electronic conducting membranes

Chem Soc Rev View Article Online

Published on 24 April 2017. Downloaded by Curtin University Library on 24/04/2017 15:23:32.

REVIEW ARTICLE

Cite this: DOI: 10.1039/c6cs00841k

View Journal

Designing CO2-resistant oxygen-selective mixed ionic–electronic conducting membranes: guidelines, recent advances, and forward directions Chi Zhang,a Jaka Sunarso

*b and Shaomin Liu*a

CO2 resistance is an enabling property for the wide-scale implementation of oxygen-selective mixed ionic– electronic conducting (MIEC) membranes in clean energy technologies, i.e., oxyfuel combustion, clean coal energy delivery, and catalytic membrane reactors for greener chemical synthesis. The significant rise in the number of studies over the past decade and the major progress in CO2-resistant MIEC materials warrant systematic guidelines on this topic. To this end, this review features the pertaining aspects in addition to the recent status and advances of the two most promising membrane materials, perovskite and fluorite-based dualphase materials. We explain how to quantify and design CO2 resistant membranes using the Lewis acid–base reaction concept and thermodynamics perspective and highlight the relevant characterization techniques. For perovskite materials, a trade-off generally exists between CO2 resistance and O2 permeability. Fluorite materials, despite their inherent CO2 resistance, typically have low O2 permeability but this can be improved via different approaches including thin film technology and the recently developed minimum internal electronic short-circuit second phase and external electronic short-circuit decoration. We then elaborate the two main future directions that are centralized around the development of new oxide compositions capable of featuring simultaneously Received 25th November 2016

high CO2 resistance and O2 permeability and the exploitation of phase reactions to create a new conductive

DOI: 10.1039/c6cs00841k

phase along the grain boundaries of dual-phase materials. The final part of the review discusses various complimentary characterization techniques and the relevant studies that can provide insights into the degradation

rsc.li/chem-soc-rev

mechanism of oxide-based materials upon exposure to CO2.

a

Department of Chemical Engineering, Curtin University, Perth, Western Australia 6845, Australia. E-mail: [email protected]; Fax: +61 892662681; Tel: +61 892669056 b Faculty of Engineering, Computing and Science, Swinburne University of Technology, Jalan Simpang Tiga, 93350 Kuching, Sarawak, Malaysia. E-mail: [email protected], [email protected]; Fax: +60 82260813; Tel: +60 82260709

Chi Zhang

Chi Zhang started his PhD study in Curtin University in 2013 under the supervision of Prof. Shaomin Liu. He received his BS in 2010 and MS in 2013 from Shandong University under the supervision of Prof. Zhonghua Zhang. His current research activities are mainly focused on the development of CO2-resistant MIEC materials, the electrochemistry of nonstoichiometric solid oxides, and oxygen permeation membranes.

This journal is © The Royal Society of Chemistry 2017

Jaka Sunarso joined Swinburne University of Technology in Malaysia (Sarawak Campus) as a Senior Lecturer in 2016. He received his PhD in Chemical Engineering from the University of Queensland in Australia in 2010. He was awarded an Alfred Deakin Postdoctoral Research Fellowship to do postdoctoral training at Deakin University in Australia between 2011 and 2013. He then secured a Banting Jaka Sunarso Fellowship for his 2nd postdoctoral training at the University of Waterloo in Canada between 2013 and 2015. His main research interests are batteries, fuel cells, and membranes.

Chem. Soc. Rev.

View Article Online

Review Article


1. Background The perpetual growth of energy demand has lead to the continual consumption of fossil fuels which is projected to increase by B27% over the next 24 year-period, leading to a parallel increase in the United States CO2 emissions from 6000 million tons per year (in 2006) to 8000 million tons per year (by 2030) according to the Energy Information Administration (EIA) within the United States Department of Energy (U.S. D.O.E.).1 Hydrocarbons in particular will still form a substantial part of the fuel supply, supplying more than 60% of the world’s energy demand over the next 20 years.2 This will contribute to a significant increase in anthropogenic carbon dioxide (CO2) emissions. The global atmospheric CO2 concentration itself has increased from about 315 ppm (in 1958) to 375 ppm (at the end of 2005).3 One of the potential approaches is oxyfuel technology, primarily based on pulverized coal combustion in a CO2/O2 atmosphere which offers facile CO2 recovery, low NOx emissions, and high desulfurization efficiency. There are several different oxyfuel power cycles, including the advanced zeroemission power plant (AZEPP), Graz and water cycle, MATIANT and Feher supercritical CO2 cycles, the zero-emission power plant (ZEPP), the zero-emission ionic transport membrane (ITM) oxyfuel plant, etc.; the details of which can be found in another review paper.2 In the oxyfuel process, fuel (e.g., coal, oil, gas, and other hydrocarbons) is burned in a mixture of oxygen and recirculated flue gas (which comprises mainly CO2 and water vapor as well as traces of N2, O2, SO2, and NOx).3 This is because fuel combustion using pure O2 leads to extremely high combustion temperatures.2 The boilers in the current existing power plants can be retrofitted to oxyfuel configurations given the match between their maximum temperature ranges, which would not otherwise be possible in the case of combustion using pure O2. Furthermore, the absence of nitrogen in the flue gas allows easy processing of CO2, i.e., its

Shaomin Liu is currently a Professor in the Department of Chemical Engineering at Curtin University. He received his PhD from the National University of Singapore in 2002 and then worked as a postdoctoral research fellow in the Department of Chemical Engineering at California Institute of Technology from 2002 to 2005. From the end of 2008, he has been working in Curtin University, Australia. He is Shaomin Liu the recipient of an ARC Future Fellowship (2012–2016) and an ARC Australian Research Fellowship (2008–2012). His main research interests are membranes for gas separation, membrane reactors for gas reactions, and nanoporous materials for advanced oxidation.

Chem. Soc. Rev.

Chem Soc Rev

purification up to 96–99% which can then be compressed and condensed into liquid CO2 for sequestration or used in other processes.3 The required oxygen can be supplied by cryogenic distillation or a ceramic-based ionic transport membrane; both of which will incur extra energy and capital penalties from additional devices such as a compressor relative to the conventional air-fired power plants. The latter technology, specifically known as the oxygen ionic transport membrane (ITM), has rapidly become one of the key technologies within clean combustion based energy generation as demonstrated by substantial investment from developed countries such as US $90 million from the Department of Energy on Ionic Transport Membrane Oxygen Development Program in the U.S. and OXYCOAL-AC in Germany.3,4 Compared to the cryogenic process, ITM technology provides up to 48% capital saving and up to 69% power saving for applications such as integrated gasification combined cycle (IGCC), decarbonized fuel, oxygen enrichment, oxyfuel, and gas to liquid.3 For the oxyfuel process, ITM, due to its reduced auxiliary power requirement, is more efficient in terms of cost and energy. Moreover, if good combustion performance can be achieved at low oxygen concentrations in a CO2/O2 mixture, the capital and operational costs of an ITM based oxyfuel plant can be reduced further.3 Within the oxyfuel context, the availability of CO2resistant membrane materials would enable oxygen separation from air in the feed side of the air separation unit directly into the recirculated flue gas (which contains mostly CO2) on the other side. This would simplify the configuration and provide lower demand on the operating pressure and the membrane area as well as higher thermal efficiency relative to the case where direct contact between the membrane and CO2 is not allowed.5 Another important application for CO2-resistant membranes lies in the area of membrane reactors for greener chemical synthesis, including partial oxidation of methane (CH4) to syngas (H2 and CO), oxidative coupling of CH4 to C2 (ethylene and/or ethane), oxidative dehydrogenation of light alkanes to olefins, CO2 thermal conversion to CO and water splitting to hydrogen (H2).6–10 In these important reactions, CO2, H2, and CH4 are inevitably involved as the feed or product gases, requiring the membrane to be at least CO2-tolerant. The best performance of such membrane reactors requires the matched properties of oxygen flux and catalytic efficiency to consume the permeated oxygen. The intermediate value of oxygen flux for CO2-resistant membranes also perfectly fits within the membrane reactor context to optimize the product yields, because otherwise a too high oxygen concentration in the permeate side will highly oxidize the products leading to low product selectivity. Single-chamber solid oxide fuel cells (SC-SOFCs) represent another application area using CO2-resistant membranes. In the SC-SOFC system, the anode and cathode are assembled in one gas chamber filled with fuel, oxidant gases, and reaction products. The presence of CO2 is unavoidable in a SC-SOFC reaction product stream, thus CO2-resistant electrode materials are required in this fuel cell system.11–14 The last 20 years have witnessed the rapid development of the ITM, especially its application in the oxygen permeation areas. Fig. 1(a) and (b) display the number of publications


View Article Online


Chem Soc Rev

Fig. 1

Review Article

The number of (a) publications and (b) patents per year between 1997 and 2017 as a function of time in the ITM area.

(articles, reviews, and proceedings) and patents as a function of time during the past 20 years. The number of published studies has consistently increased since 1997; reaching more than 120 publications per year beyond 2010. The emerging hundreds of publications on this ITM topic highlights substantial interest in this technology which has resulted in several excellent review papers that cover the background, related technologies and applications, material structures and compositions, transport mechanisms and theories, material synthesis and characterization, oxygen permeation performance overviews, and/or future directions.6,9,10,15–19 Zhang et al., for example, overviewed the structure of perovskite-related compounds, the oxygen permeation rate determining factors, and the material selection guidelines.16 Sunarso et al. provided a comprehensive account on all aspects of mixed ionic–electronic conducting (MIEC) ceramic-based membranes with an emphasis on the fluorite and perovskite-based structures, defect theory, transport mechanisms, and preparation methods.15 Still, the CO2 resistance aspect of ITM technology has been rather overlooked despite its significance to realize the wide applications of this technology within oxyfuel combustion, clean coal energy delivery, and catalytic membrane reactors. This review thus aims to address this gap by featuring the current status and advances on CO2resistant MIEC membranes as well as the underlying concepts and approaches. The emphasis has been placed on perovskite and dual-phase materials. The review is divided into six sections. This first section justifies the technology significance and its context. The second section features the pertaining aspects to develop CO2-resistant MIEC membranes according to four major categories, i.e., perovskite materials, dual-phase materials, other materials, and engineering approaches, and overviews the relevant industrial developments. The third section provides comparative insights into the performances of perovskite and dual-phase membranes. The fourth section, in turn, overviews the two latest directions for the development


of CO2-resistant MIEC membranes. The fifth section discusses the membrane degradation behavior in a CO2 atmosphere, mostly for the CO2-prone perovskite materials. Finally, the sixth section discusses the poisoning effects of SO2 and other elements.

2. CO2-resistant oxygen-selective mixed ionic–electronic conducting membranes Perovskite compounds have drawn major attention for application as mixed ionic–electronic conducting (MIEC) membranes due to their high oxygen flux determined by their inherently high ionic/ electronic conductivities and surface oxygen exchange reaction kinetics. Perovskite MIEC compounds generally have overwhelming higher electronic conductivity than ionic conductivity. Compared to perovskite, fluorite materials have lower ionic conductivity and substantially lower electronic conductivity but possess high chemical stability to tolerate these acidic or reducing gases. Thereby, a fluorite material is usually designed as a novel robust membrane but accompanied by a second electronic conducting phase to compensate its low electronic conductivity limitation. The property differences between these compounds can be understood by inspecting their specific structures as schematized in Fig. 2. 2.1.

Perovskite materials

2.1.1. Structure and basic concepts. Fig. 2(a) and (b) depicts the ideal perovskite structure and its packing arrangement. Perovskite compounds are defined as those having an ABO3 formula. A and B here are two different metal cations with different sizes. Size limitation dictates the ionic radius range of A and B where generally the A and B cations have an ionic radius between 1.1–1.8 Å and 0.62–1 Å, respectively.20 As such, the A-site cation is normally from the lanthanides, e.g., La, Pr, and Gd or alkaline earth metals, e.g., Ba and Sr while the B-site

Chem. Soc. Rev.

View Article Online


Review Article

Chem Soc Rev

Fig. 2 Ideal structure and the packing arrangement for: (a and b) perovskite ABO3 compounds represented by SrTiO3 (green atoms – A (Sr), light blue atoms – B (Ti), red atoms – O (O)); and (c and d) fluorite AO2 compounds represented by CaF2 (blue atoms – A (Ca), grey atoms – O (F)).

cation can be a transition metal, e.g., Mn, Co, Ni, Cr, and Fe or a non-transition metal, e.g., Al and Ga.15 SrTiO3 serves as a typical example. The perovskite structure consists essentially of a framework of corner-shared BO6 octahedra where a large A cation sits in a cavity between octahedra.21 Each B cation is coordinated by 6 O anions creating 6 equidistant B–O bonds while each A cation is coordinated by 12 O anions creating 12 equidistant A–O bonds. A and B cations may have different oxidation states as represented by the formula of A2+B4+O3 or A1+B5+O3 or A3+B3+O3. The framework dimensions are fixed by the B–O bond length and in order for the cube shown in Fig. 2(a) to form, the cube edge should be equal to twice the sum of the ionic radius of the B cation and the ionic radius of the O anion while the face diagonal should be equal to twice the sum of the ionic radius of the A cation and the ionic radius of the O anion, assuming that the A cation is in complete contact with all its neighboring ions. By using the Pythagorean theorem, it becomes apparent that: (RA + RO)2 = 2(RB + RO)2 pffiffiffi RA þ RO ¼ 2ðRB þ RO Þ RA þ RO 1 ¼ pffiffiffi 2ðRB þ RO Þ RA þ RO t ¼ pffiffiffi 2ðRB þ RO Þ

(1) (2) (3)

(4)

Therefore, eqn (3) determines the relationship between the ionic radius values for A, B and O. The ratio of the sum of the ionic radius of A and O to the square root of 2 multiplied by the sum of the ionic radius of B and O should ideally be equal to 1 for the cubic structure to form. This ratio was generalized as the Goldschmidt tolerance factor (eqn (4)), a simple predictor variable which has been widely used to design a specific perovskite structure from numerous metal components.21,22 The empirical ionic radius values for various metal cations have been listed by Shannon.23

Chem. Soc. Rev.

Most perovskite compounds actually exist in their ‘‘nonstoichiometric’’ analogues, i.e., as ABO3%d compounds where some oxygen anions are missing from the lattice. The d here represents the non-stoichiometry parameter created upon incorporating A and B cations with the total oxidation state of less than 6, e.g., 5 from A2+ and B3+ or A1+ and B4+. In fact, the presence of such ‘‘oxygen vacancy’’ defects is beneficial to provide a pathway for oxygen ionic diffusion (for oxygen ions hopping from one empty site to another empty site) and is the primary contributing factor for the oxygen ionic conductivity. Moreover, the electronic conductivity can also be created in perovskite by choosing transition metals as the B-site metal component. It is well-known that transition metals can afford more than one oxidation state during reduction and oxidation reactions. Oxygen vacancies are often formed at increasing temperatures in conjunction with the reduction of transition metal components which are simultaneously accompanied by electron transfer reactions.24,25 The presence of this simultaneous ionic and electronic conductivity characterizes its ‘‘mixed ionic– electronic conducting’’ definition.15 Another main advantage of using perovskite MIEC is its capability to host different metal oxides as represented by A1%x%yA 0 xA00 yB1%x%yB 0 xB00 yO3%d. To this end, the structure and properties of the resultant perovskite can be tuned by compositional tailoring to fit particular applications, e.g., high ionic conductivity, electronic conductivity or catalytic activity. As an example, if electronic conductivity is not required, a nontransition metal such as Ga can be incorporated into the B-site. Such a perovskite compound, e.g., LaGaO3 serves as an excellent ionic conductor.26 2.1.2. Quantifying the extent of reaction with CO2. Highly permeable oxygen-selective MIEC perovskite materials normally have alkaline earth metal constituents in their A-site which readily react with acidic CO2. In the perspective of obtaining larger lattice size and higher oxygen permeability, Ba that has a large ionic radius of 1.61 Å (at the coordination number of 12) becomes an attractive A-site component.23 CO2 exposure on such alkaline earth metal containing perovskites generally leads to the deterioration of the original structure and the resultant degradation of the oxygen permeation performance. For example, the O2 flux of the Ba0.5Sr0.5Co0.8Fe0.2O3%d membrane, when exposed to CO2, decreases to zero rapidly, i.e., within several minutes.27–29 Identical degradation was also observed on the BaCo0.4Fe0.4Nb0.2O3%d membrane during exposure to CO2.30 Post characterization of the CO2-exposed membrane has been used to evaluate the extent of structure decomposition which will be elaborated in more detail in the fifth section. Generally, the decomposed part on the surface is composed of a carbonate layer and a decomposed zone.27,29 2.1.2.1. Thermodynamics perspective. The susceptibility of perovskite oxide to form carbonates upon contact with CO2 can be assessed from a thermodynamics point of view. Perovskite oxide, i.e., ABO3 can be considered as one among many possible phases in the binary mixture of AO–BO2.31 Its interaction and equilibrium with gaseous species like CO2 can be determined


