in the BaCe x Zr 0.9-x Y 0.1 O 3-δ electrolytes. 145 Minimized electron leakage was observed in samples with low Ce content. We can conclude that an intermediate composition is preferred for practical applications, where a negligible electronic conductivity and a relatively high protonic conductivity are achieved simultaneously.

ELECTROLYTE-BASED PCFC DEVICES
SOFCs based on proton-conducting electrolytes are known as PCFCs. They have a sandwich structure (the same as conventional SOFCs), in which an anode and a cathode are separated by proton conductors (Figure 6). 19 PCFCs possess higher theoretical electromotive force and electrical efficiency compared with conventional oxide-ion-conducting SOFCs. Water is generated at the cathode instead of the anode, preventing fuel dilution at the anode and achieving efficient fuel managements. Furthermore, PCFCs have higher power output at low temperatures due to higher protonic conductivity, especially below 600°C. They are more suitable than oxide-ion SOFCs for lower operating temperatures.
Currently, the exploration of chemically stable proton conductors for PCFCs aims to achieve long-term, stable cell operation and reduce the costs of fuel cell systems. In addition, many novel strategies have been developed to achieve sufficient protonic conductivity and suppress the electronic conductivity in the electrolyte. Without doubt, the improvement of electrode performance is also important for proton-conducting electrolyte-based PCFCs.

| Development of chemically stable electrolytes for PCFCs
The doped perovskite of SrCeO 3 -based oxide, which can conduct protons under hydrogen-containing atmosphere, was invented by Iwahara et al in 1981. 34 This stimulated great efforts to explore this type of proton conductors with the finding of SrZrO 3 , BaCeO 3 , BaZrO 3 systems, etc. 35 These materials were then widely studied as proton-conducting electrolytes for PCFCs. 146−149 Currently, the most developed proton-conducting electrolytes are based on BaCeO 3 , which have relatively large lattice free volume and relatively high geometrical symmetry, leading to the highest conductivities. 150 For practical applications in PCFC devices, the BaCeO 3 -based perovskites are usually substituted partially by Zr to achieve sufficient chemical stability, along with doping by different elements. (1) The Zr-containing perovskite BaZrO 3 is very stable under fuel cell operating conditions, which possesses similar proton conduction mechanism with BaCeO 3 , but with lower conductivity. The introduction of Zr into the BaCeO 3 achieved an balance between conductivity and chemical stability. 55 (2) Cations of rare-earth elements, such as Yb 3+ , Y 3+ , and Nd 3+ , were typically employed as dopants to introduce more oxygen vacancies into the perovskite structure. Table 2 shows the progress of PCFC device performances by employing the BaCeO 3 -and BaZrO 3 -based proton-conducting electrolytes in the recent decade. Y-doped BaCeO 3 and BaZrO 3 (BCY and BZY) are two representative proton-conducting electrolytes at an early stage ( Table 2). For example, Zunic et al employed electrophoretic deposition technique to prepare dense BCY electrolyte films on Ni-BCY composite anode substrates. 151 The deposition time was tuned to obtain a thickness of <15 μm, achieving a peak power density of 150 mW cm −2 at 600°C ( Figure 7A,B). Soon afterward, Bi et al developed an ionic diffusion strategy to prepare anode-supported BZY electrolyte films, leading to excellent chemical stability and improved performance (169 mW cm −2 at 600°C) ( Figure 7C,D). 152 Furthermore, BCY and BZY have been combined (BZCY) as a high performance, chemically stable electrolyte for PCFCs. Based on this electrolyte, Lee et al developed a gradient anode functional layer made of NiO and BZCY deposited on the BZCY electrolyte ( Figure 8A), obtaining a power density of 240 mW cm −2 at 600°C ( Figure 8B). 153 Both the ohmic and polarization resistances were decreased due to the optimized microstructure of the anode functional layer on the BZCY electrolyte, which holds a continuously gradient interface to avoid the mismatch between electrolyte and anode. Fan 154 Simultaneous transport of intrinsic O 2− and e − as well as extrinsic H + is illustrated in Figure 8C. The H 2 O-uptake capability of this material also facilitated the oxygen reduction reaction (ORR) process between the cathode and the BZCY electrolyte, leading to a power output of 410 mW cm −2 at 650°C with the help of ZnO sintering aid ( Figure 8D). Yuan et al reported a PCFC with the BZCY electrolyte and a Cofree cathode of SrFe 0.95 Nb 0.05 O 3-δ , achieving a peak power density of 428 mW cm −2 at 600°C with a low polarization resistance (0.35 Ω cm 2 ). 