Rare Earth Tungstate: One Competitive Proton Conducting Material Used for Hydrogen Separation: A Review

Abstract

Membrane technology is an advanced hydrogen separation method that is of great significance in achieving hydrogen economy. Rare earth tungstate membranes have both high hydrogen permeability and remarkable mechanical/chemical stability, exhibiting good application prospects in hydrogen separation. This review provides the basic aspects and research progress on rare earth tungstate hydrogen separation membranes. The crystal structure, proton transport properties, and membrane stability under a chemical atmosphere are introduced. Different membrane construction designs, such as single-phase, dual-phase, and asymmetric rare earth tungstate membranes, are summarized. Lastly, the existing problems and development suggestions for tungstate membranes are discussed.

1. Introduction

As the primary energy source of the earth, fossil fuels give rise to more and more environmental problems. The combustion products of fossil fuels, such as sulfur dioxide, hydrogen sulfide, and nitric oxide, are the main sources of atmospheric pollution. Carbon dioxide is the ringleader of global warming. Developing clean and sustainable energy alternatives to fossil fuels has become a consensus worldwide. Due to its high energy density and zero-emission, hydrogen energy is considered an essential and ideal energy source for achieving a clean and sustainable society [1,2,3]. The hydrogen market increased rapidly in recent years [4]. Hydrogen is mainly produced by the methane steam reforming reaction or by electrolyzing water. Complex and expensive separation processes such as cryogenic separation are inevitable for hydrogen production [5,6]. The cost of hydrogen production can be reduced largely by improving the hydrogen separation efficiency. Attribute to the significant advantages of energy efficiency, cost-effectiveness, and environmental compatibility, membrane technology is considered one of the most promising hydrogen separation technologies [7,8,9].

According to different membrane materials, hydrogen separation membranes can be divided into polymeric membranes [10], porous inorganic membranes [11,12], dense metallic membranes [13,14], and dense mixed protonic-electronic conducting (MPEC) membranes [15]. Among these membranes, the dense metallic membrane and MPEC membrane feature absolute selectivity, allowing only hydrogen to pass through the membrane. The transport mechanism of hydrogen through the metallic membrane is solution-diffusion. Hydrogen splits into two atoms on the membrane surface. The atoms are then transported to the other side of the membrane and recombined into hydrogen [16,17]. For the metallic membrane represented mainly by the Pd membrane, the high cost and high sensitivity to carbon monoxide, chlorine, and sulfur limit their large-scale application [18,19,20].

Hydrogen separation by MPEC membrane is realized through the mixed conduction of protons and electrons. The most studied MPEC membrane is the perovskite membrane, which mainly refers to doped ACeO₃ and AZrO₃, A = Ba, Sr [21,22,23,24]. The perovskite material has a cubic structure with a formula of ABO₃, eight BO₆ octahedrons occupy the corners of the cubic structure, and one A-site cation is in the center [25,26]. Perovskite membranes show high protonic conductivity in the temperature range of 500 to 900 °C [27]. The key challenges of perovskite membranes are poor membrane stability and low mixed proton and electron conductivity [28]. Despite the great efforts made to improve performance, perovskite membranes still cannot meet the needs of practical applications. ACeO₃-based perovskite materials exhibit high mixed protonic-electronic conductivity but are unstable in CO₂ or H₂S-containing atmospheres and degrade quickly due to the formation of barium carbonate and cerium oxide [29,30]. AZrO₃-based perovskite materials have good stability against acid gases, but the total conductivity is low because of the high grain boundary resistance [31,32].

Recently, another MPEC material, rare earth tungstate, has attracted significant interest as a promising hydrogen separation material. Rare earth tungstate has considerable ambipolar protonic-electronic conductivity at high temperatures. The ambipolar conductivity at 1000 °C is about twice that of perovskite oxide [33]. Compared with the weak protonic conductivity of perovskite oxide at intermediate temperatures, rare earth tungstate has high protonic conductivity below 750 °C [33]. More interestingly, rare earth tungstate also has good thermal/chemical stability at high temperatures in CO₂, H₂S-containing atmospheres, which is critical for hydrogen separation applications [34]. Therefore, rare earth tungstate has bright application prospects in hydrogen purification. So far, the crystal structure, proton transport mechanism, proton-electron conductivity, and application in hydrogen separation of tungstate materials have been studied in detail.

This paper gives an overview of rare earth tungstate MPEC membranes. The crystal structure, proton transport mechanism, and proton conductivity characteristic of rare earth tungstate are introduced. The doping strategies, membrane configuration design, as well as membrane stability of rare earth tungstate membranes are summarized. Lastly, the challenges and future development directions of tungstate membranes are discussed.