View Article Online


Chem Soc Rev

using thermodynamics relationships. To simplify, the oxygen nonstoichiometry (d) of perovskite oxide is normally assumed to be negligible so that the effect of oxygen partial pressure can be neglected. The formation of different intermediate phases at different ratios of AO to BO2 depends on the reaction between AO and BO2 and eventually, the activity ratio of AO to BO2. In the presence of CO2, another variable, i.e., CO2 partial pressure comes into play. For example, for the binary system of SrO–TiO2, the stability diagram of SrO–TiO2–CO2 can be developed which correlates different phases as a function of the activity ratio of SrO to TiO2 and CO2 partial pressure as illustrated in Fig. 3(a) and (b). The detailed calculations to generate such stability diagrams will not be discussed here since it is available in the work of Frade and co-workers and Yokokawa et al.31–33 The stability diagrams are useful to determine the tolerance range to CO2 as well as to compare the stability between different perovskite oxides.31 Fig. 3(a) for example shows that at 500 K (227 1C), SrTiO3 is more stable to carbonation relative to SrO since SrCO3 is formed from reaction with SrTiO3 at higher CO2 partial pressure. It is worth noting that the presence of excess TiO2 actually increases the stability of SrTiO3 (Fig. 3(a)), leading to the better stability of SrTiO3 against CO2 than the single SrO phase. Within this context, the composition (its heterogeneity) may have a substantial effect on CO2

Review Article

tolerance; the variation of which can be minimized using a solution route rather than a solid-state route which requires a higher temperature for a prolonged sintering time. Upon comparing Fig. 3(a) and (c), it becomes clear that at 500 K, CaTiO3 is more stable than SrTiO3 given the higher low and high pressure (PL and PH) for CaCO3 formation from CaTiO3 (relative to SrTiO3 case). Likewise, SrTiO3 is more stable than BaTiO3 at 773 K (500 1C) when comparing Fig. 3(b) and (d). The low and high pressure limits for particular carbonate formation can also be summarized to correlate these limits with the temperature as shown in Fig. 3(e) for different titanium-based perovskite oxides. Furthermore, the relative gains in CO2 resistance of ABO3 perovskite oxide with respect to the AO metal oxide appear to be connected to the Goldschmidt tolerance factor with a higher factor favoring higher resistance (Fig. 3(f)).31 The correlation between relative gains and tolerance factor for titanium based perovskite oxides implies the general validity of the plot given the clearly established trend between the experimental points and the regression line (Fig. 3(f)). If such a relationship indeed exists and is extendable to every other perovskite oxide system, stability can be predicted using this plot for those systems where the thermodynamics data are absent. From the above discussion, it becomes clear that there is a difference between the stability of AO against CO2 relative to the stability of the perovskite where

Fig. 3 Stability diagrams for the SrO–TiO2–CO2 system at (a) 500 K and (b) 773 K; stability diagrams for the (c) CaO–TiO2–CO2 system at 500 K and (d) BaO–TiO2–CO2 system at 773 K; (e) lower and upper tolerance limits for alkaline earth titanates; and (f) relative gains in CO2 resistance for titanium and zirconium-based perovskites. Reproduced with permission from ref. 31. Copyright 2011, Elsevier Inc.


Chem. Soc. Rev.

View Article Online


Review Article

Chem Soc Rev

the A-site element becomes the lattice component together with the other elements. One example is CaO which easily reacts with water to form its hydrate. However, in our research, we find that La0.8Ca0.2FeO3%d is very chemically inert and will maintain its perovskite structure in hot water up to 100 1C. This difference is attributed to the stabilization energy of the perovskite oxides which will be discussed below. The stabilization energy of the perovskite oxides may also be taken into consideration to analyze the carbonate formation tendency.34 The stabilization energy of perovskite oxides is defined as the enthalpy change for perovskite formation from the constituent oxides (AO and BO2), which can be written as: d(ABO3) = [Df H0(ABO3) % {Df H0(AOn) + Df H0(BOm)}]

(5)

The chemical stability of the perovskite oxides against reactions with a particular chemical constituent such as CO2 is associated with the stabilization energy of the perovskites from their constituent oxides (AO and BO2). Moreover, the stabilization energy correlates with the Goldschmidt tolerance factor t with the ideal t value (t = 1) indicating the highest stabilization, while smaller t (t o 1) leads to lower stability.33,35,36 A typical ABO3 perovskite reacts with CO2 to form ACO3 as follows: ABO3 + CO2 " ACO3 + BO2

(6)

The reaction can be resolved into the following two reactions energetically: ABO3 " AO + BO2

(7)

AO + CO2 " ACO3

(8)

Fig. 4 displays the Gibbs energy change for carbonate formation for some typical strontium and barium perovskite oxides. For each oxide, the stability against carbonate formation is a function of temperature. With decreasing temperature, the Gibbs energy change becomes lower, indicating a stronger tendency

Fig. 4 Gibbs energy change for carbonate formation for (a) strontiumbased perovskites and (b) barium-based perovskites; (1) AO + CO2 (g) = ACO3; (2) ACeO3 + CO2 (g) = ACO3 + CeO2; (3) AZrO3 + CO2 (g) = ACO3 + ZrO2; (4) ATiO3 + CO2 (g) = ACO3 + TiO2 (A = Sr or Ba). Reproduced with permission from ref. 34. Copyright 1992, Elsevier Inc.

Chem. Soc. Rev.

to form carbonates. For strontium and barium perovskite oxides, the stabilization energy increases (becomes negatively larger) with decreasing B4+ ion radius (0.87 Å 4 0.72 Å 4 0.605 Å for Ce, Zr, and Ti, respectively); resulting in a higher Gibbs energy change and lower tendency for carbonate formation.23,34 The stability of the perovskite oxides was mainly determined by the less stable metal oxide constituent like SrO in the Sr-doped lanthanum perovskite oxide.33 Comparing strontium and barium, although the tendency of carbonate formation is stronger for barium oxide, the stabilization energy of barium perovskite is (negatively) larger, resulting in the relatively similar stability against carbonate formation for strontium and barium perovskites.34 This also serves as an evidence that the stability of AO against CO2 in the perovskite ABO3 is different from the stability of the single AO phase. This theory was summarized and discussed in Martynczuk’s paper.37 This stabilization energy concept was applied by Klande et al. to explain the better CO2 stability of La0.6Sr0.4Co0.8Fe0.2O3%d compared with SrCo0.8Fe0.2O3%d.38 They have proposed that the perovskite stabilization energies increase with increasing lanthanum content, leading to the higher stabilization effect for La0.6Sr0.4Co0.8Fe0.2O3%d compared to SrCo0.8Fe0.2O3%d, which prohibits the reactions in eqn (7) and (8). It is worth noting that the stabilization energy is just one of the possible aspects affecting the stability of perovskite against CO2. Other parameters may need to be accounted for. Another popular way is to estimate the CO2 resistance of a perovskite oxide based on its carbonate formation at a specific temperature and CO2 partial pressure.39–42 Fig. 5(a)–(d) represent the typical Ellingham diagrams for evaluation of carbonate formation. Ellingham diagrams have been utilized and disseminated by Feldhoff’s group to estimate the stability of metal oxides in a CO2-containing atmosphere at different CO2 partial pressures.39–42 To show its reproducibility, we re-calculated and re-plotted the Ellingham diagrams for alkaline metal carbonates (Li2CO3, Na2CO3, K2CO3, and Cs2CO3 in Fig. 5(a)), alkaline-earth metal carbonates (MgCO3, CaCO3, SrCO3, and BaCO3 in Fig. 5(b)), and transition metal carbonates (NiCO3, ZnCO3, FeCO3, CoCO3, Ag2CO3, and MnCO3 in Fig. 5(c)). The thermodynamic data from the data compilation of Barin et al. were used to obtain these three Ellingham diagrams.43 The absence of the thermodynamic data for lanthanide carbonates (La2(CO3)3, La2O2CO3, and Pr2O2CO3) requires their experimental measurements as performed by Feldhoff’s group to obtain these compounds’ diagrams (Fig. 5(d)).42 In Ellingham diagrams, the dashed lines represent the chemical potential of CO2 at a particular CO2 partial pressure while the bold continuous lines represent the chemical potential of CO2 for decomposition of a particular carbonate; both are functions of temperature. The stability of a particular carbonate is evaluated by comparing the CO2 chemical potential at a particular CO2 partial pressure and temperature against the decomposition chemical potential at the same temperature. If the former is lower than the latter, the carbonate is thermodynamically unstable. In another word, the carbonate would form if the former is higher than the latter. For example, at 800 K (527 1C) and 1 Pa, the chemical potential of CO2 is %75 kJ mol%1


View Article Online

Review Article


Chem Soc Rev

Fig. 5 Ellingham diagrams for carbonate decomposition at different CO2 partial pressures: (a) Li2CO3, Na2CO3, K2CO3, and Cs2CO3; (b) MgCO3, CaCO3, SrCO3, and BaCO3; (c) NiCO3, ZnCO3, FeCO3, CoCO3, Ag2CO3, and MnCO3; and (d) La2(CO3)3, La2O2CO3, and Pr2O2CO3. The dashed lines represent the chemical potential of CO2 at a particular CO2 partial pressure while the bold continuous lines represent the chemical potential of CO2 for decomposition of a particular carbonate; both are functions of temperature ((a)–(c) were replotted based on the Ellingham diagrams from ref. 39–42; the thermodynamic data are taken from ref. 43; (d) was reproduced with permission from ref. 42. Copyright 2013, Elsevier Inc).

while the chemical potential of decomposition of BaCO3 and CaCO3 is %130 kJ mol%1 and %53 kJ mol%1, respectively (Fig. 5(a)). Under this condition, BaCO3 may form whereas CaCO3 is not likely to form. Here, we only provide Ellingham diagrams for carbonates whose thermodynamic data are available. Diagrams for other carbonates can be identically constructed using the following procedure. First, the CO2 chemical potential at different CO2 & partial pressures is calculated using mCO2 ¼ RT Inðp CO2 =p CO2 Þ

(kJ mol%1) and is plotted as the y-axis with temperature and T as the x-axis. Second, the Gibbs free energy change, DGr for the carbonate decomposition reaction is calculated using thermodynamic data and all of the reactions are normalized to generate one mole of CO2. Taking BaCO3 as an example, the Gibbs energy


for the formation of BaCO3 (DGf,BaCO3), BaO (DGf,BaO), and CO2 (DGf,CO2) can be collected from the data compilation of Barin et al.43 The Gibbs free energy change DGr for the carbonate decomposition reaction of BaCO3 (s) - BaO (s) + CO2 (g) is calculated using the formula of DGr = DGf,BaCO3 % DGf,BaO % DGf,CO2. For each individual temperature, there is one corresponding DGr. The DGr values are then plotted in the diagram as a function of temperature; forming a line with a positive slope. Accordingly, at a certain temperature, if DGr is higher than the chemical potential of CO2 (mCO2) at a given CO2 partial pressure, the carbonate will decompose. For example, at 1000 K, the Gibbs free energy change DGr for BaCO3 (s) - BaO (s) + CO2 (g) is %101.85 kJ mol%1, which is higher than the CO2 chemical potential of %134.11 kJ mol%1 at a CO2 partial pressure of 10%2 Pa.

Chem. Soc. Rev.

View Article Online


Review Article

This indicates that BaCO3 is not stable at 1000 K when the surrounding CO2 partial pressure is 10%2 Pa. The Ellingham diagram should be viewed as an approximate thermodynamic predictor for the carbonate stability. Other factors such as the stabilization energy of the perovskites and the kinetics of the carbonate decomposition may play a role. In most carbonate cases, there is also a correlation between the stability and the carbonate size where larger ionic radius normally leads to larger crystal energy for the respective metal carbonate and thus, higher stability. For instance, Fig. 5(b) implies increasing stability from CaCO3 to SrCO3 and to BaCO3, in parallel with increasing ionic radius of the metal cations, i.e., 1.34 Å (Ca2+), 1.44 Å (Sr2+), and 1.61 Å (Ba2+) (all with a coordination number of 12).23,41 Again, the smaller ionic radius for Ni2+ (0.69 Å – coordination number of 6) compared to La3+ (1.36 Å – coordination number of 12) and Pr3+ (1.179 Å – coordination number of 9) is in accordance with the lower stability of NiCO3 with respect to La2(CO3)3, La2O2CO3, and Pr2(CO3)3 (Fig. 5(d)).23,42 The charge (or the oxidation state) of the metal cation, on the other hand has a reciprocal relationship with the stability of the carbonate with lower charge favoring higher stability.44 2.1.2.2. Chemistry perspective. Although the thermodynamic analysis may provide a glimpse into the chemical stability of perovskite oxides, it requires the presence of thermodynamic data which are not always readily available for all compounds. The extent of reaction between perovskite oxide and CO2 can also be rationalized using a Lewis acid–base concept to address this shortcoming.30,45–47 In this context, perovskite oxide is a base (or an electron pair donor) while CO2 is an acid (or an electron pair acceptor). The basicity of perovskite oxide, i.e., the tendency to give up (or donate) electrons to the adsorbed CO2 molecule which varies at different temperatures and oxygen partial pressures, becomes the determining variable. Higher basicity translates to intensified reactions. Sanderson’s electronegativity and the oxidation state of the cations are the two parameters that reflect the relative basicity of the metal oxides.48 A higher oxidation state generally signifies lower basicity (higher acidity) and accordingly, higher CO2 resistance. The descriptor variable that indicates the relative acidity of metal oxide has been given by Jeong et al. in the form of the NM % 2dM value where NM represents the formal oxidation of the metal in the compound and dM represents the partial charges of the metal ions.48 Fig. 6 depicts such NM % 2dM values for 101 metal oxides where a higher value reflects higher acidity. Moreover, the data in Fig. 6 clearly indicates that the surface acidity of a metal is a function of the oxidation state of the cation, the electronegativity of the cation, and the electronegativity of the anion. Given the explicit acidity value, Fig. 6 becomes an invaluable reference for the design of new CO2resistant oxide compositions. To obtain this comprehensive correlation, Jeong et al. measured the intramolecular chargetransfer (ICMT) energies for adsorption of two different dyes, i.e., alizarin and 4-methoxyalizarin onto various metal oxides

Chem. Soc. Rev.

