Perspective—Oxygen Separation Technology Based on Liquid-Oxide Electrochemical Membranes

The electrochemical oxygen generator operation principle is illustrated in Fig. 1b. The generator consists of two main components: electrodes (anode and cathode) separated by an electrolyte. Under the electric potential, oxygen ions migrate in the ceramic electrolyte from cathode to anode. The rare-earth stabilized bismuth oxide and rare-earth doped ceria are usually used as the oxygen generator electrolytes.^27–34 Temperature dependences of conductivities of these ceramic electrolytes are presented in Fig. 2a. To achieve a sufficient ionic conductivity of these electrolytes, oxygen generators operate in the temperature range of 700 °C–800 °C. The purity of oxygen produced by the ceramic electrolyte-based generators significantly depends on the electrolyte density. At the present time, the product grades of 98%–99.99% oxygen are available.^32–34

Zoom In Zoom Out Reset image size

Intensive study of the MIEC membrane materials was started in the mid-80 s of the last century when Teraoka et al.³⁵ reported a high oxygen permeation flux of ceramic La_1−xSr_xCo_1−yFe_yO_3−δ perovskite. The MIEC membrane operation principle is illustrated in Fig. 1a. Different perovskite-type La_1−x(Sr, Ba, Ca)_xCo_1−y(Fe, Mn, Cu, Ni)_yO_3−δ and composite mixed-conducting ceramic membrane materials have been developed.^17,21 Carolan and Dier³⁶ (Air Products & Chemicals Inc.) patented an asymmetric membrane with a thin oxygen selective perovskite film on a porous support. A number of techniques such as tape casting, slip casting, dip coating, phase inversion, screen printing and spin coating have been developed to manufacture asymmetric membranes.²¹ At the present time, there are asymmetric membranes with the highest oxygen permeability.^20,22,37 The asymmetric membrane-based semi-industrial modules have been constructed and tested under various capacities.^10,11 The tests of these membrane modules have revealed a number of disadvantages. Primarily, there are incomplete hermeticity of thin ceramic membrane films and their mechanical instability under a temperature gradient.¹⁰ Currently, the product grades of 98.9%–99.9% oxygen are available.^10,11

Thus, the ITM technology based on ionic conducting and MIEC ceramic membranes does not allow the separation of ultrahigh purity oxygen (>99.999% purity).

The oxygen permeation flux (${J}_{{O}_{2}}$) through an oxygen generator electrolyte is expressed by Eq. 1

$$\begin{eqnarray}&&{J}_{{O}_{2}}=\displaystyle \frac{i}{4F}\end{eqnarray} \tag{ 1 }$$