2. The Chemical Formula, Crystal Structure, and Transport Mechanism

Identifying the composition of rare earth tungstate oxide has undergone a long and complicated process. The chemical formula of this material was first recorded as Ln₆WO₁₂ (Ln = La, Nd, Sr, In), derived from ancient nomenclature [35]. However, researchers later found that La₆WO₁₂ could not remain a single phase unless the La/W ratio was within a specific range or the material was calcined at a particular temperature. Yoshimura et al. found that La₆WO₁₂ could not maintain thermodynamic stability below 1740 °C, and an additional La₁₀W₂O₂₁ phase would be formed [36]. To study the relationship between the La/W ratio and the formation region of LWO single-phase, Serra et al. investigated the composition of LWO powders with La/W ratios of 4.8–6.0 by XRD. The single-phase region of LWO is controlled by the La/W ratio and sintering temperature. The higher the content of lanthanum, the higher the sintering temperature required to form a single phase. At 1300 and 1500 °C, the single-phase region lies between 5.2 and 5.3, 5.3 and 5.5, respectively. The study also found that the La or W-site doping of the powders can lead to the shift of the single-phase region [37]. Through analyzing the lattice parameters of the material, Magrasó et al. also summarized the relationship between the material composition, sintering temperature, and La/W ratio [38]. The results are shown in Figure 1 and Figure 2. La₆WO₁₂ can remain a single phase at 1500 °C when 5.7 ≥ La/W ≥ 5.3. With the increase of La content, the lattice parameters of the oxide increase (region II in Figure 1). When La/W ≥ 5.7 or ≤5.3, the lattice parameter of the oxide is virtually independent of the La/W ratio (region III and I in Figure 1). The composition becomes complex, and an additional La₂O₃ or La₆W₂O₁₅ phase will be formed, respectively. As shown in Figure 2, the La/W ratio that keeps La₆WO₁₂ as a single phase increases with the increase of sintering temperature, especially at high sintering temperatures.

The crystal structure of rare earth tungstate was analyzed by many techniques, such as neutron powder diffraction and high-resolution X-ray synchrotron [39,40]. These oxides have a defective cubic fluorite structure in which the cubic cells have small distortions. Lanthanum is the largest of lanthanide cations. The crystal symmetry of rare earth tungstate decreases with the decrease of ion size. While the crystal structure of La to Pr is cubic or pseudo-cubic, Nd to Gd is pseudo-tetragonal, and Tb to Lu is rhombohedra. The crystal symmetry of rare earth tungstate with different lanthanide cations is shown in

Table 1. The crystal symmetry of rare earth tungstate with different lanthanide cations. Reprinted/adapted with permission from Ref. [41]. 2009, Sonia Escolástico, Vicente B. Vert, and José M. Serra.

			ionic radii Shannon coordination
element	atomic weight	valence	7	8	reported structure symmetry
La	138.9	3	1.100	1.160	cubic
Ce	140.12	3	1.070	1.143	cubic
		4	0.920 a	0.970
Pr	140.907	3	1.028 a	1.126	cubic
		4	0.905 a	0.960
Nd	144.24	3	1.046 a	1.109	tetragonal
Sm	150.35	2	1.220	1.270	tetragonal/cubic
		3	1.020	1.079
Eu	151.96	2	1.200	1.250	tetragonal/cubic
		3	1.010	1.066
Gd	157.25	3	1.000	1.053	tetragonal
Tb	158.924	3	0.980	1.040	rhombohedral
		4	0.820 a	0.880
Dy	162.50	3	0.970	1.027	rhombohedral
Y	88.905	3	0.960	1.019	rhombohedral
Ho	164.93	3	0.958 a	1.015	rhombohedral
Er	167.26	3	0.945	1.004	rhombohedral
Tm	168.934	3	0.937 a	0.994	rhombohedral
Yb	173.04	2	1.080	1.140	rhombohedral
		3	0.925	0.985
Lu	174.97	3	0.919 a	0.977	rhombohedral
Sc	44.956	3	0.808 a	0.870	rhombohedral

^a iInterpolated from tabulate data.

[41].

A structure model is proposed to describe the structure of rare earth tungstate with the space group F$\overline{4}$3m [38]. La occupies two different Wyckoff positions (Figure 3). One is 4a (0, 0, 0) (La1), which is fully occupied, and another is 24f (1/4, 1/4, 1/4) (La2), which is nearly fully occupied. Oxygen is in position 16e (x, x, x), x~0.13, and x~0.87, and the tungsten position is 4b (1/2, 1/2, 1/2). According to the model, the stoichiometry of the material is La₂₈W₄O₅₄, and the La/W ratio is 7, which is inconsistent with the experimentally measured value (5.3–5.7). Density Functional Theory (DFT) predicts that there must be other tungsten sites in the structure. W dissolves in the La2 site of La₂₈W₄O₅₄ as a donor dopant and acts as the “additional” tungsten, and the formula of the material should be described as La_28−xW_4+xO_54+δv_2−δ (δ = 3x/2), where x is the solubility of tungsten in La2 site, and v stands for the oxygen vacancy [39,42,43]. When x = 0, the formula is La₂₈W₄O₅₄v₂ (La/W = 7, LWO70), and the number of oxygen vacancies is 2. When x = 1, the formula is La₂₇W₅O_54+3/2v_1/2 (La/W = 5.4, LWO54), each formula unite accommodates one W in La2 sites, and the number of oxygen vacancies is 1/2. The existence of W at La2 sites was confirmed by many subsequent experiments [44,45,46]. LWO54 was proved to be a stable composition. LWO70 was unstable and could not be synthesized under normal conditions. For simplicity, the following La_28−xW_4+xO_54+δv_2−δ (5.7 ≥ La/W ≥ 5.3) is abbreviated as LWO (L = La, Nd, Gd, Er).