Chem Soc Rev

and sulfides by diffuse reflectance spectroscopy. They found correlations between the ICMT energies and NM % 2dM values which allowed them to calculate Sanderson’s electronegativity values for all tested samples including those for lanthanides (Ln2O3) and CeO2 which were not available before their work.48 Since the reactions occur effectively only at the basic sites or sites with negative charge, i.e., on oxide ions, the basicity of perovskite oxide can also be measured using the O1s binding energy (BE) from X-ray photoelectron spectroscopy (XPS) which reflects the electron density (or the oxidation state) of oxide ions.49 Higher O1s binding energy implies lower electron density (or higher oxidation state) and consequently, lower basicity.49 The O1s peak on XPS data generally consists of contributions from 3 oxygen components, i.e., oxygen in adsorbed water (at the highest BE), adsorbed oxygen (at the medium BE), and lattice oxygen (at the lowest BE) (Fig. 7(a), (d) and (e)). The binding energy for lattice oxygen in most cases can be taken as an accurate measure for basicity. For example, Ti-doped SrCo0.8Fe0.2O3%d was reported to be more stable to CO2 than SrCo0.8Fe0.2O3%d; consistent with the fact that when the Ti content (x) in Sr(Co0.8Fe0.2)1%xTixO3%d was increased from 0 to 0.4, the O1s BE of lattice oxygen increased from 528.4 eV to 529.1 eV (Fig. 7(a) and (b)).45 Partial substitution of Co and Fe in SrCo0.8Fe0.2O3%d with Ta (i.e., Sr(Co0.8Fe0.2)0.9Ta0.1O3%d) also led to the shifting of the lattice oxygen peak to a higher binding energy value, i.e., by 0.4 eV from 528.7 eV (for SrCo0.8Fe0.2O3%d) to 529.1 eV (for Ta-doped SrCo0.8Fe0.2O3%d) (Fig. 7(c)).46 Accordingly, Ta-doped SrCo0.8Fe0.2O3%d showed enhanced stability to CO2 with respect to SrCo0.8Fe0.2O3%d. Moreover, exposing SrCo0.8Fe0.2O3%d to higher oxygen partial pressure, e.g., 1 bar relative to 10%4 bar increased the CO2 resistance in agreement with increased O1s BE for lattice oxygen at higher oxygen partial pressure, i.e., 529.5 eV relative to 528.9 eV (Fig. 7(d) and (e)).47 An effective way to reduce the basicity of the perovskite oxide is by decreasing the electron density of oxide ions. This can be achieved using metal elements whose cations strongly attract the electron cloud away from oxygen anion, i.e., metal cations with a high oxidation state, e.g., Ta5+ and Ti4+ as demonstrated for Ta and Ti-doped SrCo0.8Fe0.2O3%d compounds.45,46 The bonding between such a metal cation and oxygen anion (e.g., bonding energy of Ta–O and Ti–O is 799.1 kJ mol%1 and 672.4 kJ mol%1, respectively) is substantially stronger than the bonding between a common metal cation and oxygen anion such as Co and Fe (e.g., bonding energy of Co–O and Fe–O is 384.5 kJ mol%1 and 390.4 kJ mol%1, respectively). Such a strong bond however suppresses the oxygen loss from the lattice which affects negatively the oxygen permeability. Therefore, a tradeoff exists between oxygen permeability and CO2 resistance. The average metal–oxygen bond energy (ABE) provides a simple quantitative predictor in this regard to assess the CO2 resistance.50 This parameter can be calculated as shown below for A1%xA 0 xB1%y%zB 0 yB00 zO3%d-type perovskite oxides (eqn (9)–(13)) and thus far has been exclusively used to determine the oxygen ionic diffusion in perovskite oxides.51,52


View Article Online


Chem Soc Rev

Review Article

Fig. 6 Comparison of the relative acidity of various oxides. Reproduced with permission from ref. 48. Copyright 2008, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

hABEi = hA–Oi + hB–Oi

(9)

hA–Oi = D(A–O) + D(A 0 –O)

(10)

(11) hB–Oi = D(B–O) + D(B 0 –O) + D(B00 –O) " # n DðA % OÞ ¼ ðxA =ðCNA ' mÞÞ ( DHAm On % m ' DHA % ' DO2 2 (12) " # n DðB % OÞ ¼ ðyB =ðCNB ' mÞÞ ( DHBm On % m ' DHB % ' DO2 2 (13) where xA and yB represents the mole fraction of A and B; HA(B)mOn and HA(B) represents the enthalpy of formation of one mole of A(B)mOn and the sublimation energy of A(B) metal at 25 1C, available from the thermodynamics database; CNA(B) represents the coordination number of A and B site cations (CNA = 12, CNB = 6); and DO2 represents the dissociation energy of O2 (i.e., 500.2 kJ mol%1). 2.1.2.3. Other factors like chemical adsorption. In general, two subsequent processes manifest over the course of interaction


between CO2 and oxide-based membranes, i.e., the CO2 adsorption on the membrane surface and carbonate formation. In most cases, the significant poisoning effect of CO2 as reflected by the considerable deterioration of original flux is mainly attributed to the chemical adsorption of CO2 onto the membrane surface since carbonate formation does not occur instantaneously.53 Three progressive steps are involved in the oxygen transport through MIEC membranes, i.e., (1) the surface-exchange reaction of oxygen gas molecules and oxygen vacancies; (2) bulk-diffusions of lattice oxygen ions and electrons/ electron holes across the bulk membrane; and (3) the surfaceexchange reaction of lattice oxygen and electrons.54,55 The oxygen transport rate is thus closely associated with the surface oxygen vacancies. To this end, when CO2 is used as the sweep gas, oxygen ions on CO2 may be adsorbed chemically on the oxygen vacancies of the membrane.56,57 This leads to the decrease in oxygen permeation rate with increasing time. Such a CO2 adsorption effect is responsible for the rapid deterioration of O2 flux after CO2 is introduced and also the fast recovery of O2 flux to larger values after CO2 is removed.53

Chem. Soc. Rev.

View Article Online


Review Article

Chem Soc Rev

Fig. 7 (a) X-ray photoelectron spectroscopy (XPS) spectrum of O1s binding energy (BE) for Sr(Co0.8Fe0.2)0.9Ti0.1O3%d; (b) the O1s BE of lattice oxygen as a function of Ti content (x) on Sr(Co0.8Fe0.2)1%xTixO3%d; (c) the O1s BE of adsorbed oxygen and lattice oxygen on SrCo0.8Fe0.2O3%d and Sr(Co0.8Fe0.2)0.9Ta0.1O3%d; and XPS spectrum of SrCo0.8Fe0.2O3%d quenched from 20 hour annealing at 950 1C in an oxygen partial pressure of (d) 10%4 bar and (e) 1 bar. Reproduced with permission from ref. 45–47. Copyright 2009, 2011, and 2014, Elsevier Inc.

Using temperature programmed desorption, Yan et al. showed the competitive adsorption of O2 and CO2 on the Ba0.5Sr0.5Co0.8Fe0.2O3%d (BSCF5582) surface.58 They studied the CO2 adsorption effect on BSCF5582 in the presence of O2 and H2O; highlighting CO2 adsorption as one of the factors reflecting the CO2 tolerance of perovskite. The reactivity of CO2 and Sr2+ and/or Ba2+ on the BSCF5582 surface increased with increasing temperature when only CO2 was adsorbed onto the surface. However, when CO2 and O2 were co-adsorbed on the surface, CO2 adsorption was restrained, especially when O2 adsorption became dominant at 700 1C. Furthermore, the presence of water led to a more severe CO2 poisoning effect due to bicarbonate formation. Winnubst and co-workers showed that the addition of a low amount of O2 to the CO2 atmosphere enhanced the CO2 tolerance of the SrCo0.8Fe0.2O3%d membrane.47 They found that when pure CO2 was used as a sweep gas, the oxygen flux of SrCo0.8Fe0.2O3%d reduced to zero within 25 hours. When 5 vol% O2 was introduced into predominantly CO2 sweep gas, the oxygen flux decreased by 34% over the first 10 hour period before slowly increasing to 90% of its initial flux value within the following 50 hour period and then stabilized around this value. Fang et al. proposed another factor that influences the CO2 tolerance of a fluorite–perovskite-based composite membrane, i.e., an 80 wt% Ce0.8Gd0.15Cu0.05O2%d–20 wt% SrFeO3%d (CGC–SF) dual-phase membrane. This factor is related to the chemical

Chem. Soc. Rev.

potential of oxygen (mO2) across the membrane.59 When O2-enriched air was used as the feed gas (i.e., the chemical potential of O2 increased), the membrane exhibited significantly enhanced CO2-resistance when exposed to pure CO2 sweep gas. They hypothesized that the carbonate formation is thermodynamically and kinetically limited. Therefore, higher mO2 can suppress the adsorption of CO2 significantly by generating a faster O2 releasing rate and lower oxygen vacancy concentration on the permeate side, contributing to less carbonate formation during long-term exposure to CO2.58 Nonetheless, another research work on Ba(Co,Fe,Nb)O3%d indicates that the increase of the oxygen partial pressure significantly promotes BaCO3 formation, which suggests that oxygen molecules probably become the oxygen source for carbonate formation. However, they recommended further investigation to reveal the real oxygen source for carbonate formation.30 From the above discussion, the presence of O2 in the sweep gas containing CO2 appears to enhance the stability of a CO2sensitive membrane.47 Using an oxygen-enriched feed gas has the same effect on stability.59 O2 flux was reported to increase with increasing O2 partial pressure in the feed gas.60–62 This increase was attributed to the higher driving force ( pO2 gradient) across the membrane.60 An excess amount of O2 also appears to prolong the stability of the membrane. Even in an oxy-fuel combustion reactor, where CH4 was applied as the sweep gas, the CO2-sensitive BSCF5582 membrane remained stable for more than 200 h when oxygen was present in the sweep side.63


View Article Online


Chem Soc Rev

By taking into account the gas (O2) concentration polarization effect observed during the O2 permeation process,60–62,64 we propose a plausible explanation here. During O2 permeation, if the O2 release rate on the sweep side is lower than the O2 generation rate through the membrane, the excess O2 becomes accumulated on the sweep side. Such O2 accumulation in turn manifests into a ‘gas protection layer’ that minimizes the contact of CO2 with the membrane surface, thus protecting the membrane from reaction with CO2. Still, so far, no report is available that discusses in detail the O2 polarization effect on membrane stability against CO2. It is worth noting that O2 polarization can be alleviated by increasing the operation temperature or changing the sweep conditions.60,62 To this end, more detailed studies are warranted to provide more insights into the O2 polarization effect on the membrane stability. Yi et al. investigated the oxygen permeability and stability of the Sr0.95Co0.8Fe0.2O3%d membrane in an atmosphere containing CO2 and H2O in the feed side.65 Compared with applying CO2 or H2O alone in the feed side, the combined presence of CO2 and H2O provides more negative effects on the microstructure, phase composition, and oxygen permeation performance of the Sr0.95Co0.8Fe0.2O3%d membrane. This is identical to what was observed by Yan et al. on BSCF5582.58 Such phenomena were caused by the formation of bicarbonate on the membrane surface which is enabled by the existence of both CO2 and H2O. Membrane degradation additionally correlates with the applied oxygen partial pressure and temperatures. Higher oxygen partial pressure and lower temperature (i.e., 810 1C) enhanced the degradation effect. The bicarbonate formation was associated with entropy reduction which is favored at lower temperature. 2.1.3. Recent status and advances. In this section, recent studies on CO2-resistant perovskite membranes are summarized. For comparison purpose, the performances of some typical perovskite membranes are listed in Table 1.30,39,41,45–47,53,60,62,66–76 Alkaline-earth metals containing perovskite oxides such as (Ba,Sr)(Co,Fe)O3%d are among the best candidates for oxygen permeation membranes due to their high mixed ionic and electronic conductivity.15 However, they have poor stability under a CO2 atmosphere. BSCF5582 disks, for example, have high oxygen fluxes that normally can reach more than 1 mL min%1 cm%2 at 950 1C, or even reach to 12.2 mL min%1 cm%2 at 1000 1C for an ultrathin BSCF5582 membrane with a porous substrate.61,64,77–89 The low chemical stability of BSCF5582 nevertheless resulted in the formation of barium or strontium carbonates as the surface layer and oxygen flux deterioration, upon exposure to CO2.27–29,83 As little as 300 ppm of CO2, less than the content in air (380 ppm), was sufficient to damage the BSCF5582 membrane.90 The carbonate particles could form on the surface even after a short annealing time of 3 minutes at 800 1C in the CO2 atmosphere. With prolonged treatment time, the carbonate particles grew into a compact carbonate layer.29 Such degradation was also observed on the performance of solid oxide fuel cells (SOFCs) using BSCF5582 as the cathode component in a CO2 containing atmosphere.91–94 Strong CO2 adsorption on the BSCF5582 surface and the formed carbonate layer contributed to the decrease of the oxygen surface exchange reaction that


Review Article

impaired the cell performance. Small quantities of CO2 (i.e., 0.28–3.01 vol%) negatively affected the SOFC performance at 450 1C.91 To this end, several attempts have been performed to enhance the CO2 resistance of BSCF5582 membranes such as reducing the amount of alkaline earth elements and/or the amount of cobalt and incorporating fixed-valence metal elements in the B-site. As mentioned above, the stability of carbonates increases with increasing ionic radius of alkaline earth elements.41,95,96 The stability increases in the order of Ca (1.34 Å), Sr (1.44 Å), and Ba (1.61 Å). Although Ca-containing perovskites show better stability towards CO2 than Ba- and Sr-containing ones, carbonates still formed during exposure to a CO2 atmosphere.57,97 Partial substitution of alkaline-earth cations in the A-site with rare-earth cations has been adopted to enhance the CO2 resistance of alkaline-earth elements containing perovskite oxides. For example, partial substitution of Ba with rare-earth cations, i.e., La or Pr can substantially enhance the stability against CO2.53,60,75,98 Waindich et al. compared the CO2 corrosion behavior of BSCF5582, Ba0.3Sr0.7Co0.2Fe0.8O3%d (BSCF3728), and La0.3Ba0.7Co0.2Fe0.8O3%d (LBCF3728). They found that BSCF3728 and LBCF3728 exhibit better chemical stability against CO2 corrosion relative to BSCF5582.99 The enhancement of stability against CO2 can be understood from the thermodynamics perspective as discussed above. Using the Ellingham diagram, Partovi et al. predicted that the decomposition temperature is above 1100 1C for pure SrCO3 and even higher (above 1400 1C) for BaCO3 under atmospheric pressure (101.3 kPa) of CO2,98 while the rare-earth cation carbonates decompose at lower temperatures.15,41,42 Partial substitution of La in LaFeO3%d and LaCo0.8Fe0.2O3%d with Ca was performed to take advantage of the higher Ca and La stability in a CO2 atmosphere. (La1%xCax)FeO3%d and (La1%xCax)(Co0.8Fe0.2)O3%d compounds with x = 0.4–0.6 were synthesized where only x = 0.4 compounds showed almost pure perovskite phases.41 In general, the complete substitution of Ba in the A-site with rare-earth cations leads to substantially lower oxygen permeation flux (than that obtained in Ba-containing perovskites). For example, at 800 1C, a La0.58Sr0.4Co0.2Fe0.8O3%d membrane of 1 mm-thickness displayed an oxygen flux of only around 0.15 mL min%1 cm%2 or an even lower flux value for La0.6Sr0.4Co0.2Fe0.8O3%d (LSCF6428).100–102 The lower oxygen permeation flux was a consequence of two factors. Firstly, replacing low valence cations (i.e., Ba2+ or Sr2+) with higher valence cations (i.e., La3+) led to a lower oxygen non-stoichiometry and also a lower concentration of oxygen vacancies (relative to SrCo0.2Fe0.8O3%d or BSCF5582).38,98 Secondly, a reduced lattice parameter manifested from such replacement given the smaller ionic radius of La3+ (1.36 Å) compared with Sr2+ (1.44 Å).23,98 The CO2 stability of La0.6Sr0.4Co0.8Fe0.2O3%d (LSCF6482) nonetheless increases following such substitution (relative to SrCo0.2Fe0.8O3%d and BSCF5582).38 Besides the rare earth metal substitution strategy, the chemical and structural stability of perovskites can also be related to their A-site cation stoichiometry.65,103–107 Shao et al. observed that the oxygen permeation of BSCF5582 increased slightly, i.e., from 1.1 to 1.2 mL min%1 cm%2 during a long-term test for more than 1000 h at 850 1C.108 Such increment in permeation flux is

Chem. Soc. Rev.

Chem. Soc. Rev.

Solid-state Disk reaction


BaFe0.8Nb0.2O3%d


BaCo0.4Fe0.4Nb0.2O3%d



Hollow fiber

Synthesis route Geometry


(x + y + z = 1)

BaCoxFeyZrzO3%d

Materials

Area (cm2) Coating

900 0.510 (He) 0.097 (CO2) 900 0.400 (He) 0.260 (CO2)

900 1.090 (He) 0.000 (CO2) 900 0.700 (He) 0.070 (CO2)

J(O2) (mL T min%1 Feed (1C) cm%2) side

Long term CO2 test

800 0.46 (Ar) 0–0.05 (CO2) 900 0.7 (Ar) 0.05– 0.08 (CO2) 1000 0.92 (Ar) 0.92– 0.26 (CO2)

800 2.26– 0.125 (He + CO2) 850 2.61– 0.24 (He + CO2) 900 3.05– 0.375 (He + CO2)

800 2.3– 0.7 (He + CO2) 850 2.6– 1.7 (He + CO2)

Ar CO2 Ar CO2 Ar CO2

Air Air Air Air Air Air

0– 1.275

0

0– 1.24

0– 1.25 0

0

0–0.5 150 mL 50 mL min%1 air min%1 He (10%CO2)

150 mL 50 mL 0–0.5 min%1 air min%1 He (10%CO2)

0–0.5 150 mL 50 mL min%1 air min%1 He (10%CO2)

800 CO2 : He 5–10 mm Ba(50 : 50 enriched layer vol%) and 5–15 mm Ba-depleted layer On top of the original composition layer

0–0.5 EDX 150 mL 50 mL min%1 air min%1 He (3%CO2)

Remarks

800 CO2 : N2 BaCO3 (50 : 50 formation vol%)

T Atmo(1C) sphere

Other CO2 test

0–0.5 In situ 150 mL 50 mL powder min%1 air min%1 He XRD (3%CO2)

J (O2) (mL Permeate T Permeate Time min%1 side (1C) cm%2) Feed side side (h) Method

Instantaneous performance

30

30

30

30

39

Ref.

Review Article

1

1

1

1

0.14 (OD 1.1)

Thickness (mm)

Membrane architecture

Table 1 CO2-resistance and performance of perovskite membranes


View Article Online

Chem Soc Rev



Complexa- Disk tion

Complexa- Disk tion

Complexa- Disk tion


BaFe0.65Nb0.35O3%d

La0.6Ca0.4Co0.8Fe0.2O3%d

La0.6Ca0.4FeO3%d


SrCo0.8Fe0.2O3%d

SrCo0.8Fe0.2O3%d 1

1

0.5

1

0.85

Thickness (mm) Area (cm2) Coating


Materials

(continued)


Table 1

800– 0.08– 100 mL 950 0.41 min%1 (He) air 0.04– 0.24 (CO2)

900 0.12 (He) 0.106 (CO2)


Long term CO2 test

30 mL min%1 He 30 mL min%1 CO2

0.44 (CO2)

100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2

71

51

19–23

0–5

83

73

64

28–47

4–26 100 mL 50 mL min%1 air min%1 He

10– 110

Remarks

30

Ref.