155 The fuel cell worked steady under a constant load of 140 mA cm −2 at 550°C for 50 h without degradation. The co-doping of Y and Yb created the renowned protonconducting electrolyte BZCYYb, displaying an improved conductivity and excellent tolerance to S-poisoning and Ccoking. This material has proton defects and oxide-ion vacancies, thus allowing fast transport of both ions (H + and O 2− ). 156 Kim Figure 9B). 22 The dense thin-film possesses a sufficient solubility for H + ions, which is beneficial to ORR at the oxidant/cathode interface. Consequently, the power density of the PCFC was greatly enhanced to 1098 mW cm −2 at 600°C ( Figure 9D), with a long-term operation of 700 hours. The refractory property of proton ceramic oxides is a major challenge. To solve this issue, An et al prepared a thin, dense BZCY electrolyte by anode-assisted densification, leading to a great breakthrough in PCFC performance and scalability. 21 With the help of a self-supplied transient phase from the anode (due to higher shrinkage of the anode), the BZCY electrolyte was uniformly densified on a Ni-BCZY anode at 1350°C ( Figure 10A). This technique is useful to readily produce large-area PCFCs. The BZCY-based PCFC with a size of 5 × 5 cm 2 delivered an extremely high-power density of 1302 mW cm −2 at 600°C ( Figure 10B). Very recently, Chen et al developed a facile phase-coating strategy to fabricate a microtubular fuel cell based on the BZCYYb electrolyte. 158,159 Anode-substrate, electrolyte, and cathode were formed layer-by-layer on a glass rod using corresponding ceramic slurries followed by sintering ( Figure 11A). With optimized coating times and sintering temperatures,  Figure 11B). We believe these techniques are promising for mass production of commercial PCFCs.

| Novel semiconductor-based electrolytes for PCFCs
Semiconductor materials with proton-conducting properties could play an important role in future advanced PCFC technology. Recently, many semiconductor oxides have been explored as novel electrolytes for efficient fuel cells. 160−167 These new developments from traditional theory to innovative concepts have created a promising research frontier.
A fundamental challenge for the semiconductor electrolyte is the separation of electrons to prevent short circuiting.
Changing the oxidation states of transition elements may suppress the electronic conduction. Based on this principle, a new proton-conducting electrolyte was designed using a transition element-rich, layered-structured semiconductor Li x Al 0.5 Co 0.5 O 2 . 168 Under H 2 /air operating conditions, this material experienced a significant transition, where the Co was reduced to divalent ions. Its electronic conduction was then eliminated due to the electron-insulating CoO bulk. Meanwhile, protons were readily incorporated into the interlayers of the semiconductor. The proton conductivity of this material achieved 0.1 S cm −1 at 500°C, with a power density of 160 mW cm −2 based on a thick (0.79 mm) electrolyte. Additionally, several semiconductor electrolyte systems (eg, SrTiO 3 , 161 TiO 2 , 162 and BaCo 0.4 Fe 0.4 Zr 0.1 Y 0.1 O 3-δ -ZnO composite 164 ) are based on an energy band alignment mechanism to eliminate electronic conductivity, providing new insight into the development of electrolytes. Namely, the energy band structure is carefully designed to tune (suppress) the transportation of electrons through the electrolyte in fuel cell devices, similar to some other semiconductor devices, for example, solar cells. Typically, high OCVs of 1-1.1 V and power densities of 400-650 mW cm −2 can be achieved at a relatively low temperature range of 500-550°C using these semiconductors as the electrolyte for fuel cells. Notably, almost pure proton conductivity was observed in a perovskite nickelate (SmNiO 3 ) 160 and a composite semiconductor of Na x CoO 2 -CeO 2 . 167 Both of them have intrinsic electronic conductivity. However, in hydrogen-containing atmospheres, electronic conduction can be suppressed by a Mott transition that generates an electron-insulating layer in the SmNiO 3 ( Figure 12A,B) or by a local electric field generated with the semiconductor heterojunction in the composite Na x CoO 2 -CeO 2 ( Figure 12C,D). The latter case (composite) is more significant because the local electric field can simultaneously promote proton diffusion along the two-phase interface. 169 The composite semiconductor-based PCFC devices delivered a high-power density of 1000 mW cm −2 at 520°C. 167 Conventional solid-state ionics generally uses the doping strategy to design and create high-performance electrolytes.