Rare earth tungstate has the same transport mechanism as perovskite oxides. Hydrogen permeation is achieved through the ambipolar diffusion of protons and electrons. Hydrogen splits into protons and electrons on the membrane surface. Protons and electrons diffuse simultaneously to the other side of the membrane and recombine to form hydrogen. When the LWO membrane is exposed to hydrogen or in a water vapor-containing atmosphere, the membrane will combine with protons through Equations (1) and (2). Protons exist in the form of hydroxyl in the membrane and are transported by the Grotthuss mechanism [48].

$${\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)+{\mathrm{v}}_{\mathrm{O}}^{··}+{\mathrm{O}}_{\mathrm{O}}^{\mathrm{x}}\leftrightarrow 2{\mathrm{OH}}_{\mathrm{O}}^{·}$$

(1)

$${\mathrm{O}}_{\mathrm{O}}^{\mathrm{x}}+\frac{1}{2}{\mathrm{H}}_2\leftrightarrow {\mathrm{OH}}_{\mathrm{O}}^{·}+{\mathrm{e}}^{\prime }$$

(2)

${\mathrm{v}}_{\mathrm{O}}^{··},{\mathrm{O}}_{\mathrm{O}}^{\mathrm{x}}$ are the oxygen vacancy and lattice oxygen, respectively, ${\mathrm{OH}}_{\mathrm{O}}^{·}$ is the hydroxyl ion [48].

3. The Conductivity Characteristics

The conductivity characteristics of LWO were first studied by Shimura et al. The material was found to have remarkable mixed protonic-electronic conductivity in wet hydrogen [49]. In 2007, Haugsrud et al. investigated the conduction properties of LWO specifically. The conductivity of LWO under reducing and deuterated conditions was measured by two-point ac conductivity measurements [50]. The total conductivity has a clear isotope effect. The conductivity under wet H₂ is higher than that under wet D₂ at temperatures below 1000 K (Figure 4). The conductivity of LWO is mainly controlled by protons under low temperatures (≤800 °C) and wet conditions [51,52]. The conductivity is predominated by electrons at higher temperatures and exhibits n-type electronic conduction and p-type electronic conduction under reducing and oxidizing conditions, respectively [53]. According to the conductivity characteristics as well as the hydrogen permeation results, LWO possesses significant hydrogen permeability above the temperature of 800 °C. However, at temperatures below 800 °C, the hydrogen permeability decreases due to the low n-type electron conductivity.

4. Current Status of Hydrogen Permeation Membrane

4.1. Single-Phase Membrane

As a mixed protonic-electronic conductor, LWO has high proton conductivity below the temperature of 900 °C but relatively weak electron conductivity. To obtain high hydrogen permeability, especially at intermediate temperatures, the electron conductivity of the material needs to be improved. One effective method to enhance the conductivity of LWO is the partial substitution of L or W cations, i.e., A-site doping or B-site doping. Since the protonic conductivity of LWO decreases in the order of LaWO, NdWO, GdWO, and ErWO, the conductivity and hydrogen permeation studies of LWO are focused on LaWO- and NdWO-based membranes.

Table 2. The total conductivity (S·cm⁻¹) of LWO at 800 °C under H₂ + 2.5% H₂O atmosphere, taken from Ref. [50].

Sample	LaWO	NdWO	GdWO	ErWO
Undoped	4 × 10−2	1 × 10−2	8 × 10−3	9 × 10−3
Ca-doped	1 × 10−2	9 × 10−3	-	1 × 10−3

shows the total conductivity of LWO under an H₂ + 2.5% H₂O atmosphere.

4.1.1. LaWO-Based Membrane

Lanthanum tungstate is the most studied rare earth tungstate because of its highest protonic conductivity. The protonic conductivity of undoped LaWO reaches 5 × 10⁻³ S·cm⁻¹ [54]. To improve the ambipolar conductivity of LaWO, A-site doping is initially considered. Zr⁴⁺, Ca²⁺, Y³⁺, Nd³⁺, K⁺, Ce⁴⁺, and Tb³⁺ are trying to partially replace La³⁺. However, the results are disappointing, as A-site doping does not make the electronic conductivity increase but makes the conductivity of most of the doped materials decrease. Lashtabeg et al. studied the conductivity property of (La, Y)_10+xW_2−xO_21−δ. The electronic conductivity decreased with the increase of Y content at high temperatures [55]. Shimura et al. investigated the electronic conductivity of Zr- and Nd-doped La_xWO_3+1.5x. The doping of Nd had little impact on the electronic conductivity of the material, while Zr doping reduced the conductivity [49]. For K-doped (La_1−xK_x)_27.08W_4.92O_55.38−θ,the conductivity initially increased with slight K doping but decreased with increasing K content [56]. Conductivity changes due to A-doping can be ascribed to two factors. First, compared with La, the smaller volume of the doped ion leads to cell shrinkage and a more ordered cation sublattice arrangement. Second, the substitution of La reduces the La/W ratio of the material, leading to the increase of $W_{La}^{\cdots }$ and a decrease in oxygen vacancy.