45

46

66

30– CO2 : air No carbonate 41 1000 (50 : formation 50 vol%) Rhombohedral to cubic phase transition at 700 1C

T Atmo(1C) sphere

Other CO2 test

Chem Soc Rev

0.88 (CO2)

1.72 (CO2)

950 1.86 (CO2)

0 (CO2)

0.18 (CO2)

0.59 (CO2)

900 1.35– 1.5 (He) 1.06 (CO2)

150 mL 29 mL min%1 air min%1 CO2


900 0.27 (He) 0.155 (CO2)


850 0.09 (CO2)

110– In situ 185 powder XRD




View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

Complexa- Disk tion


Sr(Co0.8Fe0.2)0.95Ti0.05O3%d

Complexa- Disk tion

SrCo0.8Fe0.2O3%d

Sr(Co0.8Fe0.2)0.9Ta0.1O3%d

Complexa- Disk tion

SrCo0.8Fe0.2O3%d

Area (cm2) Coating


Long term CO2 test

1.17 (CO2)

950 1.61 (CO2)

900 0.98– 1.34 (He) 1.25– 1.22 (CO2) 1.16 (CO2)

1.76 (CO2 + O2)

1.64 (CO2 + O2)

1.28 (CO2 + O2)

950 1.94 (CO2 + O2)

0.04 (CO2)

0.07 (CO2)

0.25 (CO2)

0.84 (CO2)

950 1.78 (CO2)

100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2

100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 (5%O2) 100 mL 50 mL min%1 air min%1 CO2 (5%O2) 100 mL 50 mL min%1 air min%1 CO2 (5%O2) 100 mL 50 mL min%1 air min%1 CO2 (5%O2) 100 mL 50 mL min%1 air min%1 He

20–21

0–6

72– 103

57–61

5–56

90

50

8

0

25

15

10

3

0



T Atmo(1C) sphere

Remarks

Other CO2 test

45

46

47

47

Ref.

Review Article

1

1

1

1

Thickness (mm)


Materials

(continued)


Table 1


View Article Online

Chem Soc Rev




Solid-state Tube reaction


Solid-state Multireaction channel hollow fiber


Sr(Co0.8Fe0.2)0.9Ti0.1O3%d

Sr(Co0.8Fe0.2)0.8Ti0.2O3%d


Sr(Co0.8Fe0.2)0.9Ti0.1O3%d

SrFe0.8Nb0.2O3%d

SrFe0.8Nb0.2O3%d

SrFe0.8Zr0.2O3%d

Area (cm2) Coating

1

0.65

0.25 2.4 (ID0.71)

1

1.6 11 (ID9.4)

1

1

Thickness (mm)


Materials

(continued)


Table 1

750– 0.135– Ambient 100 mL 900 0.525 air min%1 He (He)

80 mL min%1 CO2

800– 0.52– 120 mL 80 mL 900 1.24 min%1 min%1 He (He) air 0.38– 1.12 (CO2)

Long term CO2 test

100 mL min%1 CO2 100 mL min%1 He

0.3– 0.27 (CO2) 0.3 (He)

850 0.35 (He)

1.06 (CO2)

80 mL min%1 CO2 120 mL 80 mL min%1 air min%1 (20%CO2) CO2 Ambient 100 mL air min%1 He

120 mL 80 mL min%1 air min%1 He



100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL min%1 air min%1 CO2 100 mL 50 mL %1 min air min%1 CO2

1.12 (CO2)

900 1.24 (He)

900 0.31 (CO2)

1.59 (CO2) 950 1.34 (CO2)

950 1.47 (CO2)

0.63 (CO2)

0.73 (CO2)

0.92 (CO2)

T Atmo(1C) sphere

13.3– 16.7

850 CO2

No carbonate formation

69

68

67

45

45

45

Ref.

Chem Soc Rev

3.3– 13.3

0–3.3 Treating at 850 1C

400– 600

75– 390

Very slight increase


Remarks

Other CO2 test

0–210 Annealing 900 CO2 in pure CO2 Isothermal 900 CO2 : O2 (99 : gravimetry 1vol%) 0–75

0–83

13–88

0–4

69

51

30


7 mL 900 0.44 3 bar (CO2) pressure min%1 CO2 0.84 43 mL (CO2) min%1 CO2




View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

(continued)

BaCo0.7Fe0.24Zr0.06O3%d

Complexa- Disk tion

1

1.3

1.3

Complexa- Disk tion


1

0.65

1


Area (cm2) Coating

SrFe0.8W0.2O3%d

Thickness (mm) 0.65



SrFe0.8Mo0.2O3%d

Materials

Table 1

Long term CO2 test

750– 1.36– 300 mL 100 mL 925 2.7 min%1 min%1 He (He) air

300 mL min%1 air 0.13– 0.03 (He + CO2)

2.7 (He)

Remarks

69

69

Ref.

10–20 300 mL 85 mL min%1 air min%1 He + 15 mL min%1 CO2 20–30 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 100 mL 30–40 min%1 air min%1 He

Review Article

0.25– 0.08 (He + CO2)

2.12– 0.85 (He + CO2)

2.65 (He) 925 2.7 (He)

300 mL min%1 air 300 mL min%1 air

300 mL min%1 air

1.8– 0.5 (He + CO2)

13.3– 16.7 0–10 TG-DSC

3.3– 13.3

T Atmo(1C) sphere

Other CO2 test

40– CO2 : N2 Effectively 70 1150 (10 : improved 90vol%) 85 mL 10–20 SEM of 8–12 mm corromin%1 He he spent sion layer on + 15 mL membrane top of the orimin%1 ginal composiCO2 tion layer 50 mL 20–30 %1 min He + 50 mL min%1 CO2 100 mL 30–40 min%1 He 100 mL 0–10 70 min%1 He

Ambient 100 mL air min%1 CO2 Ambient 100 mL air min%1 He 300 mL 100 mL min%1 air min%1 He

0.19 (He) 925 2.65 (He)

13.3– 16.7

Ambient 100 mL 0–3.3 air min%1 He

0.17 (CO2)

850 0.22 (He)

0.16 (He)

100 mL min%1 CO2 100 mL min%1 He

0.15 (CO2) 3.3– 13.3

100 mL 0–3.3 min%1 He

0.21 (He)



750– 0.149– Ambient 900 0.25 air (He)

750– 0.148– 900 0.27 (He)




View Article Online

Chem Soc Rev


Complexa- Disk tion

Complexa- Disk tion




1.6

1

1

1

1.3

1.3

1.3

Area (cm2) Coating 300 mL 100 mL min%1 min%1 air He


750– 950

Long term CO2 test

300 mL 85 mL min%1 air min%1 He + 15 mL min%1 CO2 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 100 mL min%1 air min%1 He 200 mL 50 mL min%1 air min%1 He

1.63– 0.87 (He + CO2)

2.2 (He) 800 1.375 (He)

0.24– 0.125 (He + CO2)

2.43 (He) 925 2.2 (He)

0.37– 0.19 (He + CO2)

300 mL 85 mL min%1 air min%1 He + 15 mL min%1 CO2 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 100 mL %1 min air min%1 He 300 mL 100 mL min%1 air min%1 He

0– 0.25

30–40

20–30

10–20

0–10

30–40

20–30

10–20

300 mL 100 mL 0–10 min%1 air min%1 He

10–20 300 mL 85 mL min%1 air min%1 He + 15 mL min%1 CO2 20–30 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 100 mL 30–40 %1 min air min%1 He


1.95– 1.15 (He + CO2)

925 2.43 (He)

2.53 (He)

0.31– 0.14 (He + CO2)

1.88– 1.05 (He + CO2)

925 2.53 (He)



750– 1.3– 925 2.53 (He)



T Atmo(1C) sphere

Remarks

Other CO2 test

71

70

70

70

Ref.

Chem Soc Rev

Ba0.5Sr0.5Solid-state Disk Co0.78Fe0.2W0.02O3%d reaction

Complexa- Disk tion


Thickness (mm)


Materials

(continued)


Table 1


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

Commercial

Commercial

La0.6Sr0.4Co0.2Fe0.8O3%d


0.8

Area (cm2) Coating

180 mL 60 mL min%1 min%1 air He

200 mL 50 mL min%1 min%1 air He

180 mL 90 mL min%1 air min%1 He + 10 mL min%1 CO2 180 mL 80 mL min%1 air min%1 He + 20 mL min%1 CO2 180 mL 60 mL min%1 air min%1 He + 40 mL min%1 CO2 180 mL 20 mL %1 min air min%1 He + 80 mL min%1 CO2

3.4– 3.3 (He + CO2)

0.5– 0.2 (He + CO2)

1.5– 0.9 (He + CO2)

3.1–2 (He + CO2)

950 3.57 (He)

0.18 (CO2)

200 mL 40 mL min%1 air min%1 He + 10 mL min%1 CO2 200 mL 50 mL min%1 air min%1 CO2 180 mL 100 mL min%1 air min%1 He

0.75 (He + CO2)

7.75– 9.4

6.7– 7.75

2.75– 6.7

1.3– 2.75

0–1.3

1.75– 2.0

0.25– 1.75

750– 0.45– 300 mL 300 mL 1000 4.32 min%1 min%1 Ar air (Ar)

TGA

CO2

Perovskite structure and its crystallinity were preserved 20– CO2–air No carbona60 1000 (95 : tion/dec5 vol%) arbonation process occurs

72

Ref.

XRD of spent membrane

Remarks

20– CO2–air 73 No detectable 1000 (95 : carbonation/ 5 vol%) decarbonation process occurs

T Atmo(1C) sphere

Other CO2 test

TGA


Long term CO2 test

600– 0.07– 300 mL 300 mL 900 2.5 (Ar 300 mL 150 mL 0– + CO2) min%1 air min%1 Ar 41.7 1000 6.8 min%1 min%1 Ar air + 150 mL (Ar) min%1 CO2

775– 1.5– 900 2.86 (He)

1.0– 2.5 (He)



Review Article

Asymmetric 0.03 disk

Asymmetric 0.03 disk


BaBi0.05Co0.8Ta0.15O3%d

Thickness (mm)


Materials

(continued)


Table 1


View Article Online

Chem Soc Rev


(continued)


Commercial

Disk


0.1

Thickness (mm)

4.95 (Ar)

4.9 (Ar + CO2)

5 (Ar + CO2)

5.2 (Ar + CO2)

1000 5.52 (CO2)

2.95 (Ar)

2.25 (Ar + CO2)

2.1 (Ar + CO2)

900 2.02 300 mL (CO2) min%1 air 2 (Ar + 300 mL CO2) min%1 air

300 mL 150 mL 0–90 min%1 air min%1 Ar + 150 mL min%1 CO2

XRD


Long term CO2 test

300 mL min%1 CO2 225 mL min%1 CO2 + 75 mL min%1 Ar 300 mL 150 mL min%1 min%1 air CO2 + 150 mL min%1 Ar 300 mL 75 mL min%1 min%1 air CO2 + 225 mL min%1 Ar 300 mL 300 mL min%1 min%1 Ar air 300 mL 300 mL min%1 min%1 air CO2 300 mL 225 mL min%1 min%1 air CO2 + 75 mL min%1 Ar 300 mL 150 mL min%1 min%1 air CO2 + 150 mL min%1 Ar 300 mL 75 mL min%1 min%1 air CO2 + 225 mL min%1 Ar 300 mL 300 mL min%1 min%1 Ar air 300 mL 300 mL 1000 1.41– 1.25 min%1 min%1 Ar air (Ar + CO2)



LSCF6428 800– 1.9– 1000 7.0 (Ar)

Area (cm2) Coating


750 CO2

T Atmo(1C) sphere

Absence of carbonates or single oxides

Remarks

Other CO2 test

62

Ref.

Chem Soc Rev


Materials

Table 1


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

0.25 (ID0.77)

Multichannel hollow fiber Solid-state Disk reaction

Multichannel hollow fiber Complexa- Disk tion

SrFe0.8Sb0.2O3%d

Pr0.6Sr0.4Co0.2Fe0.8O3%d

0.6

0.21 (ID0.87)

1

1


SrFe0.9Ta0.1O3%d

Thickness (mm)

0.6

Area (cm2) Coating



(continued)

Materials

Table 1

750– 0.1– 1000 1.3 (He) 0.08– 1.24 (CO2) 150 mL min%1 air 150 mL min%1 air

750– 0.12– 120 mL 950 0.47 min%1 (He) air 0.09– 120 mL 0.32 min%1 (CO2) air 0.3– 120 mL min%1 1.3 (CO2) air

750– 0.17– 120 mL 950 0.63 min%1 (He) air 0.12– 120 mL 0.38 min%1 (CO2) air 0.41– 120 mL 1.51 min%1 (CO2) air

Long term CO2 test


60 mL min%1 He 60 mL min%1 CO2 60 mL min%1 CO2

60 mL min%1 He 60 mL min%1 CO2 60 mL min%1 CO2

340– 450

215– 340

175– 215

CO2

T Atmo(1C) sphere

No carbonate formed

Remarks

Other CO2 test

75

74

74

Ref.

Review Article

850 0.42 (CO2)

1000 1.24 (CO2)

850 0.42 (CO2)

125– 175

150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL %1 min air min%1 CO2 1000 1.24 (CO2)

0–130

0–130 XRD of spent membrane




1000 1.3 (He)

900 0.22 (CO2)

900 0.3 (CO2)


0.15– 300 mL 300 mL min%1 min%1 7.2 CO2 (CO2) air




View Article Online

Chem Soc Rev


Complexa- Hollow tion fiber

Complexa- Disk tion

Complexa- Disk tion


Sm0.6Ca0.4CoO3%d


Sm0.6Ca0.4FeO3%d

0.5

0.5

0.64 (ID1.4)

Thickness (mm)

1

1

Sm0.6Ca0.4FeO3%d

Sm0.6Ca0.4CoO3%d

Area (cm2) Coating



(continued)

Materials

Table 1

800– 0.1– 1000 3.15 (He) 0.01– 1.2 (CO2) 200 mL min%1 air 200 mL min%1 air


Long term CO2 test


0.017 (CO2)

0.02 (He)

0.01 (CO2)

950 0.02 (He)

0.78 (CO2)

900 0.2 (CO2)

0.78 (CO2)

2.44 (He)

0.79 (CO2)

950 2.6 (He)

200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 100 mL min%1 air 100 mL min%1 air 100 mL min%1 air 100 mL min%1 air 30 mL min%1 CO2

100– 200

30 mL 100– min%1 200 CO2 0–100 30 mL min%1 He

151 mL 0–7.8 SEM of the spent min%1 He membrane 151 mL 7.8– %1 min 31 CO2 151 mL 31–40 min%1 CO2 151 mL 40–59 min%1 CO2 151 mL 59–79 min%1 CO2 151 mL 79– min%1 100 CO2 0–100 30 mL min%1 He



T Atmo(1C) sphere

Ref.

76

76

1.4 mm erosion 53 layer

Remarks

Other CO2 test


View Article Online

Chem Soc Rev Review Article

Chem. Soc. Rev.