However, fast ionic conduction can also be generated in non-doping semiconductors. Recently, a novel non-doping CeO 2 has been designed as an electrolyte, with a CeO 2-δ @ CeO 2 core-shell structure formed by surface hydrogenation. 163 It was found that this material exhibited high protonic conductivity. 165 High-concentration oxygen vacancies in the surface layer (CeO 2-δ ) can promote proton conduction, creating continuous proton shuttles. The CeO 2-δ @CeO 2 electrolyte-based fuel cells have achieved ~700 mW cm −2 at 520-550°C. Proton conduction in natural mineral is also promising. Electrolyte materials using delafossites, 170 hematites, 171 and mixed rare-earth carbonates 172 have shown their potential for developing cost-effective PCFCs. Power densities of 200-400 mW cm −2 were achieved at ~550°C by these mineral electrolyte-based fuel cells and could be further enhanced by their composites with other ionic materials. Proton transport through multi-phase interfaces (eg, CuO, Fe 2 O 3 , SiO 2 , and CaCO 3 ) is the main transport mechanism in these mineral oxides (Figure 13), which enables a low activation energy and rapid transport kinetics. 171

| 997
ZHANG ANd HANG HU

| Cathode materials
Although PCFCs have shown great potential to be operated at lower temperatures than traditional SOFCs, their power output is still insufficient for practical applications, especially in the temperature range of 300-500°C. This is primarily limited by the lack of ideal cathodes that are compatible with proton-conducting electrolytes. As shown in Figure 14, ORR takes place at cathode, where O 2 molecules are converted to O 2− ions by receiving electrons, which further combined with H + to form water. 157 Cathode materials for PCFCs should possess high activity for ORR, allowing simultaneous transport of H + , O 2− , and e − under fuel cell operating conditions, thus extending the active sites at the cathode/electrolyte interface. Mixed ionic-electronic conductors (MIECs) were often selected as cathode for conventional SOFCs because of their excellent electronic conductivity and catalytic activity. However, MIECs are not suitable for PCFC cathode due to their poor proton conductivity, in which the electrochemically active sites are limited to the interface between the cathode and proton-conducting electrolyte ( Figure 14). 157 Thus, the triple-conducting (ie, H + , O 2− , and e − ) cathodes and the composite cathodes (eg, the mixture of MIECs and proton-conducting oxides) have been developed for PCFCs.