Compared with A-site doping, B-site doping can improve the electronic conductivity of LaWO more effectively. Nb, Re, Mo, Zr, and Mn are the main B-site ions that partially substitute W. Zayas-Rey studied the conducting features of Nb-doped LaWO. The total conductivity of La₂₇NbW₄O₅₅ at 800 °C was 0.01 S·cm⁻¹, which was higher than that of the undoped LaWO56 (7 × 10⁻³ S·cm⁻¹) and undoped LaWO54 (4 × 10⁻³ S·cm⁻¹) [57]. Escolastico studied the conductivity and hydrogen permeability of the Re-doped LaWO55 membrane at low temperatures. The total conductivity of LaWO55 at temperatures below 700 °C was improved by Re-doping. At 700 °C, the hydrogen flux of the doped membrane reached 0.095 mL·min⁻¹·cm⁻² when swept with humidified gas [58].

Mo-doped LaWO is proven to be a truly competitive hydrogen permeation material. It is considered to be the next-generation ceramic membrane material with high hydrogen permeability. First, Mo-doping can effectively improve the electronic conductivity of hydrogen permeation membranes or oxygen permeation membranes [59,60]. Mo doping is expected to be the most effective doping strategy to improve the electronic conductivity of LaWO. The conductivity of Mo-doped LaWO is one order of magnitude higher than that of an undoped one [61]. Second, Mo has a similar ionic radius to W. The replacement of W with Mo has little effect on the material’s structure. Third, Mo has different oxidation states, and the conversion between different oxidation states can generate additional oxygen vacancies inside the membrane. Furthermore, Mo doping exhibits higher sintering activity than the Mo-free compound. Figure 5 shows the conductivity of La_28−y(W_1−xMo_x)_4+yO_54+δ. The ambipolar conductivity of the material has a strong enhancement due to the Mo-doping, and the ambipolar conductivity increases with the increase of Mo-content [62]. At 1000 °C, the ambipolar conductivity of 20% Mo-doped LaWO is twice that of LaWO. The ambipolar conductivity of 40% Mo-doped LaWO at 600 °C is almost equivalent to that of the undoped material at 1000 °C. Mo-doping can significantly improve the electronic conductivity of LaWO and has no damage to the protonic conductivity of the material. Therefore, Mo-doped LaWO is considered to possess excellent hydrogen permeability, especially at intermediate temperatures. Escolastico et al. studied the hydrogen permeation performance of the La_5.5W_0.8Mo_0.2O_11.25−δ membrane. At 700 °C, the hydrogen flux was 0.04 mL·min⁻¹·cm⁻² when the feed and sweep gases were humidified. The flux was much higher than that of the undoped membrane and nearly ten times higher than that of the SrCe_0.7Zr_0.2Eu_0.1O_3−δ membrane at the same conditions [58]. Vøllestad et al. studied the hydrogen permeability of 30% Mo-doped lanthanum tungsten membrane, La₂₇W_1.5Mo_3.5O_55.5. The n-type conductivity of the membrane was significantly increased by Mo-doping, and the protonic conductivity became the rate-limiting of hydrogen separation. The hydrogen flux reached 6 × 10⁻⁴ mL·min⁻¹·cm⁻¹ at 700 °C. The hydrogen permeability of the membrane was believed to be significantly higher than that of SrCeO₃-based perovskite membranes [63]. Chen et al. fabricated a Mo and Nb co-doped La_5.5WO_11.25−δ membrane and studied its hydrogen permeation performance. A flux of 0.195 mL·min⁻¹·cm⁻² was obtained at 1000 °C, and the flux had no decline during the 80 h operation [64]. The excellent hydrogen permeation character of Mo-doped LaWO makes it a broader application prospect. Xue et al. constructed a catalytic reactor based on La_5.5W_0.6Mo_0.4O_11.25−δ membrane for nonoxidative methane dehydroaromatization. Compared with the fixed-bed reactor, CH₄ conversion was improved, and the aromatics yield was increased by ~50–70% due to the in situ removal of hydrogen [65].