View Article Online


Review Article

associated with the A-site cation diffusion.109 The resultant A-site cation deficiency caused by diffusion was charge compensated by the creation of additional oxygen vacancies which increased oxygen permeation flux. For Sr1%yFe1%xTixO3%d compounds, creation of A-site deficiency suppressed the oxygen-vacancy ordering and increased the structure disorder, leading to the stability of the cubic perovskite structure. In the case of A-site stoichiometric SrFe0.8Ti0.2O3%d, exposure to CO2 for 40 hours at 200 1C led to the formation of SrCO3 as revealed by the infrared (IR) absorption spectroscopy results. SrCO3, on the other hand, did not form in the case of A-site deficient Sr1%yFe0.8Ti0.2O3%d ( y = 0.03–0.06). This observation indicates that structure and oxygen permeation flux stability may, in specific cases, be enhanced via the creation of A-site deficiency.106 Another representative example is A-site deficient Sr0.95Nb0.1Co0.9O3%d in the application context of cathodes for SOFCs. After introducing CO2containing air, the increase in the area specific resistance (ASR) of the Sr0.95Nb0.1Co0.9O3%d cathode was less than that of the SrNb0.1Co0.9O3%d cathode. The lower oxygen reduction reaction performance degradation for the former compound indicates that the A-site deficiency minimized CO2 poisoning on the surface of the cathode.110 In addition, for the dual-phase (composite) membrane of SrFe(Al)O3%d–SrAl2O4, partial dissolution of SrAl2O4 in the SrFe(Al)O3%d perovskite phase generated A-site deficiency in this phase. The resultant strontium deficiency contributed towards the improved composite stability against reaction with CO2.111 The mechanism behind the enhanced CO2 resistance from A-site deficiency is still elusive. Generally speaking, the additional oxygen vacancies generated from A-site deficiency is likely to increase the oxygen permeation flux and the structure stability. Such higher concentration of oxygen vacancies tends to promote CO2 surface adsorption and eventually, the formation of carbonates. From the perspective of the Lewis acid–base concept, partial substitution of B-site elements with fixed valence cations serves as an alternative option to improve the stability of perovskite membranes. The work of Czuprat et al. was among the earliest studies to report the effect of exposing perovskite to a CO2 atmosphere.39 For the BaCoxFeyZrzO3%d compound, exposure to 50 vol% CO2 in He for 5–10 hours led to the decomposition of the perovskite structure up to a depth of 25 mm. Scanning electron microscopy indicates the formation of an up to 10 mm BaCO3-enriched layer followed by an up to 15 mm Ba-depleted zone on top of the original perovskite layer. They also showed that the original microstructure and oxygen fluxes can be recovered in a CO2-free atmosphere (i.e., He). Similar results were obtained on the BaCo0.7Fe0.3%xZrxO3%d membrane of which a corrosion layer with an average thickness of 12 mm for x = 0.04 and 8 mm for x = 0.10 appeared.70 CO2 resistance is enhanced by increasing the amount of Zr dopant although with inevitable reduction of oxygen permeation flux. Doping with other high valence metal cations such as Ta, Nb, and Ce gives the identical effect towards the oxygen permeation, methane partial oxidation, and chemical stability behaviors of Ba fullyoccupied A-site BaCoFeO3%d membranes.30,112–118 Nb, in particular, is an effective dopant to enhance the stability of perovskite while

Chem. Soc. Rev.

Chem Soc Rev

maintaining high oxygen permeability.30,119 Yi et al. studied the oxygen permeation behavior and the degradation mechanism of Ba(Co,Fe,Nb)O3%d perovskite in the CO2 atmosphere.30 Partial substitution of Fe and Co by Nb leads to the higher CO2 resistance (relative to the original compound) and the stability enhancement mechanisms are explored using the Lewis acid–base perspective as discussed above. Upon relating this observation with the acidity trend reported by Jeong et al. (in increasing acidity order), i.e., Fe2+, Co2+, Fe3+, Co3+, Co4+, and Nb5+; such acidity increase appears to correlate with the increasing oxidation state.48 Increasing content of Nb5+ or Fe3+ contributes towards higher acidity relative to the case of B-site cations fully occupied by Co2+. The high valence metal doping approach resulted in several other compounds such as BaCo0.7Fe0.2Sn0.1O3%d, BaCo0.7Fe0.3%xInxO3%d, BaTi0.2Co0.5Fe0.3O3%d, and BaBixCo0.2Fe0.8%xO3%d which were tested as oxygen permeation membranes but without evaluating their stability in the presence of CO2.120–123 Despite the stability enhancement obtained via such doping, Ba-containing perovskites nonetheless show performance deterioration over time in a CO2-containing atmosphere as observed in the case of tantalum and bismuth co-doped perovskite BaBi0.05Co0.8Ta0.15O3%d.72 BaBi0.05Co0.8Ta0.15O3%d indeed demonstrated higher stability than BSCF5582 but the oxygen flux still declined under long-term operation with a CO2-containing atmosphere as a sweep gas. With the exception of the BaCo1%xFexO3%d system, other compositions of B-site doping have mostly been performed. By substituting as low as 2 mole% of tungsten into BSCF5582, the phase transition and O2 permeation performance degradation in the CO2 atmosphere was effectively suppressed on such Ba0.5Sr0.5Co0.78Fe0.2W0.02O3%d membranes,71 which exhibited higher stability under a sweep gas containing 20 vol% CO2 within a 1.5 hour-test period and 100 vol% CO2 within a 0.25 hour-test period, compared with BSCF5582 without tungsten doping. Chen and co-workers developed Ti-doped SrCo0.8Fe0.2O3%d, i.e., Sr(Co0.8Fe0.2)1%xTixO3%d (x = 0.05, 0.1, 0.2, and 0.4) to enhance the CO2 resistance of SrCo0.8Fe0.2O3%d.45 Upon exposure to CO2, the oxygen permeation flux of SrCo0.8Fe0.2O3%d was reduced to only 30% of the initial flux after a 70 hour-period. For Sr(Co0.8Fe0.2)0.95Ti0.05O3%d after the same period of exposure, up to 43% of the initial flux was retained. In the case of Sr(Co0.8Fe0.2)0.9Ti0.1O3%d and Sr(Co0.8Fe0.2)0.8Ti0.2O3%d, on the other hand, CO2 exposure did not affect the oxygen flux. Ta-doped SrCo0.8Fe0.2O3%d, i.e., Sr(Co0.8Fe0.2)0.9Ta0.1O3%d was also demonstrated to be stable to CO2 where only a 5.3% (from 1.32 mL min%1 cm%2 to 1.25 mL min%1 cm%2) drop in oxygen flux was observed on switching the sweep gas from helium to CO2. The flux stabilized at 1.14 mL min%1 cm%2 after 50 hours of exposure to CO2.46 Although partial substitution of B-site cations with more acidic (less basic) metal cations can enhance CO2 resistance, the formation of an alkaline-earth carbonate layer is unavoidable after prolonged exposure to pure CO2. A Ta-doped SrFeO3%d membrane, for example, in spite of its higher CO2 resistance than non-doped SrFeO3%d, i.e., capability to retain its oxygen permeation flux for at least 71 hours under 10 vol% CO2 atmosphere, showed 33.9% decay in oxygen flux within 26 hours when pure CO2 was used as the sweep gas.124 Zhu et al. evaluated CO2 reaction mechanisms on SrCo0.9%xFexNb0.1O3%d


View Article Online


Chem Soc Rev

(SNCFx) cathodes.50 No carbonate phase was detected on the powder X-ray diffraction patterns for all perovskite compositions after one hour of CO2 exposure at 600 1C. Fourier Transform Infra-Red spectra nonetheless displayed the presence of carbonate functional groups on SNCF0.1, SNCF0.2, and SNCF0.3 which were absent on SNC, SNCF0.5, and SNF. In the cases of SNCF0.1, SNCF0.2, and SNCF0.3, Zhu et al. hypothesized the simultaneous occurrence of CO2 adsorption and strontium carbonate formation whereas for the other compositions, only CO2 adsorption occurred. These distinct behaviors appeared to correlate with the oxygen non-stoichiometries of the perovskites. Reducing Co content in the B-site of the perovskite can enhance CO2 resistance despite the resultant reduction in O2 permeability as discussed in detail by Partovi et al.75 The oxygen permeation properties and the chemical stabilities of cobalt-free SrFe0.8M0.2O3%d perovskites (SrFe0.8Zr0.2O3%d, SrFe0.8Mo0.2O3%d, and SrFe0.8W0.2O3%d) were probed under a CO2-containing atmosphere.69 Powder X-ray diffraction measurements combined with Rietveld analyses showed that the dominant phase for CO2exposed SrFe0.8M0.2O3%d is cubic perovskite. A very small peak emerged near the (110) plane peak that can be attributed to the pyrochlore phase or single oxide phase. When 10 vol% CO2 was introduced into the sweep gas, SrFe0.8Mo0.2O3%d and SrFe0.8W0.2O3%d displayed stable oxygen fluxes (0.15 and 0.17 mL min%1 cm%2, respectively) for 10 hours whereas the oxygen flux of SrFe0.8Zr0.2O3%d gradually decreased. The latter deterioration was related to the presence of Zr3+ in CO2-exposed SrFe0.8Zr0.2O3%d as evidenced by X-ray photoelectron spectroscopy results.69 The stabilizing effect of divalent zinc-doped (Ba,Sr)FeO3%d (BSFZ) was investigated under a CO2 atmosphere.37 Even when cobalt was totally substituted by zinc, the O2 flux of BSFZ still decreased to almost zero when pure CO2 was used as the sweep gas. Martynczuk et al. found that the formation of an 8% carbonate phase on the BSFZ surface was sufficient to induce oxygen permeability degradation at 750 1C.37 Powder X-ray diffraction, scanning electron microscopy, and transmission electron microscopy results evidenced the co-existence of cubic, hexagonal, and tetragonal perovskite phases as well as a carbonate layer at the surface after CO2 treatment. The tetragonal and cubic phases were found to be favorable for oxygen permeation while the hexagonal and carbonate phases were not. Furthermore, they also showed that the original O2 permeability can be partially recovered by sintering CO2-exposed membranes at high temperature, i.e., 950 1C under helium for a short time. Yi et al. discussed the strategy to choose metal elements in CO2-resistant perovskite ceramics.67 In terms of A-site cation candidates from alkaline earth metals, the presence of a large ionic radius cation such as Ba2+ usually leads to high oxygen permeability but has a tendency to form carbonate with CO2. Incorporating Ca2+, on the other hand, would provide CO2 tolerance although with the anticipated decrease of oxygen flux. Sr2+ that is located between those two metals in the same column in the periodic table is expected to balance the oxygen permeability and CO2 stability. For the B-site cation, Co and Fe have been extensively used while the former has poor stability under CO2 and the latter shows better stability and is thus preferred. The inclusion of a low


Review Article

amount of metal with high valence such as Nb5+ is beneficial to stabilize the perovskite framework and improve its resistance to CO2. These considerations lead to the development of Sr(Fe,Nb)O3%d perovskite membranes possessing good oxygen permeability and CO2 tolerance. Zhang et al. evaluated the doping effect on the phase structure, chemical stability, sintering behavior, conductivity, and oxygen permeability of SrFe0.8M0.2O3%d (M = Ti4+, Nb5+, and Cr6+) perovskites.125 They found that the dopants with different oxidation states affect the oxidation state of Fe ions in the perovskite, i.e., a higher oxidation state for the dopants resulted in a lower average oxidation state of Fe ions and also a lower concentration of oxygen vacancies. The CO2 resistances of these compounds were correlated with the average metal–oxygen bond energy (ABE) in the perovskite where higher ABE values increase CO2 resistance. Based on these aforementioned studies, we can delineate several major approaches that have been undertaken to enhance the stability of perovskite-based membranes to CO2, i.e., (1) adopting a rare earth metal cation as the A-site cation; (2) reduction in the Co content in the B-site; and (3) partial substitution of the original transition metal components in the B-site with more acidic (less basic) or more stable transition metal components. In doing these approaches, the Goldschmidt tolerance factor consideration (eqn (4)) becomes crucial to ensure that a suitable perovskite structure (i.e., cubic) with sufficiently high oxygen permeability can be obtained. Still, these approaches inevitably involve trade-off between the CO2 resistance and O2 permeability. The other option that does not induce such a trade-off represents the engineering approach that can be attained via optimizing the membrane configuration (i.e., area to volume ratio) and/or reducing the effective membrane transport thickness. A more practical membrane configuration such as hollow fibers with a thinner wall thickness can improve the oxygen permeation fluxes. To increase the oxygen permeation fluxes achieved by the SrFe0.8Nb0.2O3%d (SFN) disk-shaped membrane, Zhu et al. developed SFN hollow fibers that demonstrated up to 1.12 mL min%1 cm%2 oxygen flux under an air to CO2 gradient and stable operation for more than 500 hours.68 Likewise, Zhu et al. obtained approximately quadruple O2 flux enhancement for SrFe0.9Ta0.1O3%d and SrFe0.8Sb0.2O3%d at 900 1C via modification of the membrane geometries from a disk to hollow fiber form.74 Tan’s group contributed numerous studies on the morphological evolution and the associated oxygen permeation properties of LSCF6428 hollow fibers.53,126–131 They also studied the influence of CO2 on the oxygen permeation behavior of the LSCF6428 hollow fibers.53 CO2 chemical adsorption on the LSCF6428 surface was the main poisoning effect of CO2 in the short term. The adsorbed CO2 occupied the oxygen vacancy sites, thus suppressing the oxygen flux of the fiber. After operation under CO2 for more than 100 hours at 900 1C, the carbonates were gradually formed due to the reaction of Sr with the adsorbed CO2 causing Sr segregation on the surface. However, such CO2-induced corrosion on the surface did not lead to the membrane collapse given that the eroded surface layer was only 1.4 mm after long-term operation. Such thickness is substantially thinner than the 50 mm-thick erosion depth observed

Chem. Soc. Rev.

View Article Online


Review Article

on the BSCF5582 membrane exposed to pure CO2 at 875 1C for only 70 hours.27 The oxygen permeation flux of the fiber remained stable at 0.78 mL min%1 cm%2 at 950 1C after exposure to pure CO2 sweep gas for over 100 hours. In addition to the hollow fiber membranes, other configurations have also been developed. Gaudillere et al. fabricated a hierarchicallystructured LSCF6428 membrane by freeze-casting with a maximum O2 flux of 6.8 mL min%1 cm%2 at 1000 1C.73 Upon introducing 50 vol% CO2 into Ar sweep gas, the O2 flux at 900 1C decreased quickly from 4.6 to 2.5 mL min%1 cm%2. The O2 flux nonetheless remained stable at around 2.5 mL min%1 cm%2 during the 48 hourtest. The results indicated that the hierarchically-structured LSCF6428 membranes were robust and stable, at least identical in this aspect to the monolithic LSCF6428 membranes. An asymmetric LSCF6428 membrane consisting of a porous LSCF6428 support, a homogeneous dense thin LSCF6428 layer and a porous LSCF6428 activation layer was fabricated using a tape casting method.60 The oxygen permeation test was carried out under CO2-containing sweep gas at 900 and 1000 1C. At 900 1C, the O2 flux dropped 32% from 2.98 to 1.94 mL min%1 cm%2 when Ar was completely replaced by CO2, which was attributed to the competitive adsorption of CO2 and O2 species on the active sites of membrane surface. However, the test result at 1000 1C showed the opposite counterintuitive trend, displaying an increase in O2 flux from 4.95 to 5.52 mL min%1 cm%2 when the CO2 content was increased from 0 to 100%. Gaudillere et al. also fabricated a highly porous LSCF6428 freeze-cast asymmetric membrane with a 30 mmthick LSCF6428 porous activation layer over the dense top layer.62 The porous activation layer provided protective features against CO2, as evidenced by the lower O2 flux deterioration for the activated membrane with respect to the bare membrane. The highest obtained O2 flux of 7.2 mL min%1 cm%2 was observed at 1000 1C using 100% CO2 as the sweep gas. This value was even higher than that obtained from using Ar as a sweep gas (7.0 mL min%1 cm%2), which was consistent with the results of Serra et al.60 The long-term test in the CO2-rich atmosphere (50% CO2 in Ar) over 92 hours led to a degradation rate of only 4.17 ( 10%2 mL min%1 cm%2 per day (from 1.41 to 1.25 mL min%1 cm%2). The above studies from Serra’s group have shown that the O2 flux decreased with increasing CO2 content in the sweep gas at temperatures lower than 900 1C.60,62,73 It is quite counterintuitive that at high temperature (1000 1C), the O2 flux actually increased with increasing CO2 concentration where the membrane showed higher performance in CO2- rather than in an inert gas (Ar)-containing sweep gas. These phenomena were also observed in dual-phase membranes which will be introduced later.132,133 Two most plausible reasons were postulated, i.e., (1) the CO2 adsorption rate at 1000 1C was lower compared to that at lower temperatures; and (2) the lowered oxygen partial pressure in the permeate side from using CO2 sweep gas (relative to Ar sweep gas), or in other words, higher sweeping capacity for CO2 (relative to Ar). In the previous study on asymmetric LSCF6428,62 from the activation energy perspective, the Arrhenius analysis showed that the increase of CO2 concentration induced a progressive increase in the activation energy (Ea). The change in Ea revealed the change in the rate

Chem. Soc. Rev.

Chem Soc Rev

limiting step. The Ea values increased with the increase of CO2 contents (from 0 to 100%); indicating that the surface exchange rate became the rate limiting step at higher CO2 content.134,135 In another study on dual-phase membranes, the enhanced O2 flux when CO2 was used as sweep gas at 1000 1C was attributed to the better fluid dynamics and thermal emissivity of CO2 gas.133 It is noteworthy that the membranes with higher O2 permeation performance when exposed to pure CO2 generally display any or a combination of the three features, i.e., (1) featured an activated surface layer and/or a porous support; (2) operated at 1000 1C; and (3) used 100% CO2 as the sweep gas. Under such conditions, the fluid dynamics and gas emissivity may be different. More detailed future studies in this context are required. 2.2.