Doping is a useful strategy for developing electrochemically active triple-conducting cathode. Different alter-valent cations can be doped into a proton conductor (eg, BaZrO 3 or BaCeO 3 ) to enhance its electrochemical activity for oxide ions and electrons. However, this material showed insufficient protonic contribution with dominant p-type electronic conduction under oxidizing conditions. It possesses proton and oxide-ion conductivities, but its ionic transport numbers were decreased by the doping. Chemically stable BaZrO 3 doped by transition elements (eg, Co and Fe) has been demonstrated as an effective route for designing the triple-conducting cathode. In the BaCo 0.4 Fe 0.4 Zr 0.2 O 3-δ (BCFZ) cathode, both proton and oxide-ion diffusions reduce the polarization resistance of the cathode, achieving significantly lower activation energy than the renowned La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3-δ (LSCF). 174 A power density of 225 mW cm −2 was obtained at 600°C with the BZCYYb electrolyte. Furthermore, Y-doped BaCo 0.4 Fe 0.4 Zr 0.1 Y 0.1 O 3-δ (BCFZY) showed greatly improved ORR kinetics at low temperatures due to increased lattice parameter and free volume. 175 Under the same test conditions and cell configuration, the PCFC power density was improved to ~400 mW cm −2 at 500°C with a stable operation for over 1000 hours. Very recently, Ren et al designed a novel A-site-deficient B x CFZY (Ba x Co 0.4 Fe 0.4 Zr 0.1 Y 0.1 O 3-δ , x = 1, 0.95, 0.9) to promote the oxygen diffusion and hydration kinetics in the BCFZY material ( Figure 15). 176 The asdesigned Ba-deficiency introduced more oxygen vacancies and protonic defects due to the charge compensation mechanism. The PCFC based on the Ba-deficient B x ZFCY cathode (Ni-BZCY/BZCY electrolyte/B 0.9 CFZY cathode) showed a power density of 376.27 mW cm −2 at 500°C, demonstrating an effective strategy to tune the triple-conducting properties of perovskite materials.
Triple-conducting cathodes can also be derived from some MIECs by the absorption of water to obtain desired proton conductivity. Hydration and proton transport properties of the Pr 2-x Sr x NiO 4+δ were explored for PCFC cathodes. 177 The triple-conducting properties of Pr 2 NiO 4+δ without Srdoping (ie, x = 0) were clearly demonstrated under different pH 2 O conditions. The substitution of Pr 3+ by Sr 2+ in the Ruddlesden-Popper Pr 2 NiO 4+δ could enhance its structural stability but with decreased protonic conduction. Large water uptake at high pH 2 O in Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ (BSCF) and PrBaCo 2 O 5+δ was also obtained, showing considerable proton conduction and therefore converting them to potential triple-conducting cathodes. 178 Other double-perovskite MIECs (eg, BaGd 0.8 La 0.2 Co 2 O 6-δ , BaGdCo 1.8 Fe 0.2 O 6-δ , and BaPrCo 1.4 Fe 0.6 O 6-δ ) were also explored as the cathode for PCFCs. 179 Furthermore, it was proposed that layeredstructured perovskites not only provide pore channels for Copyright 2020 The American Association for the Advancement of Science oxide ions, but also exhibit better proton-conducting properties than other perovskite-related oxides due to their unique interlayers. 178,180 Some typical layered perovskite cathodes, for example, NdBa 0.5 Sr 0.5 Co 1.5 Fe 0.5 O 5+δ (NBSCF) and PrBa 0.5 Sr 0.5 Co 1.5 Fe 0.5 O 5+δ (PBSCF), have achieved power densities as high as 1000-1200 mW cm −2 at 650°C with the BZCYYb electrolyte and Ni-BZCYYb anode operated under H 2 /air conditions. 22,157 Notably, the lithiation of some transition metal oxides may introduce extrinsic proton conduction by forming a layered structure. Transition metal cations can be reduced to lower valence states in hydrogen-containing atmospheres, which allows the intercalation of protons in the lattice for charge compensation. 168 A lithiated transition metal oxide, LiNi 0.8 Co 0.2 O 2 , showed higher ORR activity than most commonly used cathodes. 154 Extrinsic proton conduction originated from the layered structure was confirmed by water storage capability (hydration) of this oxide. Following this route, we believe that a series of triple-conducting cathode materials can be developed from cost-effective transition metal or alkaline-earth metal (eg, Ca, Cu, Fe, and Mn) oxides.