4.1.2. NdWO-Based Membrane

NdWO is another important MPEC tungstate oxide besides LaWO. Since the ionic radius of Nd is smaller than that of La, the crystal structure of NdWO changes from cubic to rhombohedral. The structure change may lead to changes in conductivity and hydrogen permeability. Due to the higher proton mobile enthalpy caused by the reduction of crystal symmetry, NdWO was predicted to have lower hydrogen permeability than LaWO [66]. NdWO-based membranes are mainly studied by the Serra group from Spain. U, Re, and Mo are commonly used elements to substitute W. Doping Re in NdWO can significantly enhance the oxygen ionic conductivity and n-type electronic conductivity of the compound. Mo-doping can also improve the n-type electronic conductivity of the material. Meanwhile, it can improve the protonic conductivity of the material below 700 °C and the oxygen ion conductivity above 700 °C. The introduction of U can improve both the electronic conductivity and ionic conductivity of the material when the dopant acceptor level is low. The protonic conductivity drops and the total conductivity stagnates with increasing U concentration [67]. Escolástico et al. investigated the hydrogen permeability of the Nd_5.5W_0.5Mo_0.5O_11.25−δ membrane. At 700 °C, the hydrogen flux was 3.8 × 10⁻³ mL·min⁻¹·cm⁻¹ [68]. The permeability was higher than that of La_5.5W_0.8Mo_0.2O_11.25−δ (2.7 × 10⁻³) and SrCe_0.95Tm_0.05O_3−δ (2.6 × 10⁻³), but lower than that of La_5.5W_0.8Re_0.2O_11.25−δ (5.9 × 10⁻³) and BaCe_0.80Y_0.10Ru_0.10O_3−δ (4.3 × 10⁻³, 800 °C). At 1000 °C, the hydrogen flux of the membrane reached 0.3 mL·min⁻¹·cm⁻² [69]. Wang et al. fabricated Nd_5.5W_0.5Mo_0.5O_11.25−δ hollow fiber membrane via phase-inversion technique. At 975 °C, the hydrogen permeation flux of a 170 μm-thick membrane reached 1.29 mL·min⁻¹·cm⁻² [70]. The experimental results show that doped NdWO membranes possess excellent hydrogen permeability, no less than any other state-of-the-art tungstates or cerates membranes. The NdWO-based membrane is among the best H₂-permeating MPEC membranes. The hydrogen permeation performances of several representative single-phase LWO membranes are listed in

Table 3. Hydrogen permeation performances of several representative single-phase LWO membranes.

Composition	Hydrogen Flux (mL·min−1·cm−2)	σ × 103 (S·cm−1)	T (°C)	Thickness (μm)	Feed/Sweep Gas	Ref.
Nd5.5WO11.25−δ	0.03		1000	900	50%H2-He/Wet Ar	[69]
La5.5WO11.25−δ	0.136		1000	900	50%H2-He/Wet Ar	[71]
La5.5WO11.25−δ	0.037		1000	500	50%H2-He/Ar	[72]
A-Site Doping
La0.95Ca0.05W1/6O~2		1.0	800		Wet H2	[33]
YLa9W2O21		4.56	1000		Air	[55]
(La0.98K0.02)27.08W4.92O55.38−θ		3.59	800		Wet air	[56]
(La5/6Nd1/6)5.5WO12−δ	0.046		1000	900	Wet2.5%H2-Wet Ar	[73]
B-Site Doping
La27NbW4O55		10	800		Wet N2	[57]
La5.5W0.8Re0.2O11.25−δ	0.095		700	760	50%H2-He/Wet Ar	[58]
La5.5W0.8Mo0.2O11.25−δ	0.04		700	850	Wet50%H2-He/Wet Ar	[58]
La27.07Mo1.97W2.95O54+δ		1.6	600		Wet H2	[62]
La27Mo1.5W3.5O55.5	6 × 10−4 mL·min−1·cm−1		700	-	50%H2-Ar/Ar	[63]
La5.5W0.45Nb0.15Mo0.4O11.25−δ	0.195		1000	500	50%H2-He/Ar	[64]
La5.5W0.45Nb0.15Mo0.4O11.25−δ	0.233		1000	500	50%H2-He/Wet Ar	[64]
La5.5W0.8Mn0.2O11.25−δ	0.07		1000	500	50%H2-He/Ar	[74]
La5.5W0.6Mn0.2 Mo0.2O11.25−δ	0.12		1000	500	50%H2-He/Ar	[74]
La5.5W0.8Cr0.2O11.25−δ	0.046		1000	500	50%H2-He/Ar	[75]
La5.5W0.45Nb0.15Mo0.4O11.25−δ (Pt coated)	0.483		1000	500	Wet50%H2-He/Wet Ar	[76]
La5.4W0.55Nb0.15Mo0.3O11.25−δ (Pt coated)	0.01 mL·min−1·cm−1		1000	-	Wet50%H2-He/Wet Ar	[77]
La5.5W0.6Mo0.4O11.25−δ−x/2Cl0.1	0.15		1000	500	50%H2-He/Ar	[78]
Nd5.5W0.9U0.1O11.25−δ (Pt coated)	0.015		740	530	Wet50%H2-He/Wet Ar	[67]
Nd5.5W0.5Mo0.5O11.25−δ (Pt coated)	0.07		700	550	Wet50%H2-He/Wet Ar	[68]
Nd5.5W0.5Mo0.5O11.25−δ	0.3		1000	900	Wet50%H2-He/Wet Ar	[69]
Nd5.5W0.5Mo0.5O11.25−δ	1.29		975	170	80%H2-He/Wet Ar	[70]
Nd5.5W0.5Mo0.5O11.25−δ	0.05		1000	500	50%H2-He/Ar	[79]
La5.5W0.9Ti0.1O11.25−δ		9.2	800	-	Wet N2	[80]
La5.5W0.95Al0.05O11.25−δ		9.8	800	-	Wet N2	[80]
Nd5.5W0.5Re0.5O11.25−δ	0.08		1000	900	Wet50%H2-He/Wet Ar	[81]

.