where i is the current density which is defined as

$$\begin{eqnarray}&&i={\sigma }_{i}E\end{eqnarray} \tag{ 2 }$$

where ${\sigma }_{i}$ is the oxygen ionic conductivity and $E$ is the electric field strength ($E=U/L$ where $U$ is the voltage and $L$ is the electrolyte thickness). In accordance with Eqs. 1 and 2, the electrolyte should have maximum ionic conductivity, since the oxygen generator voltage power losses are due to the electrolyte and electrode resistance as well as electrode polarization. Of the known electrolytes, ceramic δ-Bi₂O₃ electrolyte shows the highest oxygen ionic conductivity of 2.3 s·cm⁻¹ at 780 °C (Fig. 2a).²⁷ However, it cannot be used as an oxygen generator electrolyte, since the polymorphic transformation of monoclinic α-Bi₂O₃ into cubic δ-Bi₂O₃ at 730 °C is accompanied by a large volume change (∼12 vol%) leading to the cracking of the ceramic δ-Bi₂O₃. This cubic phase can be stabilized by the partial replacement of Bi³⁺ with Y³⁺, Er³⁺, Dy³⁺ and W⁶⁺ in the crystal lattice. However, ionic conductivity of the stabilized Bi_0.75Y_0.25O_1.5, Bi_0.8Er_0.2O_1.5 and Bi_0.88Dy_0.08W_0.04O_1.56 is substantially lower than that of δ-Bi₂O₃ (Fig. 2a). In order to improve the purity of industrial oxygen, we suggest to use containing liquid grain boundaries (GBs) δ-Bi₂O₃-0.2 wt% B₂O₃ composite as an oxygen generator electrolyte.^25,38 The intergranular liquid layers predominantly consisting of molten δ-Bi₂O₃ provide ionic conductivity, gas tightness and ductility to the composite. This solid/liquid composite demonstrates the highest oxygen ionic conductivity (Fig. 2a), which is due to the high disordering of oxygen sublattice in fluorite structure of solid δ-Bi₂O₃³⁹ and short-range disorder of liquid Bi₂O₃.³⁸ The highest oxygen selectivity of the solid/liquid δ-Bi₂O₃-0.2 wt% B₂O₃ composite is due to the fact that the liquid phase (molten Bi₂O₃-1.2 wt% B₂O₃) with almost zero solubility of nitrogen is an excellent high-temperature sealing component. Moreover, the intergranular liquid layers provide ductility to the composite and thus overcomes the problem of brittleness (Fig. 2c). Note that the volume fraction of the liquid cannot exceed 30%–35%.²⁵ A further increase in the volume fraction of the liquid leads to the mechanical instability of the composite. GB wetting plays a critical role in the manufacturing of the solid/liquid composite (regarding the thermodynamics of GB wetting in ceramic oxide materials, the reader is referred to the comprehensive review⁴⁰).

To assess the potential of the solid/liquid δ-Bi₂O₃-0.2 wt% B₂O₃ composite as an oxygen generator electrolyte, a symmetric cell based on this composite has been manufactured and tested. The electrochemical reactions took place within porous Bi₂Ru₂O₇ + 1.2 wt% Pt electrodes deposited on the electrolyte. The electrodes were selected for their conductivity and catalytic activity.⁴¹ The oxygen generation rate as a function of applied current is presented in Fig. 2b. The purity of the generated oxygen has been more than 99.999%. The measured area specific resistance (ASR) of the cell was 0.55 Ω cm² at 750 °C. The cell was operating for 5 h with electrolyte Faradaic efficiency of 97%. These initial results show that the δ-Bi₂O₃-0.2 wt% B₂O₃ composite with liquid GBs can be used as an oxygen generator electrolyte for production of ultrahigh purity oxygen.

The gastight and mechanically stable (under a temperature gradient) solid/liquid Co₃O₄-36 wt% Bi₂O₃ composite MIEC membrane material with a record oxygen selectivity (O₂/N₂ > 10⁵)²³ and competitive oxygen permeability (Fig. 3f) has been developed. In this composite membrane material, the liquid phase predominantly consisting of molten Bi₂O₃ provides oxygen ionic conductivity, gas tightness and ductility, while the solid phase consisting of cobalt oxide ensures the electronic conductivity. The ionic conductivity of the composite is increased with the volume fraction of the liquid. However, due to mechanical requirements, the volume fraction of the liquid does not exceed 35%.²³ In the membrane thickness range of 1–3 mm, the oxygen permeation flux through the solid/liquid Co₃O₄-36 wt% Bi₂O₃ composite membrane material is limited by diffusion of oxygen ions in the liquid phase and described by Eq. 3⁴²

$$\begin{eqnarray}&&{j}_{{O}_{2}}=\displaystyle \frac{{D}_{i}{\varphi }_{l}{\rho }_{l}}{8MNL}\,\mathrm{ln}\,\displaystyle \frac{{P}_{{O}_{2}}^{{\prime} }}{{P}_{{O}_{2}}^{{\prime\prime} }}\end{eqnarray} \tag{ 3 }$$

where ${\varphi }_{l}$ is the volume fraction of liquid, ${D}_{i}$ is the diffusion coefficient of oxygen ions in liquid, $N$ = 1 and ${\rho }_{l}$ and $M$ are the density and molar mass of liquid phase, respectively. The simplicity and cheapness of the manufacture of solid/liquid Co₃O₄-36 wt% Bi₂O₃ composite membrane material combined with the highest oxygen selectivity and competitive oxygen permeability makes it most suitable for the efficient separation of ultrahigh purity oxygen from air.