Composite cathodes can allow ones to extend the electrochemically active sites at the cathode/electrolyte interface for PCFCs. Typically, they can be formed by mixing a proton-conducting oxide and a MIEC material. 181−183 (1) Sufficient proton conductivity should be introduced into the MIECs; (2) Chemically stable composite cathodes under fuel cell operating conditions are required. Thus, proton conductors of barium cerates are not ideal candidates for the composite cathode, because the atmosphere at the cathode generally contains CO 2 and H 2 O. Zr-containing proton conductors can enhance the stability of the as-formed composite cathode with retaining decent proton conductivity. Progress in nanoparticle-embedded fibrous materials has been applied to manufacturing composite cathodes. For example, nanosized BZCY particles were embedded into the LSCF by electrospinning, forming a one-dimensional structured fibrous composite ( Figure 16). 184 The porous and continuous structure is beneficial for mass transport and charge transfer in the cathode, leading to 239 mW cm −2 at 550°C by a Ni-BCZY/BCZY/ LSCF-BCZY cell configuration. Highly efficient composite cathodes can also be achieved by mixing MIECs with layered double-perovskite oxides, for example, PrBaCo 2 O 5+δ (PBC), SmBaCo 2 O 5+δ (SBC), PrBaMn 2 O 5+δ (PBM), PBSCF, and NBSCF due to their considerable proton-conducting properties. 185−188 For example, single-perovskite Sm 0.5 Sr 0.5 CoO 3-δ (SSC) was combined with the layered SBC as a composite cathode for PCFCs, generating 1570 mW cm −2 at 750°C using the BZCYYb electrolyte, with a low area specific resistance (ASR) of 0.021 Ω cm 2 . 189 High ORR activity can be achieved by the mixtures due to the triple-conducting properties of the layered perovskite, which improves the diffusivity of oxide ions and protons simultaneously. In addition, the operation stability of composite cathode-based PCFCs may be improved by controlling the porosity gradient in the cathode. For example, a gradient porous LSCF-BCZY cathode was developed for BCZY electrolyte-based fuel cells to accelerate the emission of water vapor from the cathode. 190 The porosity of the cathode was increased from ~24% (at the cathode/electrolyte interface) to ~38% (at the cathode surface) by electrostatic slurry spray deposition. Based on the gradient cathode, PCFC operated steady for 100 hours at a load of 175 mA cm −2 at 600°C. A similar study using the Sm 0.5 Sr 0.5 CoO 3-δ (SSC) to replace the LSCF achieved enhanced PCFC performances (over 400 mW cm −2 at 700°C) because of enhanced surface diffusion of oxide ions. 191 In situ exsolution is another promising method for the fabrication of high-performance composite cathodes. A composite cathode material based on the nominal composition of La 0.5 Ba 0.5 CoO 3-δ -BaZrO 3 was prepared by exsolution of a single-phase La 0.3 Ba 0.7 Zr 0.4 Co 0.6 O 3-δ , resulting in a La 0.8 Ba 0.2 CoO 3-δ (40 mol%)-BaZr 0.6 Co 0.4 O 3-δ (60 mol%) composite, where the phase composition was tuned by thermal treatment. 192 The composite consists of larger BaZr 0.6 Co 0.4 O 3-δ grains with the embedded, smaller La 0.8 Ba 0.2 CoO 3-δ grains ( Figure 17A,C), showing a high degree of contiguity between two phases, which is beneficial for fast proton conduction. In contrast, directly mixed (twophase) sample without in situ exsolution shows almost the same grain size and a low degree of mixing ( Figure 17B,D). The reported ASRs of this composite cathode are competitive   193 The Nb-doped catalyst aims to increase the concentration of interstitial oxygen and enhance the structural stability. The resulting composite facilitates oxygen surface exchange and overall cathode reaction. Very recently, a nanoscaled self-assembly technique was explored for the preparation of triple-conducting composite cathodes to address the issues of inhomogeneity between different phases. 