4.2. Dual-Phase Membrane

Another effective method to improve the ambipolar conductivity of the tungstate membrane is to construct a dual-phase membrane. The dual-phase membrane is fabricated by LWO and is an electronic conducting phase with high electronic conductivity. Due to the addition of an electronic conducting phase, the ambipolar conductivity of the membrane can be improved greatly. Dual-phase membranes can be divided into two types depending on the electronic conducting phase. One is a ceramic-metal (cermet) dual-phase membrane in which one metal or alloy metal is the electronic conducting phase, and the other is the ceramic-ceramic (cercer) dual-phase membrane in which one oxide with high electronic conductivity is the electron conductor. Xie et al. fabricated a cermet dual-phase membrane, Ni-La_5.5WO_11.25−δ. The membrane had high ambipolar conductivity, and the hydrogen permeation flux at 1000 °C reached 0.18 mL·min⁻¹·cm⁻², which was one order of magnitude higher than that of the La_5.5WO_11.25−δ single-phase membrane [82]. Bespalko et al. prepared a Cu_0.5Ni_0.5-La_5.5WO_11.25−δ cermet asymmetric membrane via hot pressing techniques. When the membrane was used as a membrane reactor for ethanol steam reforming, the H₂ flux through the membrane at 900 °C reached 2.0 N mL·min⁻¹·cm⁻² [83].

The main issue that needs to be solved for cermet membranes is the poor thermal compatibility between metal and ceramic phases, which can lead to the fracture of the membrane at high temperatures. In addition, the oxidation of metal at high temperatures makes the preparation of cermet membrane complicated. To overcome the disadvantages of cermet membranes, a cercer dual-phase membrane using ceramic oxides instead of metal as the electron conductor is developed. The ceramic oxide electronic conducting phase used in the dual-phase membrane should meet these criteria: (1) has high electronic conductivity, especially at intermediate temperatures; (2) has good thermal compatibility with LWO at high temperatures and reducing atmospheres; (3) no inter-phase reaction with LWO during the membrane preparation and hydrogen permeation processes. Escolástico et al. prepared a cercer dual-phase membrane using La_5.5WO_11.25−δ (LWO) as the protonic conducting phase and La_0.87Sr_0.13CrO_3−δ (LSC) as the electronic conducting phase. The mixing of LWO and LSC achieved high membrane densities and well-balanced ambipolar conductivity. The total conductivity of the composite was higher than that of a single LWO or LSC. The effect of phase composition ratio on hydrogen permeation was investigated. 50LWO-50LSC presented the highest hydrogen permeability, while 20LWO-80LSC exhibited the highest total conductivity. The hydrogen flux of a 370 μm-thick 50LWO-50LSC membrane at 700 °C was 0.15 mL·min⁻¹·cm⁻², which was among the highest fluxes achieved by MPEC membranes [84]. Liang et al. prepared a triple-conducting composite membrane La_5.5WO_11.25−δ (LWO)-La_0.8Sr_0.2FeO_3−δ (LSF) for pure hydrogen production, LWO was used for proton transport, and LSF was used for electron and oxygen ion transport. After sweeping with humidified Ar, the hydrogen flux of the membrane at 900 °C was 0.14 mL·min⁻¹·cm⁻². When swept with diluted steam (87.5% H₂O + Ar), the hydrogen flux changed to 0.25 mL·min⁻¹·cm⁻² [85]. The hydrogen permeation results of La₂₇W_3.5Mo_1.5O_55.5−δ (LWM)-La_0.87Sr_0.13CrO_3−δ (LSC) ceramic composites with phase ratios of 7:3, 6:4, and 5:5 showed that, 70LWM-30LSC membrane in which a continuous LSC phase did not form exhibited the highest hydrogen flux [86]. It can be concluded that a continuous electronic conducting phase network may not be necessary for dual-phase membranes. The high hydrogen flux is due to the enhanced membrane ambipolar conductivity through adding the extra electronic conducting phase. The hydrogen permeation performances of several representative dual-phase LWO membranes are listed in

Table 4. Hydrogen permeation performances of several representative dual-phase LWO membranes.

Composition	Hydrogen Flux (mL·min−1·cm−2)	T (°C)	Thickness (μm)	Feed/Sweep Gas	Note	Ref.
60Ni-40La5.5WO11.25−δ	0.18	1000	500	50%H2-He/Ar		[82]
Ni0.5Cu0.5-40Nd5.5WO11.25−δ	2	900	150	H2O-EtOH-Ar/Ar	catalyst coated; reactor	[83]
Ni0.5Cu0.5-Nd5.5WO11.25−δ	0.27	900	50	H2O-EtOH-Ar/Ar	catalyst coated; reactor	[87,88]
50La5.5WO11.25−δ-50La0.87Sr0.13CrO3−δ	0.15	700	370	50%H2-He/Wet Ar		[84]
50La5.5WO11.25−δ-50La0.8Sr0.2FeO3−δ	0.14	900	500	50%H2-N2/Wet Ar		[85]
50La5.5WO11.25−δ-50La0.8Sr0.2CrO3−δ	0.05	900	500	50%H2-N2/Wet Ar		[85]
70La27W3.5Mo1.5O55.5−δ-30La0.87Sr0.13CrO3−δ	0.001 mL·min−1·cm−1	700	1430	Wet50%H2-He/ Wet Ar	Pt coated	[86]
50La27W3.5Mo1.5O11.25−δ-50La0.87Sr0.13CrO3−δ	0.004 mL·min−1·cm−1	900	-	Wet49%H2-He/ Wet Ar		[89]
60La5.5WO11.25−δ-40La0.87Sr0.13CrO3−δ	0.22	725	360	Wet49%H2-He/ Wet Ar	Pt coated	[90]
33La0.87Sr0.13CrO3−δ-67La5.4WO12−δ	0.10	750	40–70	Wet50H2-Ar/ Wet Ar		[91]

.