Dual-phase materials

The dual-phase membrane comprises an ionic conducting phase and an electronic conducting phase. This concept is displayed in Fig. 8(a) and (b) for two different composites, i.e., ionic conductor–electronic conductor and ionic conductor– mixed ionic–electronic conductor, respectively. More details were discussed by Zhu and Yang on the critical factors required for appropriate selection of each phases.136 Here, we have summarized the results from recent publications on dualphase membranes in Tables 2a and b66,76,132,133,137–162 and discussed the current research status on this topic, but placing more efforts on the strategies to battle the influence of CO2 erosion at high temperatures. The following definitions of terms are used throughout this section. A composite or dual phase membrane consisting of 45% A and 55% B by weight is

Fig. 8 Dual-phase membrane concepts: (a) ionic conductor–electronic conductor (IC–EC) dual-phase membrane; (b) ionic conductor–mixed ionic–electronic conductor (MIEC) dual-phase membrane. Reproduced with permission from ref. 66. Copyright 2015, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.



Disk

Disk

Ce0.8Gd0.2O2%d– Powder mixing PrBaCo0.5Fe1.5O5+d (60 : 40 wt%)

Disk

Powder mixing

Ce0.9Gd0.1O2%d– Ba0.5Sr0.5Co0.8 Fe0.2O3%d (60 : 40 wt%)

Powder mixing

1

0.5

0.5

0.5

ThickSynthesis ness Area route Geometry (mm) (cm2) Coating


0.38 (CO2)

150 mL 40 mL min%1 air min%1 CO2 150 mL 60 mL min%1 air min%1 CO2 150 mL 80 mL min%1 air min%1 CO2 150 mL 100 mL min%1 air min%1 CO2 300 mL 20 mL %1 min air min%1 CO2 300 mL 40 mL min%1 air min%1 CO2 300 mL 60 mL min%1 air min%1 CO2 300 mL 80 mL %1 min air min%1 CO2


100 mL 30 mL min%1 air min%1 CO2 150 mL 40 mL min%1 air min%1 CO2

925 0.40– 0.38 (He + CO2) 0.39 (He + CO2) 0.38– 0.37 (CO2) 0.36 (CO2)

0.67 (CO2)

950 0.59 (CO2)

100 mL 30 mL 900 0.47 min%1 air min%1 (CO2) CO2 100 mL 30 mL min%1 air min%1 He

800 0.15 100 mL 30 mL (CO2) min%1 air min%1 He

0–1

150– 250

0–72

50–100

0–50

300 mL 100% CO2 14–25 min%1 air

300 mL 100% CO2 12.5–14 min%1 air

300 mL 50% CO2 + 1–12 min%1 air 50% He

150 mL 40 mL min%1 air min%1 CO2 150 mL 40 mL min%1 air min%1 CO2 300 mL 50% CO2 + %1 min air 50% He

100 mL 30 mL min%1 air min%1 CO2 100 mL 30 mL min%1 air min%1 CO2

Method

T Atmo(1C) sphere

Other CO2 test

Remarks

140

137

66

Ref.

Chem Soc Rev

0.37 (CO2)

0.35 (CO2)

0.33 (CO2)

0.38 (CO2)

0.36 (CO2)

0.35 (CO2)

0.33 (CO2)

925 0.3 (CO2)

800– 0.25– 950 0.87 (He) 0.15– 0.7 (CO2) 0.11– 0.52 (He) 0.06– 0.46 (CO2) 825– 0.20– 975 0.85 (CO2)

J(O2) (mL Permeate Time min%1 cm%2) Feed side side (h)

Long term CO2 test

J(O2) (mL T Permeate T min%1 (1C) cm%2) Feed side side (1C)


(a) CO2-resistance and performance of dual-phase membranes (fluorite–perovskite). (b) CO2-resistance and performance of dual-phase membranes (fluorite and other oxides)

(a) Ce0.85Gd0.1Cu0.05- One-pot Disk O2%d–La0.6Ca0.4- (EC) FeO3%d (75 : 25 wt%)

Materials

Table 2


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.


0.5

0.5

Ce0.8Sm0.2O1.9– One-pot Disk Sm0.8Ca0.2Mn0.5- (SSR) Co0.5O3 (75 : 25 vol%)

Ce0.8Sm0.2O1.9– One-pot Disk SmMn0.5Co0.5O3 (SSR) (75 : 25 vol%)

0.6

One-pot Disk (EC)

0.8

0.8

0.11– 1.01 (CO2)

800– 0.24– 1000 1.08 (He)

0.4 (CO2)

SDC-SMC 800– 0.15– 940 0.52 (He) 0.13– 0.48 (CO2)

0.1– 0.51 (CO2)

850 0.2 (CO2)

100 mL 30 mL min%1 air min%1 CO2 100 mL 30 mL %1 min air min%1 CO2 100 mL 30 mL %1 min air min%1 He

0.41 (CO2)

220– 475

165– 220

155– 165

T Atmo(1C) sphere

TG

850 CO2

850 CO2

In situ pow- 30– CO2 : air der XRD 1000 (50 : 50 vol%)

Method

Other CO2 test

142

Ref.

CO2-stable 144

CO2-stable 143


Remarks

Review Article

850 0.21 (CO2)

100 mL 30 mL min%1 air min%1 CO2 100 mL 30 mL %1 min air min%1 CO2 100 mL 30 mL min%1 air min%1 CO2


0–45

220– 475

155– 220


0–10

0–300

450– 510

350– 450

0.46 (He)

100 mL 100 mL 940 0.36– min%1 air min%1 He 0.42 (He) 0.42 100 mL 100 mL (He) min%1 air min%1 CO2 0.37– 0.39 (CO2) 0.39 (CO2)


800 0.19– 0.13 (CO2) 940 0.39– 150 mL 100 mL 0.46 min%1 air min%1 He (He)

950 0.7 (CO2)

1000 1.04 (CO2)

60–350

150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL min%1 air min%1 CO2 150 mL 50 mL %1 min air min%1 CO2 100 mL 30 mL min%1 air min%1 He

150 mL 50 mL min%1 air min%1 CO2 950 0.7 (CO2)


950 0.7– 150 mL 50 mL 0.84 min%1 air min%1 He (He)



Long term CO2 test



SDC-SMC 800– 0.13– 940 0.61 (He)


(continued)

Ce0.9Pr0.1O2%d– Pr0.6Sr0.4Fe0.5Co0.5O3%d (60 : 40 wt%)

Materials

Table 2


View Article Online

Chem Soc Rev




Y0.8Ca0.2Cr0.8Co0.2O3–Sm0.2Ce0.8O1.9 (50 : 50 wt%)

One-pot Disk (GN)

1.3

1

Ce0.8Sm0.2O2%d– Powder Ba0.95La0.05Fe0.85- mixing Zr0.15O3%d (60 : 40 wt%)

Disk

0.6

30 mL 30 mL min%1 air min%1 (1.5% H2 + 48.5% N2 + 50% CO2) 30 mL 30 mL min%1 air min%1 N2

300 mL 85 mL min%1 air min%1 He + 15 mL min%1 CO2 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 300 mL 100 mL %1 min air min%1 He 300 mL 100 mL min%1 air min%1 He 300 mL 100 mL min%1 air min%1 CO2 30 mL 30 mL min%1 air min%1 (1.5% H2 + 48.5% N2 + 50% CO2) 0–350

18–105

0–18

52–65

39–52

26–39

12–26


150 mL 49 mL 0–150 min%1 air min%1 CO2 + 1 mL min%1 Ne

T Atmo(1C) sphere

XRD of the 950 H2 : N2 : spent CO2 (1.5 : membrane 48.5 : 50 vol%)

XRD of the 925 He : CO2 spent (50 : 50 membrane vol%)

In situ pow- 30– CO2 : N2 der XRD 1000 (50 : 50 vol%)

Method

Other CO2 test

Ref.

No indication of reaction

No cabonate observed

147

146

Thermally 145 and chemically stable

Remarks

Chem Soc Rev

0.05– 0.34 (N2)

0.37– 3.08 (H2 + N2 + CO2)

30 mL 700– 0.58– 30 mL 950 3.08 950 6.7 (H2 min%1 air min%1 (CO2) + N2) (3% H2 + 97% N2)

0.3 (He) 925 0.3 (He) 0.24 (CO2)

0.24 (He + CO2)

0.255 (He + CO2)

0.275 (He + CO2)

925 0.3 750– 0.1–0.3 300 mL 100 mL (He) 925 (He) min%1 air min%1 He

150 mL 49 mL 950 0.48 min%1 air min%1 (CO2) CO2 + 1 mL min%1 Ne


Long term CO2 test



La0.6Sr0.4- 850– 0.21– CoO3%d 950 0.48 (CO2)


(continued)

Ce0.9Nd0.1O2%d– One-pot Disk Nd0.6Sr0.4FeO3%d (EC) (60 : 40 wt%)

Materials

Table 2


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

Disk

Disk

One-pot Disk (GN)

Ce0.9Gd0.1O2%d– Powder La0.7Sr0.3MnO3)d mixing (80 : 20 vol%)

Ce0.9Pr0.1O2%d– Pr0.6Sr0.4FeO3%d (60 : 40 wt%)


0.06

0.6

1

0.785

850– 0.1– 950 0.26 (He)

0.27 (He)

0.195 (CO2)

0.21 (He + CO2)

0.23 (He + CO2)

925 0.275 (He)

950 0.18 150 mL 29 mL (CO2) min%1 air min%1 He + 1 mL min%1 Ne

0.06– 0.17 (CO2)

4–88 Synthetic 200 mL air min%1 He + 200 mL min%1 CO2 Synthetic 400 mL 88–150 air min%1 CO2

0–4

0–100

140– 163

Ref.



XRD of the 925 CO2 spent membrane

151

150

No weight- 148 increase peak No carbonate formation

100– CO2 : N2 1100 (80 : 20 vol%)

Remarks

XRD after 900 CO2 exposure to CO2

TG-DSC

T Atmo(1C) sphere

0–137.5 In situ pow- 30– CO2 : N2 der XRD 1000 (50 : 50 vol%)

5–95

0–5

Method

Other CO2 test

Review Article

0.43– 0.3 (CO2)

1.21– 0.5 (He + CO2)

150 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 29 mL %1 min air min%1 CO2 + 1 mL min%1 Ne 150 mL 7.32 mL min%1 air min%1 CH4 + 1 mL min%1 Ne Synthetic 400 mL air min%1 He


300 mL 15 mL min%1 air min%1 CO2 + 85 mL min%1 He 300 mL 50 mL min%1 air min%1 CO2 + 50 mL min%1 He 300 mL 100 mL min%1 air min%1 CO2 300 mL 100 mL min%1 air min%1 He 0.19 (CO2)


925 0.20– 300 mL 100 mL 0.19 min%1 air min%1 He (CO2)


Long term CO2 test



150 mL 29 mL 1000 0.28 min%1 air min%1 (CO2) CO2 + 1 mL min%1 Ne 2.2–4.4 150 mL 7.32 mL 950 4.4 (CH4) (CH4) min%1 air min%1 CH4 + 1 %1 mL min Ne La0.6Sr0.4- 700– 0.4–1.7 Synthetic 400 mL 800 1.7 (He) CoO3%d 850 (He) air min%1 He


(continued)

Ce0.8Gd0.2O2%d– Powder Ba0.95La0.05Fe0.9- mixing Nb0.1O3%d (60 : 40 wt%)

Materials

Table 2


View Article Online

Chem Soc Rev




Disk

One-pot Disk (EC)

Ce0.8Sm0.2O2%d– Sm0.6Ca0.4CoO3 (75 : 25 wt%)

0.8

SDCLSCF

800– 1000

875– 950

0.024– 0.104 (CO2) 0.323– 0.780 (He) 0.251– 0.659 (CO2) 0.045– 0.329 (He)

800 0.65 (He) 0.32– 0.28 (CO2) 800 0.48 (He) 0.44 (CO2)

950 0.13– 0.18 (He) 0.19 (CO2)

950 0.23– 0.16 (He) 0.16 (CO2)

60 mL 950 0.49– min%1 He 0.53 (CO2)

CO2

He

CO2

He

0.15 (CO2)

100 mL 30 mL min%1 air min%1 CO2 900 0.15 Ambient 40 mL (He) air min%1 He

100 mL 30 mL min%1 air min%1 CO2 100 mL 30 mL %1 min air min%1 He

875– 0.025– 100 mL 30 mL 950 0.104 min%1 air min%1 He (He)

SDC-SCC 800– 0.0675– 100 mL 950 0.167 min%1 air (He) 0.055– 100 mL 0.168 min%1 air (CO2) 1 SDC-SCC 800– 0.0825– 100 mL 950 0.18 min%1 air (He) 0.075– 100 mL 0.188 min%1 air (CO2) 0.44 800– 0.51– 120 mL 950 1.54 min%1 air (He)

1

0.78

STF-GDC

Sr0.97Ti0.5Fe0.5O3%d

7–100

0–7

7–100

0–7


100 mL CO2 min%1 air

100 mL He min%1 air

100 mL CO2 min%1 air

100 mL He min%1 air

Ambient 40 mL air min%1 CO2

0–120

100– 200

0–100

100– 200

0–100

150– 300

Ambient 40 mL 0–150 air min%1 He

400 mL min%1 He 400 mL min%1 CO2 Synthetic 400 mL air min%1 He Synthetic 400 mL air min%1 CO2

Synthetic air Synthetic air


Long term CO2 test



900 CO2

T Atmo(1C) sphere

In situ XRD 30– CO2 : N2 1000 (50 : 50 vol%) XRD of the spent membrane

XRD

Method

Other CO2 test

156

156

153

152

Ref.

No carbo- 157 nate formation Generation of impurity phase

Excellent chemical stability

Remarks

Chem Soc Rev

Ce0.8Sm0.2O2%d– Powder SrCo0.9Nb0.1O3%d mixing (60 : 40 wt%)

0.5

One-pot Disk (EC)

Ce0.8Sm0.2O2%d– Sm0.8Ca0.2CoO3 (75 : 25 wt%)

0.5

1.1

0.03

One-pot Disk (GN)

Disk

Powder mixing

0.03

0.09

0.09

Ce0.8Sm0.2O2%d– La0.9Sr0.1FeO3%d (70 : 30 vol%)

Disk

Disk

Powder mixing

Powder mixing

Disk

Powder mixing


(continued)

Ce0.9Gd0.1O1.95– (La0.8Sr0.2)0.95 MnO3%d (60 : 40 vol%)

Materials

Table 2


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

Nd0.5Sr0.5Al0.2Fe0.8O3%d–Ce0.8Nd0.2O2%d (40 : 60 wt%)

One-pot Disk (EC)

Nd0.6Sr0.4Al0.2Fe0.8O3%d–Ce0.9Nd0.1O2%d (40 : 60 wt%)

One-pot Disk (EC)

Disk

0.6

0.6

0.1

Pt

LSC

Tape casting

0.1

Y0.08Zr0.92O2%d– La0.7Sr0.3MnO3

Disk

Tape casting

Ce0.9Gd0.1O2%d– La0.7Sr0.3MnO3


Materials

(continued)


Table 2

0.03– 0.81 (CO2)

800– 0.15– 1000 1.0 (He)

0.02– 0.52 (CO2)

800– 0.08– 1000 0.6 (He)

800– 0.15– 950 0.5 (He)

150 mL 49 mL 1000 min%1 air min%1 He + 1 mL min%1 Ne 150 mL 49 mL 1000 min%1 air min%1 CO2 + 1 mL min%1 Ne 950

900

150 mL 49 mL 1000 min%1 air min%1 He + 1 mL min%1 Ne 150 mL 49 mL 1000 min%1 air min%1 CO2 + 1 mL min%1 Ne 950

0.27 (He + CO2)

XRD of the spent membrane


Method

T Atmo(1C) sphere

Other CO2 test

158

Ref.