194 Namely, a nominal single-phase composition of BaCo 0.7 (Ce 0.8 Y 0.2 ) 0.3 O 3-δ was self-assembled into several phases during calcination and fuel cell operating, where the new phases are homogeneous and closely contacted with each other ( Figure 18B). Different functionalities were achieved by the as-produced three phases, namely, BaCe x Y y Co z O 3-δ (H + /e − conducting), BaCo x Ce y Y z O 3-δ (O 2− / e − conducting), and BaCoO 3-δ (e − conducting). Based on this composite cathode, high-power densities (eg, ~1000 mW cm −2 at 650 ºC, Figure 18C) were obtained, while a robust operational stability was maintained for more than 800 hours at 550 ºC ( Figure 18D). There are some other issues regarding the chemical and thermal compatibility of PCFC cathodes. During the cosintering on the electrolyte surface, the diffusion of cations between cathode and electrolyte may form an interfacial phase, which is detrimental to the long-term operation of PCFCs. For example, several compounds, such as La 0.1 Sr 0.1 Ce 0.8 O 2-δ and La 0.4 Ce 0.6 O 2-δ , were formed between LaMO 3 (M = Mn, Fe, Co) cathode and proton conductor SrCeO 3 during longterm heat treatment at 1150°C. 195 The dissolution of SrO from SrCeO 3 into LaMO 3 produced these interfacial phases, which is driven by the high activity of SrO. To suppress the formation of impurity phases, the Ruddlesden-Popper structured cathodes (eg, La 2 NiO 4 ) have been identified as better candidates for SrCeO 3 -based electrolytes, where the interfacial reaction is less pronounced. 195 On the other hand, the interfacial reaction may enhance the electronic conductivity in some cathodes. Goupil 196 For the Ba-containing compounds, the diffusion of elements from cathode to electrolyte induced significant interfacial electronic conduction, extending the electrochemical reaction region. Lin et al evaluated the compatibility of Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ and PrBaCo 2 O 5+δ cathodes with BaCeO 3 -based electrolytes (BCY/BZCY). 197,198 During the heat treatments at 900-1100°C, Ba 2+ -enriched BSCF, Ba 2+ -deficient BCY, Y 3+ -enriched PBC, and Co 3+enriched BZCY were generated due to the diffusion of cations between electrolytes and cathodes. However, the diffusion of Ba 2+ significantly decreased the proton conductivity of the electrolyte due to a significant blocking effect on proton transfer from electrolyte to cathode. The effects of Ba migration at the cathode/electrolyte interface should be further investigated. Besides, some cobaltite-based cathodes have high thermal expansion coefficients that are usually 2-3 times larger than the values of proton-conducting electrolytes. 45 The big difference generally leads to peeling, lamination, and cracking of the fuel cell components, resulting in the degradation of the whole cell system. To address this issue: (1) Partial substitution of Co by other elements (eg, Ni, Cu, and Fe) can mitigate the drawback of Cocontaining cathodes, achieving reduced thermal expansion and maintaining electrochemical activities. 199 (2) The use of cathode-electrolyte combined composite cathodes, as we discussed above, is also an effective method to achieve thermal compatibility. 200 (3) The development of Co-free cathodes, such as La 0.6 Sr 0.4 Fe 0.9 Cr 0.1 O 3-δ , PrBaCuFeO 5+δ , GdBaCuFeO 5+δ , and GdBaFeNiO 5+δ , 201−204 are favorable for reducing the cost of materials.