4.3. Asymmetric Membrane

Besides composition optimization, engineering approaches can also be used to enhance the hydrogen permeability of the LWO membrane. Membrane thickness is an important factor affecting the permeability of the hydrogen separation membrane. A thicker membrane means greater bulk diffusion resistance. Membrane thickness can be reduced by constructing asymmetric membranes. An asymmetric membrane comprises a thin dense layer and a porous support layer. The porous substrate provides sufficient mechanical strength for the membrane and allows gases to be transported through its porous channels. To maintain the mechanical stability of the membrane, the porous substrate and the dense layer should have similar thermal expansion coefficients and cannot react with each other at high temperatures. The main preparation techniques of asymmetric membranes include dry pressing, spin coating, tape casting, and phase inversion [92]. Using carbon black as the pore former, Weirich et al. prepared LWO56 self-supported asymmetric membrane through tape casting. A single LWO56 dense layer was placed on top of five layers of LWO56 tape which contained optimized carbon black content. While the total membrane thickness was 300 μm, the dense layer thickness was only 40 μm [93]. The asymmetric structure of the membrane reduces the actual membrane thickness and bulk diffusion resistance greatly. The commonly used pore-forming agents in the porous support are rice starch and carbon black. By comparing the permeability coefficient of the porous substrates, the permeability of carbon black is higher than that of rice starch with identical addition [93]. Gil et al. prepared LWO56 asymmetric membrane via dip coating, the porosity of the support layer with different amounts of carbon black was 25 vol. % and 40 vol. %, respectively. The hydrogen flux of the membrane at 1000 °C was 0.14 mL·min⁻¹·cm⁻² [94].

Without using pore former, Xie et al. fabricated a LaWO asymmetric membrane through the phase inversion technique. The membrane had a unique composite pore microstructure. The support layer of the membrane contains a large number of finger-like pores and small irregular pores, which were formed during the phase inversion process [95]. Compared with traditional asymmetric membranes, the two layers of the hollow fiber membrane have the same composition and are prepared by a single process, which provides the membrane with better thermomechanical compatibility. A hollow fiber membrane simplifies the membrane fabrication technique and is more conducive to the large-scale production of asymmetric membranes. The hydrogen permeation performances of several representative asymmetric LWO membranes are listed in

Table 5. Hydrogen permeation performances of several representative asymmetric LWO membranes.

Composition	Hydrogen Flux (mL·min−1·cm−2)	T (°C)	Dense Layer Thickness (μm)	Preparation Method/ Pore Former	Feed/Sweep Gas	Ref.
La6-xWO12−δ	-		40	tape casting/carbon black		[93]
La5.6WO11.25−δ	0.14	1000	25	dip coating/carbon black	10%Wet H2-He/Ar	[94]
La5.5W0.6Mo0.4O11.25−δ	0.273	1000	300	phase inversion	50%H2-He/Wet Ar	[95]
La5.6WO11.25−δ	0.14	1000	20	dip coating/carbon black	10%Wet H2-He/Ar	[96]
La5.4WO12−δ	-		20–30	tape casting/rice starch		[97]
La5.5W0.6Mo0.4O11.25−δF0.05	0.16	975	67	dry pressing/soluble starch	50%H2-He/Ar	[98]
La28−xW4+xO54+δ	0.4	825	60	tape casting/rice starch	50%H2-He/Wet Ar	[99]
La28−yW4+yO54+δ	-		20	dip coating/carbon black		[100]
Nd5.5Mo0.5W0.5O11.25−δ/Nd5.5Mo0.5W0.5O11.25−δ-Ni	0.26	900	26	phase inversion	50%H2-He/N2	[101]

.

5. The Membrane Stability

In addition to hydrogen permeability, membrane stability is also the focus of attention for hydrogen separation membranes. Considering the application environment in the industry, hydrogen-permeable membranes should have (1) sufficient chemical stability under reducing conditions; (2) sufficient chemical stability at CO₂ and other acidic gas-containing atmospheres; (3) sufficient thermal stability at temperatures higher than 800 °C. As excellent hydrogen permeation substances, perovskite oxides such as BaCeO₃ and SrCeO₃ possess good ambipolar conductivity and high hydrogen permeability. However, the chemical stability of cerate membranes has always been a critical issue. These cerate oxides react rapidly in a CO₂ or H₂S-containing atmosphere. The formed carbonates or sulfides can seriously damage the conductivity and permeability of the membrane.