160

Phase 160 remain unchanged

Phase 159 remain unchanged

Remarks

Review Article

75–170

50–75

0–50

125– 225

85–125

60–85

0–60

50–200

400 mL He 17–21 min%1 air 100 mL 100 mL 0–50 min%1 air min%1 He

0–7 400 mL He min%1 air 400 mL 50% He + 7–17 min%1 air 50% CO2

100 mL 50 mL min%1 air min%1 He + 50 mL min%1 CO2 0.61 150 mL 49 mL (He) min%1 air min%1 He + 1 mL min%1 Ne 150 mL 49 mL 0.52 (CO2) min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL 0.31 (CO2) min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL 0.15 %1 (CO2) min air min%1 CO2 + 1 mL min%1 Ne 1 (He) 150 mL 49 mL min%1 air min%1 He + 1 mL min%1 Ne 150 mL 49 mL 0.81 (CO2) min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL 0.5 (CO2) min%1 air min%1 CO2 + 1 mL min%1 Ne

700 0.55 (He) 0.42 (He + CO2) 0.55 (He) 900 0.28 100 mL 100 mL (He) min%1 air min%1 He


Long term CO2 test




View Article Online

Chem Soc Rev




Sm0.5Sr0.5Cu0.2Fe0.8O3%d– Ce0.8Sm0.2O2%d (40 : 60 wt%) 0.6

0.5

One-pot Disk (EC) 0.6

0.9

0.9

0.9

1

1

1

1

SDC-SCF

SDC-SCF

SDC-SCC

SDC-SCC

0.21– 0.43 (He) 0.2– 0.34 (He) 800– 0.29– 1000 1.27 (He)

800– 0.34– 950 0.7 (He)

0.35 (He)

0.47 (He)

1000 1.27 150 mL 49 mL (He) min%1 air min%1 He + 1 mL min%1 Ne



0.63 (CO2)

950 0.7 100 mL 30 mL (He) min%1 air min%1 He

950 0.22– 0.41 (He) 0.34 (CO2)

950 0.36– 0.45 (He) 0.37 (CO2)

950 0.1– 0.14 (He) 0.14– 0.16 (CO2) 950 0.45– 0.55 (He) 0.45 (CO2)

900 0.24– 0.19 (CO2)

150 mL 49 mL 0–25 min%1 air min%1 He + 1 mL min%1 Ne


100 mL 30 mL 140– min%1 air min%1 280 CO2 100 mL 30 mL 0–140 min%1 air min%1 He






150 mL 49 mL 170– min%1 air min%1 300 CO2 + 1 mL min%1 Ne


Long term CO2 test



T Atmo(1C) sphere

SEM of the spent membrane

XRD after 850 CO2 : O2 exposure to (99 : 1 CO2 vol%) XRD of the spent membrane

Method

Other CO2 test

Remain dense

Stable

Stable

Remarks

149

161

76

76

76

76

Ref.

Chem Soc Rev

One-pot Disk (EC)

0.5

0.5

One-pot Disk (CP)

Disk

Ce0.8Sm0.2O1.9– Powder Sm0.6Ca0.4FeO3%d mixing (67 : 33 wt%)

0.5

0.5

Disk

Ce0.8Sm0.2O1.9– Powder Sm0.6Ca0.4FeO3%d mixing (75 : 25 wt%)

0.5

0.5

One-pot Disk (SSR)

Disk

Ce0.8Sm0.2O1.9– Powder Sm0.6Ca0.4CoO3%d mixing (67 : 33 wt%)

Ce0.85Sm0.15O1.925–Sm0.6Sr0.4Al0.3Fe0.7O3 (75 : 25 wt%)

Disk


(continued)

Ce0.8Sm0.2O1.9– Powder Sm0.6Ca0.4CoO3%d mixing (75 : 25 wt%)

Materials

Table 2


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.

(b) Ce0.8Gd0.2O2%d–NiFe2O4 (40 : 60 vol%)


0.5

Powder mixing

0.6

0.08– 0.2 (He)

0.09– 0.27 (CO2)

La0.6Sr0.4- 900– 0.11– CoO3%d 1000 0.31 (He)

0.11– 1.15 (CO2)

800– 0.34– 1000 1.35 (He)

0.08– 1.12 (CO2)

150 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 29 mL min%1 air min%1 He

0.3 (CO2)

150 mL 29 mL 1000 0.27 min%1 air min%1 He (CO2) + 1 mL min%1 Ne

900 0.46– 0.38 (CO2)

150 mL 49 mL 1000 1.35 min%1 air min%1 He (He) + 1 mL %1 min Ne 150 mL 49 mL 1000 1.17 min%1 air min%1 (CO2) CO2 + 1 %1 mL min Ne 950 0.77– 0.70 (CO2)

900 0.29– 0.25 (CO2)

150 mL 49 mL 1000 1.15– min%1 air min%1 1.11 CO2 + 1 (CO2) mL min%1 Ne 950 0.70– 0.67 (CO2)

155– 230

115– 155

40–115

0–40

200– 290

75–200

25–75

Method

T Atmo(1C) sphere

Other CO2 test

150 mL 29 mL 0–40 In situ pow- 30– CO2 : N2 min%1 air min%1 der XRD 1000 (50 : 50 CO2 + 1 vol%) mL min%1 Ne 150 mL 29 mL 40–100 min%1 air min%1 CO2 + 1 mL min%1 Ne

150 mL 49 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL min%1 air min%1 He + 1 mL min%1 Ne 150 mL 49 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 49 mL %1 min air min%1 CO2 + 1 mL min%1 Ne


Long term CO2 test




Remarks

138

149

Ref.

Review Article

Disk

0.5

0.6

One-pot Disk (EC)

One-pot Disk (EC)


(continued)

Sm0.3Sr0.7Cu0.2Fe0.8O3%d–Ce0.8Sm0.2O2%d (40 : 60 wt%)

Materials

Table 2


View Article Online

Chem Soc Rev




Mn1.5Co1.5O4%d– One-pot Disk Ce0.9Pr0.1O2%d (GN) (40 : 60 wt%)

One-pot Disk (Pechini)

Ce0.8Tb0.2O2%d– NiFe2O4 (40 : 60 vol%)

Disk

Powder mixing

One-pot Disk (EC)

0.3/ 0.5

0.68

1.1

0.6

0.785

0.06– 0.17 (CO2)

GDCLSM-Pd

0.27– 0.31 (CO2)

1000 0.29– 0.31 (He)

0.22 (N2) 0.18– 0.16 (CO2) 940 0.27 (CO2)

0.18 (CO2)

0.19 (CO2)

0.14 (CO2) 1000 0.22 200 mL 29 mL (He) min%1 air min%1 CO2 + 1 mL min%1 Ne

Synthetic CO2 (No flow rate) air

T Atmo(1C) sphere

0–50

480– 580

130– 275 275– 480

100– 130

30–100

0–30

0– 101.667

200 mL 29 mL 0–1.25 min%1 air min%1 He + 1 mL min%1 Ne

Heating during 7 days

Annealed for 120 hours

139

Ref.

Chemically 141 stable


Remarks

154

800 CO2 : O2 : No reaction 132 SO2 : H2O (91.5 : 5 : 1 : 2.5 vol%)

850 CO2

0– In situ pow- 30– CO2 : air 36.667 der XRD 1000 (50 : 50 vol%)

Synthetic CO2 (No 50–76 air flow rate)

100 mL 150 mL min%1 air min%1 CO2 100 mL 150 mL min%1 air min%1 CO2 100 mL 150 mL min%1 air min%1 N2 100 mL 150 mL min%1 air min%1 CO2 100 mL 150 mL min%1 air min%1 CO2 Synthetic CO2 (No air flow rate)

150 mL 29 mL min%1 air min%1 He + 1 mL min%1 Ne 150 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 100 mL 150 mL min%1 air min%1 N2

Method

Other CO2 test

Chem Soc Rev

900– 0.26– 1000 0.48 (CO2)

0.01– 0.2 (CO2)

700– 0.015– Synthetic He (No 900 0.13 1000 0.17 air flow rate) (CO2) (He)

0.057– 100 mL CO2 min%1 air ln(pO2 0 / 0.34 pO200 ) = 8 (CO2)

100 mL N2 860 0.22 min%1 air ln(pO2 0 / (N2) 00 pO2 ) = 5

+ 1 mL min%1 Ne 150 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne


Long term CO2 test



La0.6Sr0.4- 700– 0.06– CoO3%d 940 0.39 (N2)


(continued)

Ce0.9Gd0.1O1.95%d– Zn0.96Al0.02Ga0.02O1.02 (50 : 50 vol%)

Materials

Table 2


View Article Online

Review Article

Chem. Soc. Rev.

Chem. Soc. Rev.


900– 0.15– 1000 0.54 (He) 0.26– 0.84 (He)

One-pot Hollow (EC) fiber

One-pot Hollow (EC) fiber

150 mL 29 mL 1000 0.2 min%1 air min%1 He (CO2) + 1 mL min%1 Ne

BSCF

0.02– 0.21 (CO2)


0–4

0–150

0–60

4.6–6.0

3.0–4.6

1.25– 3.0

150 mL 4–44 min%1 CO2 150 mL 44–50 min%1 Ar 150 mL 0–200 min%1 CO2

200 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 200 mL 29 mL min%1 air min%1 He + 1 mL min%1 Ne 200 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 200 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 150 mL 29 mL min%1 air min%1 CO2 + 1 mL min%1 Ne 100 mL 150 mL min%1 air min%1 Ar

0.13– 100 mL 0.14 min%1 air (CO2) 0.11 Ar 100 mL min%1 air 950 0.36 100 mL 100 mL 150 mL (CO2) min%1 air min%1 air min%1 He


850 0.11 GDC-LSM 750– 0.034– 100 mL 300 mL (Ar) 1000 0.34 min%1 air min%1 Ar (Ar)

Pr0.1Gd0.1Ce0.8O2%d– CoFe2O4 (50 : 50 wt%)

0.59

One-pot Disk (Pechini)

0.785 La0.6Sr0.4- 900– 0.06– CoO3%d 1000 0.18 (He)

NiFe2O4– Ce0.8Tb0.2O2%d (50 : 50 vol%)

0.5

0.2 (CO2)

0.2 (CO2)

0.22 (He)

0.2 (CO2)


Long term CO2 test



One-pot Disk (EC)


(continued)

Fe2O3–Ce0.9Gd0.1O2%d (40 : 60 wt%)

Materials

Table 2

T Atmo(1C) sphere

SEM of the spent membrane


In situ XRD 30– CO2 : Ar 1000 (50 : 50 vol%)

Method

Other CO2 test


CoCO3 found

Chemical stability


Remarks

162

133

155

Ref.

View Article Online

Review Article Chem Soc Rev


View Article Online


Chem Soc Rev

expressed in short as A–B (45 : 55) and vice versa for other compositions, unless otherwise specially stated if the percentage is by volume. 2.2.1. Oxygen ionic conductor phase. In choosing the ionic conducting materials, three criteria should be fulfilled.136 First, the materials have high ionic conductivity. Second, the materials remain stable under acidic gas or reducing atmosphere to ensure that no phase transition would occur. Third, the materials demonstrate chemical compatibility against the electronic conducting materials. Oxide compounds with a fluorite structure fulfill these criteria and are often used as an ionic conductor (IC, an electrolyte) in solid oxide fuel cells given their predominant ionic conductivity at a certain temperature range. Fig. 2(c) and (d) illustrate the ideal fluorite structure and its packing arrangement. Fluorite compounds are defined as those having a general formula of AO2; typified by calcium fluoride (CaF2). Such a structure is created by cubic-close packing of the A cations where the A cations occupy the face-centered cubic positions while the O anions occupy the tetrahedral interstices.21 Zirconia (ZrO2) exemplifies the classical example of oxides with a fluorite structure which was once an important series of electrolytes in the old generation of solid oxide fuel cells. This compound requires a temperature beyond 700 1C to provide substantial conductivity. Upon heating, ZrO2 undergoes phase transitions from monoclinic to tetragonal (at 1100–1200 1C) and tetragonal to cubic (2300 1C).163 Partial substitution of zirconia with (approximately 10 mole% of) various lower valence metal cations, e.g., Ca2+, Mg2+, and Ln3+ (lanthanides) effectively stabilizes the cubic phase to room temperature. The tetragonal phase can also be stabilized using approximately 3 mole% of Y3+. The ionic conductivity of the monoclinic ZrO2 is quite negligible, only about 10%7–10%6 S cm%1 which is improved on its tetragonal phase to 10%3–10%2 S cm%1 and approaches 10 S cm%1 for cubic ZrO2 beyond 2100 1C. Higher conductivity can actually be realized at lower temperature by partially substituting ZrO2 with CaO or Y2O3; leading to a conductivity value close to 10%1 S cm%1 at 1300 1C. Still, given the very high temperature required to achieve substantial oxygen ionic conductivity, the use of zirconia is considered impractical. Ceria (CeO2) represents another fluorite compound of interest. Strictly speaking, pure CeO2 actually behaves more as a mixed conductor showing almost similar magnitude of ionic and electronic conductivities.164 This is related with the change in the oxidation state of the cerium ion as a function of oxygen partial pressure, i.e., Ce4+ may be reduced to Ce3+ under reducing atmosphere (and vice versa), due to the formation of small polarons.165,166 This problem can be solved by partially substituting Ce with other metal cations having lower and fixed valence, e.g., Ln3+ (lanthanides, to promote the formation of oxygen vacancies so that the ionic conductivity far outweighs the electronic conductivity). What makes ceria-based materials attractive is their high oxygen ionic conductivity which at 750 1C is analogous to that of yttria-doped zirconia at 1000 1C. Although Bi2O3 based materials have superior conductivity, their disadvantages lie in their instability and reducibility in a reducing atmosphere. Ceria based materials on the other hand are quite stable.


Review Article

These materials have in fact replaced yttria-doped zirconia as an electrolyte for lower temperature operation (less than 800 1C).164 Various dopants can be used to improve the ionic conductivity, e.g., Ba, Ca, Dy, Gd, Ho, La, Mg, Nd, Sm, Sr, Y, and Yb.164,167 Among these dopants, Ln elements, which can be accommodated in ceria up to 40% (by mole), have attracted special attention.168 Balaguer et al. have provided a good summary in one of their studies on Ln doped ceria.169 We incorporate a portion of their summary here to provide more insight on the tailoring of ceria. Among the Ln dopants, Sm and Gd provide the highest ionic conductivities in addition to high stability.170 An La dopant favors the mobility of oxygen ions by offsetting the original resistance in a reducing atmosphere.170,171 Dy and Er act as the co-dopant with Y in ceria and as a dopant in a ceria–zirconia system, respectively.172,173 Tb and Pr are two special dopants with two valence states, which makes them suitable for O2 permeation membranes due to the improved p-type electronic conductivity of the resultant doped ceria.132,174 As discussed above, doping with Ln elements generally provides the best conductivity with Gd, Nd, and Sm-doped ceria showing the highest conductivity values.164,167,169,175 This is attributed to the high concentration of mobile oxygen vacancies in the cubic lattice. Fig. 9 suggests that the grain ionic conductivity of the lanthanide-doped ceria increase with increasing ionic radius of the dopant.175 Two special aspects in this figure deserve further discussion. Firstly, within this lanthanide-doped ceria series (except for Y-doped ceria), the lowest ionic conductivity was shown by Lu-doped ceria despite the almost similar ionic radius of Lu3+ (0.977 Å) and Ce4+ (0.97 Å). This is due to the restrained oxygen vacancy diffusion, most likely due to the attractive electrostatic interaction between

Fig. 9 Correlation of the temperature-dependent grain ionic conductivity of doped ceria (in air) with the ionic radius of the dopant (all contain 10 mole% lanthanide dopant (except for Y) and were synthesized using the solid-state route). Reproduced with permission from ref. 175. Copyright 2009, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

Chem. Soc. Rev.

View Article Online


Review Article

Lu3+ and oxygen vacancies. Secondly, Y-doped ceria shows an ionic conductivity which deviates from the trend indicated by the plot. This can be understood by the fact that Y is not the lanthanide element and the electronegativity of the dopant may influence the ionic conductivity.175 Although Nd-doped ceria demonstrates the highest conductivity among all lanthanide-doped ceria, Sm-doped ceria, i.e., Ce0.8Sm0.2O1.9 is the most attractive one to replace yttriadoped zirconia given its lowest reducibility among the doped ceria compounds.167,175 Balaguer et al. prepared a series of Ln-doped ceria (Ln = Gd, La, Tb, Pr, Eu, Er, Yb, Nd) and elaborated the effects of these dopants on the structure, oxygen vacancy concentration, and transport properties, as well as their relationships.169 Using Raman spectroscopy, they showed that Ln3+ dopants created disorder due to the formation of oxygen vacancies, the concentration of which increased with the cell parameter. The less distorted cells i.e., Gd- and Nd-doped ceria provided the highest ionic conductivity. Moreover, the less distorted cell parameters represent the best balance between Coulomb interactions, steric effects, and vacancy distribution. Another work probes the effect of Tb dopant concentration on the mixed ionic–electronic conductivity of ceria.176 The ionic conductivity was dominant at low Tb content (10%). When the Tb content was increased to 50% (Ce0.5Tb0.5O2%d), the materials displayed p-type electronic conductivity with higher ionic conductivity. Ce0.5Tb0.5O2%d also exhibited a higher surface exchange rate compared with those for ceria with a lower content of Tb.176 Furthermore, the addition of cobalt oxide improved the total conductivity of doped ceria, which was ascribed to the generation of electronic paths along the grain boundaries. The addition of transition metal oxides, e.g., cobalt oxide, iron oxide and copper oxide, as sintering aids, has been studied. These aids can decrease the sintering temperature on the doped ceria and enhanced their electronic conductivities (relative to the non-doped ceria).174,176–180 Another promising alternative is bismuth oxide (Bi2O3). This compound, upon heating, also exhibits phase transition from monoclinic (a-phase) to cubic (d-phase) at 730 1C before its melting point at 825 1C.181 Its cubic d-phase has a fluorite structure with ordered defects in the oxygen sub-lattice and shows 1–2 orders of magnitude higher conductivity than the yttria-doped zirconia, i.e., 1 S cm%1 at 650 1C. The superior ionic conductivity is characterized by the deficit of one-fourth of the oxygen anions in the lattice, the electronic configuration of Bi3+ consisting of 6s2 lone pair electrons which results in the high polarisability of the cation (and high oxide ion mobility) and the capability of Bi3+ to afford highly disordered surroundings.181 Likewise, the cubic phase of Bi2O3 can be stabilized at room temperature by partial substitution of Bi with Dy, Er, Gd, Ho, Lu, Tb, Tm, Y, and Yb. Despite some dispute regarding the resultant structure, Dy, Er, Ho, and Tb-doped ones show the highest conductivity of 0.1–0.4 S cm%1 between 650–700 1C. La1%xSrxGa1%yMgyO3%d is another popular oxygen ionic conductor, frequently applied as an electrolyte layer in solid oxide fuel cells due to its high oxygen ionic conductivity. It is also suitable as the ionic conductor phase in dual-phase membranes for oxygen permeation. However, La and Sr in La1%xSrxGa1%yMgyO3%d have the tendency to diffuse into the electronic conductor phase at high temperature, forming a new

Chem. Soc. Rev.