| Anode materials
An anode provides the place for hydrogen oxidation reaction (HOR). Furthermore, hydrocarbons (eg, CH 4 ), which are commonly used as fuels, are converted to hydrogen via reforming in PCFCs. The detailed anode reactions (including the HOR) were illustrated in Figure 19. 205 The anode must meet the following requirements: high activity for fuel oxidation, excellent electronic and protonic conductivity, sufficient porosity and mechanical strength (if the PCFC is anode-supported), and decent chemical/thermal stabilities. The most widely studied PCFC anodes are Ni-based cermet anodes. They are made of Ni with proton-conducting electrolytes to increase the active sites for fuel oxidation and improve the compatibility with electrolytes. 8,206 The stateof-the-art Ni-BZCYYb composite has been identified as an excellent PCFC anode with high catalytic activity and enhanced sulfur and coking tolerance under various hydrocarbon fuels. 156 Compared with oxide-ion-conducting SOFCs, PCFCs are less affected by the porosity of an anode, because water is formed at the cathode and thus less porosity is required for water removal from the anode. A low-porosity (37%) Ni-BZCYYb anode without using the pore former achieved a high-power density of 1200 mW cm −2 at 750°C due to sufficient channels of pores formed during the anode reduction. 207 Since the development of Ni-based cermet anodes, especially the mechanism and strategies regarding the tolerance to carbon coking and sulfur poisoning, has been well reviewed in previous literature, 30,208,209 these anodes are not covered here. Therefore, we would like to introduce some new insights on recent development of in situ exsolution techniques for PCFC anodes.
Compared to cathodes, anodes have attracted less attention. Nevertheless, it is important to reduce the anode polarization that causes considerable performance loss at low temperatures. Recent advances in oxide-ion-conducting SOFCs provided feasible strategies for improving the PCFC anode performance. The exsolution of B-site elements from A-deficient perovskite (A 1-x BO 3 ) can significantly enhance the electrochemical activity of SOFC anodes. 210,211 For example, a composite anode, which consists of Pr 0.8 Sr 1.2 (Co, Fe) 0.8 Nb 0.2 O 4+δ backbone with homogenously exsolved Co-Fe nanoparticles, was prepared from perovskite Pr 0.4 Sr 0.6 Co 0.2 Fe 0.7 Nb 0.1 O 3-δ by annealing and hydrogenation, showing almost the same catalytic activity as Ni-based cermet anodes. 210 Similar results were also demonstrated with a perovskite Sr 0.95 (Ti 0.3 Fe 0. 63 Ni 0.07 ) O 3-δ . Exposure to fuel atmospheres led to exsolution and nucleation of uniformly dispersed Ni-Fe nanoparticles on the original perovskite surface, yielding reduced anode polarization resistances. 211 Neagu et al further revealed that the in situ exsolved nanoparticles from a perovskite substrate could be well controlled by tuning the stoichiometry of the perovskite, providing an attractive route for tailoring the functions of the resulting products. 212 216 The anode with exsolved Ni particles achieved higher performances (500-700 mW cm −2 ) at a lower temperature (600 ºC) compared to the Co-exsolved anode. Furthermore, nanoalloys were also designed as catalysts in perovskite-type oxide anodes (without Ni) to enhance anode performance. Namely, Luo and co-workers prepared a double-layered perovskite (Pr 0.4 Sr 0.6 ) 3 (Fe 0.85 Mo 0.15 ) 2 O 7 anode with in situ exsolved Co-Fe particles via annealing the cubic perovskite Pr 0.4 Sr 0.6 Co 0.2 Fe 0.7 Mo 0.1 O 3-δ under slightly reducing conditions. 217 The highly efficient catalysis of Co-Fe alloy in the anode leads to excellent electrocatalytic activity for ethane-fueled PCFCs, with enhanced cogeneration of ethylene as well. The PCFC device with a BZCY electrolyte achieved peak power densities of 496.2 mW cm −2 and 348.84 mW cm −2 at 750°C in H 2 and ethane, respectively. 217 F I G U R E 1 9 Schematic illustration of a fuel cell device showing the detailed anode reaction processes (AFL: anode functional layer; ASL: anode supporting layer; ARL: anode reforming layer). In the anode, CH 4 is firstly reformed to H 2 and CO as fuels, followed by the oxidation reactions of H 2 (HOR) and CO at the AFL. Reprinted with permission from Ref. 205

| CONCLUSIONS AND OUTLOOK