The stability of the LWO membrane has been intensively studied, and it is believed that the stability of the LWO membrane is better than that of the perovskite membrane [102]. Seeger et al. inspected the stability of Mo- and Re-doped LWO membranes at reducing conditions. The membrane was treated in wet hydrogen for 12 h at 800 °C. Comparing the XRD and Raman spectroscopy patterns of the samples before and after reduction, except for the slight shifts of the structural peak caused by the reduction of Re and Mo, no other phases were formed during the reduction treatment. The SEM images also presented no change after the treatment [37]. The stability against CO₂ of LWO membranes was evaluated in much literature. Escolastico et al. investigated the stability of some Mo-, Re-doped LaWO and NdWO membranes in CO₂-rich environments. TG measurements were made by heating up to 1000 °C in Ar with 5% CO₂. Except for a little mass loss due to the sample dehydration and oxygen release, no carbonates were formed (mass gain) in these compounds observed from the TG curve [58,69]. After treating the membranes at 400 °C and 34.5 atm under a mixed gas atmosphere consisting of COS, HCN, CO₂, CO, H₂, and H₂O, the XRD results showed that the membrane had no degradation, and the formed carbonate, sulfide, and sulfate phases could be negligible. The stability of the Nd_5.5W_0.5Mo_0.5O_11.25−δ membrane exposed to 1500 ppm H₂S at 827 °C was evaluated. No Nd₂O₂S or other phases were formed during the treatment. In contrast, BaCeO₃ decomposed under the same condition, and Ce₂O₂S and BaS were detected [69].

Chen et al. studied the hydrogen permeable stability of Nb, Mo co-doped La_5.5WO_11.2−δ membrane under CO₂-containing atmospheres. When the feed gas was changed from 50% H₂-He to 25% CO₂-25% H₂-He at 827 °C, the hydrogen flux of the membrane initially decreased from 0.155 to 0.064 mL·min⁻¹·cm⁻² and then kept steady as time elapsed. The decrease in the hydrogen flux was attributed to two reasons: first, the water gas shift reaction occurred at the feed side and consumed some hydrogen, thus leading to a decrease of hydrogen permeation driving force. Secondly, CO₂ absorption on the membrane surface hindered the surface exchange of hydrogen and weakened the hydrogen permeation. When the feed gas recovered, the hydrogen flux recovered immediately. At 950 °C, the hydrogen flux of the membrane exhibited almost no change during the 80 h permeation test. The membrane microstructure remained intact, and no carbonate was detected after the stability test [64].

6. Challenges and Outlook

6.1. Challenges

Despite the great efforts made to improve performance, LWO membranes still cannot meet the needs of industrial applications. The hydrogen flux needs to be improved by at least one order of magnitude. There are many different preparation techniques for the raw powder or the membrane. The use of these techniques is now chaotic and random. Even using the same composition, the hydrogen permeability of the membranes prepared by different techniques may be different. This makes it difficult to evaluate the true hydrogen permeation performance of materials. Under a CO₂-containing atmosphere, the hydrogen permeability of LWO membranes decreases due to the surface adsorption of CO₂. Reducing the surface occupation of CO₂ under a CO₂-containing atmosphere is a crucial issue to be solved for the LWO membrane, which is important for the practical application of the membrane.

6.2. Future Insights

First, the permeability should be further enhanced through composition optimization. Compared with other hydrogen separation membranes, such as metal membranes or perovskite membranes, the hydrogen separation research on LWO membranes is still not thorough enough. The doping strategy is efficient in enhancing the permeability of ceramic membranes. At present, only a few metal cations like Mo, Re, and Nb have been used as dopants for LWO. More doping ions, cations, or anions should be tried. The relationship between the doped ion structure, the doping amount, and the membrane performance should be summarized. Compared with the perovskite membrane, the LWO membrane exhibits good protonic conductivity at intermediate temperatures (<800 °C). Since many hydrogen production reactions occur at intermediate temperatures, future research on LWO membranes should focus on improving the hydrogen permeability at intermediate temperatures to facilitate the coupling with reactions.

Second, permeability through engineering perspectives should be enhanced. Membrane configuration is an important factor affecting membrane performance. Since reducing the membrane thickness can effectively enhance hydrogen permeability, it is of great significance to construct ultra-thin membranes with an asymmetric or hollow fiber structure. The dense layer thickness of the hollow fiber membrane can be reduced to 10~20 μm, which can reduce the permeation resistance of the membrane significantly. To promote the diffusion of transport species inside the membrane, Meng et al. prepared a laminated composite membrane by using the tape-casting technique. Due to the formation of regular and independent transport channels in the membrane, the bulk-diffusion distance of protons and electrons is reduced greatly [103].

Third, new techniques for membrane fabrication should be developed. The preparation process of membranes, especially composite membranes, is complicated and laborious. New preparation techniques should be developed. Mathematical modeling can assist in designing materials with excellent performance and feasible technology. Simulation calculation instead of the experiment can economize on manpower and material resources. Three-dimensional printing technology can help prepare membranes with special structures or properties, such as asymmetric membranes or membranes with catalytic functions, which can greatly expand the application field of membranes.

Fourth, new characterization techniques for the raw materials and the membranes should be developed. On the one hand, more advanced characterization techniques need to be developed to accurately evaluate the structure of LWO and the membranes, such as neutron and synchrotron scattering techniques. On the other hand, research on hydrogen permeation membranes is an expensive process, and more economical characterization techniques need to be developed to reduce the research costs.