Zoom In Zoom Out Reset image size

A promising candidate for ultrahigh purity oxygen separation from air is the copper and vanadium oxide-based bilayer MIEC-Redox solid/liquid composite membrane material with combined diffusion-bubbling oxygen mass transfer.²⁴ This original membrane material is spontaneously formed during thermal treatment of ceramic CuO-25 wt% Cu₅V₂O₁₀ composite in a "chemical field" (under an oxygen chemical potential gradient) above 816 °С. This temperature corresponds to the peritectic transformation of Cu₅V₂O₁₀ compound by reaction: Cu₅V₂O₁₀ → solid CuO + liquid (Cu₂O + V₂O₅ melt) + gas O₂.²⁴ The result is a bilayer MIEC-Redox composite membrane material consisting of solid, liquid, and gas phases (Fig. 3). The oxygen permeation flux through this membrane material is limited by diffusion of oxygen ions in the V₂O₅-based liquid phase in the external layer of the membrane (Fig. 1c) and is described by Eq. 3. According to the dynamic polymer chain model,⁴³ oxygen ion transfer in the V₂O₅-based liquid phase occurs by a stochastic disconnection-connection chain mechanism. In the internal layer of the MIEC-Redox membrane material, the redox reactions and nucleation, growth and transport of oxygen gas bubbles take place. The characteristic nucleation time, density and average radius of oxygen gas bubbles are 10⁻⁴ s, 10¹² m⁻³ and 1.4·10⁻⁵ m, respectively. The oxygen permeation flux through the internal layer of the membrane is described by Eq. 4⁴⁴

$$\begin{eqnarray}&&{j}_{{O}_{2}}=\displaystyle \frac{4\pi {N}_{b}{\bar{R}}_{b}^{5}g{\rho }_{l}}{9{\eta }_{l}}\end{eqnarray} \tag{ 4 }$$

were ${N}_{b}$ is the oxygen gas bubble density, ${\bar{R}}_{b}$ is the average radius of oxygen gas bubbles, g is the gravity, and ${\eta }_{l}$ is the liquid phase viscosity. The calculated by Eq. 4 oxygen permeation flux value of 6.7·10⁻⁸ mol cm⁻² s⁻¹ at 830 °С (${\bar{R}}_{b}$ ≈ 1.4 10⁻⁵ m (Ref. 42), ${\eta }_{l}$ ≈ 0.01 Pa s [Ref. 45], ${N}_{b}$ ≈ 4.0 10¹² m⁻³ [Ref. 42], and ${\rho }_{l}$ ≈ 5.6 10³ kg m⁻³) is in agreement with the experiment (Fig. 3f).

For a process to be cost-beneficial, it has been suggested^37,46,47 that the oxygen permeation flux through MIEC membrane should be higher than 6.9·10⁻⁷ mol cm⁻² s⁻¹. However, the oxygen permeation flux through the CuO-25 wt% Cu₅V₂O₁₀ MIEC-Redox membrane (Fig. 3f) is an order of magnitude lower than the targeted value. It was shown⁴⁸ that the targeted value could be achieved by reducing the thickness of the external layer of the membrane or increasing the ionic conductivity of this layer. Currently, we cannot purposefully change the thickness of the external layer of the membrane, since a relationship between the gradient of the chemical potential of oxygen and the thickness of the external layer is not established. This is a subject for further research. The systems containing liquid Bi₂O₃ or TeO₂ are also most suitable for further research, since the conductivity of oxygen ions in these liquid oxides is an order of magnitude higher than that of liquid V₂O₅.^49,50 Other systems containing liquid oxides with high oxygen ionic conductivity are also attractive.⁵¹ Upon reducing the external layer thickness of the membrane, the surface exchange reactions can limit the oxygen permeation flux.¹⁸ This problem can be solved by choosing the appropriate catalyst.¹⁹