The PCFC, which operates at a low temperature with high performance and reliable durability, is a highly promising technology to address the increasing demand for clean energy production. In this article, progress in proton-conducting electrolyte-based PCFCs has been reviewed from material to device aspects, with the following conclusions and perspectives: Various classes of solid oxide materials, which exhibit intriguing proton-conducting properties, have been explored as electrolytes. The mechanisms of proton conduction were deeply discussed for the commonly used A 2+ B 4+ O 3 -type perovskites (eg, BaCeO 3 , BaZrO 3 -based), A 3+ B 3+ O 3 -type perovskites (eg, LaScO 3 -based), perovskite-related proton conductors (eg, Ba 3 Ca 1.18 Nb 1.82 O 8.73 ), and other proton conductors with different crystal structures (LaPO 4 -based, RE 1-x A x MO 4 -structured, etc). Currently, the A 2+ B 4+ O 3type cerate and zirconate-based perovskites are dominant in practical applications of PCFC devices due to their higher proton conductivity and decent structural stability under optimized conditions. However, the A 3+ B 3+ O 3 -type proton conductors (eg, doped LaScO 3 ) have shown their potential to be an alternative to the cerate/zirconate-based materials, especially for enhancing the operational stability of PCFCs. In these ABO 3 -structured proton conductors (both A 2+ B 4+ O 3 and A 3+ B 3+ O 3 ), protonic defects are generated via absorbed water, where oxide-ion vacancies in these oxides are highly required. Thus, the doping strategy is typically employed to achieve enhanced proton conductivities. Furthermore, the perovskite-related proton conductors provide more opportunities for designing new proton conductors due to their tunable structures, namely, oxygen vacancies can be generated by changing the stoichiometry (eg, increasing the B'/B'' ratio in the A 2 (B'B'')O 6 -type perovskite) to create new proton conductors. Additionally, they are more stable under reducing atmospheres compared with the ABO 3 -type perovskites, showing pure proton conductivity without electronic shortcircuit issue. Other proton conductors without a perovskite structure, including phosphates, niobates, and tantalates, have also been developed. In these oxides, charge-compensating proton flow can be formed from water by introducing dopant ions into their crystals. However, limited solubility of dopant ions is a major barrier for increasing their proton conductivities. Overall, the insufficient conductivity of the second and the third types of proton electrolytes limits their applications. Various techniques are expected to improve the conventional and novel proton-conducting oxides for their applications in practical PCFCs, including structural doping, stoichiometry tuning, novel synthesis methods, and processing routes.
There are several critical issues associated with electrolytes in PCFCs, namely, insufficient proton conductivity at low temperatures, nonideal chemical stability under some operating conditions, sinterability, and electron leakage. Codoping strategy has been explored for achieving a balance between high conductivity and sufficient stability. However, sinterability and electron leakage issues deserve more efforts for the fabrication of electrolyte ceramics: (1) Variables that determine the sintered microstructure of compacted powders should be controlled from raw materials to sintering conditions; (2) Besides conventional strategies to inhibit the electron leakage in proton ceramics, the effects of energy band design and alignment mechanism on the elimination of electronic conduction need further studies.
The state-of-the-art BZCY/BZCYYb electrolytes with satisfactory performances (both proton conductivity and chemical stability) have been well developed for PCFC devices. Ni-based cermets are the dominant anodes for PCFCs, even though advanced manufacturing techniques on anodes are expected to develop efficient alternative anodes. In contrast, triple-conducting cathodes are of diversity, with two major directions of single-phase and composite materials. Doping, hydration, lithiation, and rational design of cation deficiency play important role in tuning the performance of single-phase triple-conducting cathodes. More research efforts are recommended to be made on particle-embedding, in situ exsolution, nano-infiltration, and the facile self-assembly for the fabrication of nanocomposite cathodes for PCFCs. Furthermore, more attentions are expected to the evaluation of proton transport mechanisms in semiconductor electrolytes, where the redox and proton transport occur at oxide interface or surface instead of the bulk. Finally, challenges regarding scale-up, mass production, long-term stability, and cost-effectiveness need to be solved before the integrated PCFC devices are ready for a commercial assembly line.

ACKNOWLEDGMENTS
This work was partially supported by National Science Foundation (CMMI-1661699).