Chem Soc Rev

perovskite oxide phase with inferior permeability. For example, in the case of the La0.8Sr0.2Ga0.8Mg0.2O3 (LSGM)–La0.8Sr0.2Co0.2Fe0.8O3 (LSCF8228) dual-phase membrane, sintering at temperatures between 1320 and 1410 1C induced fast cation interdiffusion between the two phases leading to the formation of a single perovskite phase with local inhomogeneities. The ionic and electronic conductivities of the formed perovskite phase were lower than those of the original phases (LSGM and LSCF8228) thus unfavorable for oxygen permeation.182 2.2.2. Electronic conductor phase. The fluorite phase cannot be utilized as an oxygen permeation membrane given its limited electronic conductivity. An electronic conductor (EC) phase having high electronic conductivity is required to form the continuous electronic conducting path within the membrane. The EC phase should ideally have good stability, and high thermal and chemical compatibility with respect to the ionic conductor (IC) phase, and be relatively inert to the reaction with the IC phase. The EC phase mainly comes from three categories, i.e., noble metals, electronic conducting oxides, and mixed conducting (MIEC) oxides. Noble metals (i.e., Au, Pt, Pd, Ag, etc.) were the initial EC candidate for dual-phase membranes.183–187 Both the individual EC and IC phases should be percolative throughout the membranes for the oxygen ions and electrons to be conducted from the feed to the permeate side. Thus, the percolative threshold which refers to the minimum content of the electronic phase is generally 40 vol% or above using the conventional preparation protocols, e.g., mixing, pressing, and sintering. For the Bi1.5Y0.3Sm0.2O3 (BYS)–Ag membrane, for example, 40 vol% of Ag was required to achieve a percolative EC phase as indicated by 104–105 higher conductivity for BYS–40 vol% Ag than that for BYS–30 vol% Ag.185 Conceptually, since there is practically no oxygen conductivity in the noble metal phase, the oxygen conducting path in an IC-noble metal composite is negatively blocked by the EC phase (Fig. 8(a)). The IC phase also blocks the electronic conducting path given the very limited electronic conductivity in the IC phase (such as Ce0.8Gd0.2O1.9 (GDC)). This blocking effect thus limits the global conducting efficiency between these two phases. In addition to this so-called ‘‘mutual blocking effect’’, the high cost of noble metals renders this cermet dual-phase membrane approach less attractive. Pure electronic conducting oxides or those with extremely low ionic conductivity and substantial oxygen reduction catalytic activity can be exploited as the EC phase as represented by the Ce0.8Gd0.2O1.9 (GDC)–La0.7Sr0.3MnO3 (LSM) membrane.188 Given its high electronic conductivity, the electronic conducting oxide phase is generally perovskite-type materials (such as La1%xSrxMn(Cr)O3)136,189–192 or spinel oxides (such as MnFe2O4, CoFe2O4, FeCo2O4, MnCo2O4, CuCo2O4, NiFe2O4, etc.).132,138,139,154,155,162,193,194 Besides the blocking effect from the EC phase (Fig. 8(a)), interphase reaction becomes another important consideration in dual-phase membrane design. Although mixing with LSM solved the limited electronic conducting issue, the long-term test (up to 800 hours) revealed a continuous oxygen flux decrease due to the formation of poorly ionic conducting layers at the boundary of GDC grains from the diffusion of lanthanum and strontium.188 In some cases, new phases formed on the


View Article Online


Chem Soc Rev

grain boundary can also promote ionic transport via modification of the phase structure and compositions.195 A recent work on dualphase Ce0.8Gd0.2O2%d (GDC)–CoFe2O4 (CF) showed a new crystalline phase containing Gd, Fe, Ce, and Co (denoted as GFCCO) formed along the grain boundaries during sintering.196 On one hand, the perovskite GFCCO has a high level of oxygen vacancies which promotes oxygen ionic conductivity. On the other hand, the formation of the GFCCO phase mitigates the space charge effect near the GDC–GDC grain boundaries which can also improve the oxygen ion transport. Their work demonstrates a novel direction in EC phase selection, i.e., the possibility to generate in situ a beneficial phase reaction with IC that creates a new phase with high ionic conductivity.196 Another example of this direction comes from dualphase membranes of Ce0.8Sm0.2O2%d (SDC) + SrCO3 + Co3O4, where a highly conductive tetragonal perovskite phase Sr0.95Sm0.05CoO3%d formed along the SDC grain boundary. The permeation flux through a 0.5 mm-thick SDC + 20 wt% SrCO3 + 10.89 wt% Co3O4 reached 1.19 mL min%1 cm%2 at 900 1C under an air/helium gradient. This composite membrane furthermore displayed a stable oxygen permeation flux during a 110 hour-test using pure CO2 as the sweep gas.197 Mixed ionic–electronic conducting (MIEC) oxides such as (Lnx(Ba,Sr,Ca)1%xCoyFe1%yO3%d; Ln = La, Pr, Nd, Sm, and Gd) are the other EC phase option.143,144,150,198–204 The oxygen permeability of an IC-MIEC composite is typically higher than that of an IC-EC composite due to the higher amount of oxygen ion conducting paths through the membranes and the enhanced surface exchange reaction rate (Fig. 8(b)). The presence of a MIEC phase within the dual-phase composite substantially enlarges the triple phase boundaries (TPBs), reaction contact areas between oxygen ions, electrons, and oxygen gases on the membrane surfaces as well as the contact areas between oxygen ions and electrons within the bulk membrane. Note that both oxygen ions and electrons can pass through the MIEC phase. This is unlike the IC-EC composite case where the TPBs and the oxygen ions to electron contact areas are limited only to IC-EC interfaces (Fig. 8(a)).205 A representative example is Ce0.9Gd0.1O2%d (GDC)–Ba0.5Sr0.5Co0.8Fe0.2O3%d (BSCF5582) which consists of a GDC IC phase and a BSCF5582 MIEC phase. BSCF5582 has a high oxygen ionic conductivity of about 1 S cm%1 (above 300 1C) and an even higher electronic conductivity.206 BSCF5582 is a popular oxygen permeation membrane material that has good compatibility with GDC.77,207 A quite high oxygen permeation flux of 2.1 mL min%1 cm%2 was obtained from a 0.5 mm-thick 60 wt% GDC–40 wt% BSCF5582 membrane under an air/He gradient at 975 1C.208 This is significantly much higher than that from the GDC (ionic conducting phase)–LSM (electronic conducting phase) composite membrane.188 When the helium sweep gas was swapped to pure CO2, a stable oxygen permeation flux of 0.67 mL min%1 cm%2 at 950 1C was retained for at least 250 hours.137 Such long-term stability to CO2 is counterintuitive and was attributed to the protection of BSCF5582 by GDC.28 Accordingly, Ce0.8Sm0.2O2%d (SDC)–SrCo0.9Nb0.1O3%d (SCN) (60 : 40) also showed excellent CO2 stability for at least 120 hours under 50 vol% CO2 in N2.157 CO2 resistance can further be improved by replacing the alkaline earth metal containing perovskite with a non-alkaline earth


Review Article

metal containing perovskite, such as Ln1%xSrxFeO3%d (Ln = La, Pr, Nd, Sm, and Gd).145,150,153,209 For example, La1%xSrxFeO3%d showed appreciable oxygen permeation flux under He, N2, and even reducing CO, CO2, and H2 atmospheres,210–212 indicating its good stability in harsh atmospheres. IC-MIEC dual-phase membranes can simultaneously provide high oxygen permeation flux and good chemical stability in the presence of acidic gas. Moreover, the thermal expansion coefficient mismatch and the interphase reaction between the two phases should also be accounted for in the design of IC-MIEC dual phase membranes. Other electronic conducting oxides such as double perovskite oxides and transition metal oxides (such as Fe2O3) were employed in dual-phase oxygen permeation membranes.155,198,200 2.2.3. Preparation routes. Ceramic-based dual-phase membranes offer different preparation routes. For single-phase membranes, the preparation methods mainly affect the grain size and defects of the membranes. The different preparation methods may also influence the distribution of the two phase compositions for the dual-phase membranes.213 Powders of each individual phase can initially be synthesized using solidstate reactions or sol–gel methods and then combined by manually mixing using a mortar and pestle or by mechanical mixing using ball-milling. It is also possible to use an in situ one-pot sol–gel synthesis to prepare the dual-phase directly from the precursors. One of the best examples comparing the distinct morphology and oxygen permeability of dual-phase membranes arising from the different preparation routes is given by Luo et al.139 For NiFe2O4 (NF)–Ce0.9Gd0.1O2%d (GDC) (40 : 60), they showed that powder mixing by hand using a mortar and pestle led to the formation of the largest grains (Fig. 10(a) and (b)) followed by mixing using ball-milling (smaller grains – Fig. 10(c) and (d)) and then by a one-pot route (smallest grains – Fig. 10(e) and (f)). Their results showed close correlation between the grain size and the oxygen flux with the composites having the smaller grain size exhibited higher fluxes. As such, the one-pot route is favored over these physical mixing methods. According to the precursors and synthesis routes, the onepot method can be divided into several types, i.e., one-pot solid state reaction (SSR), one-pot co-precipitation (CP), one-pot glycine–nitrate combustion (GN), and one-pot sol–gel (SG). For example, Luo et al. applied one-pot glycine–nitrate combustion techniques to prepare a Mn1.5Co1.5O4%d–Ce0.9Pr0.1O2%d (40 : 60) dual-phase membrane.154 Glycine–nitrate combustion is a common method to synthesize ceramic oxide powders with an atomic scale uniform composition and fine grain size, i.e., to obtain an improved sintering effect.154,214 The powders obtained from rapid glycine–nitrate combustion were pressed into disk membranes followed by sintering at 1300 1C for 10 hours. The sintered disk membrane consisted of well-distributed grains of Mn1.5Co1.5O4%d and Ce0.9Pr0.1O2%d with the average grain size areas of 0.265 mm2 and 0.329 mm2, respectively. Zhu et al. studied the effect of preparation methods on dual-phase membranes. They synthesized Ce0.85Sm0.15O1.925 (SDC)–Sm0.6Sr0.4Al0.3Fe0.7O3 (SSAF) (75 : 25) dual-phase membranes using three different methods, i.e., the EDTA–citrate (EC) sol–gel process, SSR method and CP

Chem. Soc. Rev.

View Article Online


Review Article

Fig. 10 Scanning electron microscopy images (left column) and backscattered electron microscopy images (right column) of the grain morphology of Ce0.9Gd0.1O2%d (GDC)–NiFe2O4 (NFO) (40 : 60) prepared by powder mixing using (a and b) a mortar and pestle and (c and d) ballmilling; and (e and f) by an in situ one-pot sol–gel route. Reproduced with permission from ref. 139. Copyright 2011, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

method.161 The two phases were most homogeneously mixed using the EC method. The SSR method gave the most inhomogeneous phase mixtures. Nonetheless, the composite obtained from the EC method displayed the lowest total electrical conductivity and oxygen permeation flux whereas that obtained from the SSR method showed the highest total conductivity and oxygen flux. Such discrepancy was attributed to the volume fraction of the perovskite phase that was lower than the percolation threshold (30 wt%) which becomes more important for homogenously mixed phases (EC case). For an inhomogeneous composite system (SSR case), on the other hand, a continuous conduction network was likely to form over random sections within the membrane. 2.2.4. Internal and external short circuit concepts. A dualphase or composite membrane is a manifestation of an internal short-circuit concept given the separate pathways for the ionic and the electronic conductions within such membranes. Such an approach combines the advantages of individual phases to attain high oxygen permeability and high chemical stability to acidic gas (CO2). One of its drawbacks lies in the relatively high amount of the individual phases required; for instance, at least 40 vol% of each phase is necessary to form a continuous network across the membrane, particularly when prepared by the conventional preparation route with steps of mixing, pressing, and sintering. Moreover, numerous isolated areas

Chem. Soc. Rev.

Chem Soc Rev

Fig. 11 Cross-section schematic diagram for (a) a dual-phase membrane; (b) a dual-phase membrane with a minimum electronic conducting phase (internal short-circuit); (c) a membrane with external short-circuit decoration; and (d) a membrane with dual-phase external short-circuit decoration. Reproduced with permission from ref. 222. Copyright 2015, The Royal Society of Chemistry.

may form within the membrane (marked by red circles in Fig. 11(a)) which leads to ionic and electronic species blocked by each other and eventually a low oxygen permeation flux. This isolation issue can be circumvented by increasing the amount of IC phase to its maximum (which is equivalent to decreasing the amount of EC phase to its minimum) while ensuring that both phases are still continuous. To this end, Fig. 11(b) illustrates one of the possible design concepts as realized in Chen et al.’s work.215 They dispersed fiber-shaped PrBaCo2O5+d (PBC) into a SDC matrix. Only 20 vol% of the PBC phase was required in the dual-phase membrane to attain a relatively high oxygen permeability (for dual-phase membranes) of 0.56 mL min%1 cm%2 at 940 1C. Twenty percent by volume can be regarded as the lowest volume fraction of EC reported so far. What they illustrated essentially leads to the internal short-circuited membrane design with the lowest amount of electronic conducting pathway (and phase). Zhang et al. showed another good example of this minimum EC phase approach using a SDC membrane.216 A channel through the disk SDC membrane was first made by an ultra-high speed drill, after which a thin Ag wire was implanted into the channel and sealed with silver paste. A single internal short circuit can provide an oxygen flux as high as those attained by putting a larger number of shortcircuit wires through the membranes. Still along this approach, Joo’s group developed a novel ‘internal short circuit’ concept, i.e., a modification of the traditional dual-phase membrane


View Article Online


Chem Soc Rev

concept, which they defined as a segmented structure.158,159 Instead of mixing the two ceramic powders together, they fabricated a fluorite membrane (YSZ or GDC) with segmented electron conducting oxides (LSM) in the membrane, which manifested into the identical structure shown in Fig. 11(b). Such a segmented structure did not display the phase-to-phase reaction issue regularly observed in dual phase membranes such as GDC–LSM and YSZ–LSM enabled by the presence of numerous interfacial areas between the two phases that promote such reaction.188,217 In membranes with a segmented structure, there is a limited interfacial contact area between the two phases. The long-term constant performance of this segment membrane highlights its stability. Another concept was recently devised to address the limited electronic conductivity of the IC membrane via an external short circuit.218–221 Fig. 11(c) represents the external short circuit approach. An electronic conductor layer (such as noble metal, EC, and/or MIEC) and a metal sealant over the membrane surface are employed as an electron pathway from the high oxygen potential side to the low oxygen potential side. Relatively high oxygen permeation fluxes through the Ag decorated SDC membrane can be obtained at relatively low temperatures even at 600 1C by employing such an approach using Ag.218 EC decoration on the surface has two functions. The first function is to provide an external electron pathway so the electrons released from the surface oxygen exchange reactions at the permeate side can be shuttled to the other membrane surface to consume the incoming electrons and complete oxygen reduction from the molecular state to the lattice oxygen. The second function specifically can be best achieved using MIEC materials that display high surface oxygen exchange reaction rates such as BSCF5582 and La0.8Sr0.2CoO3%d.219,221 The utilization of MIEC for the external short circuit approach is attractive to replace noble metals and to increase oxygen permeability. Most perovskite-based MIEC materials nevertheless are not stable to acidic gas (CO2).221 However, in real applications, the feed side is in a relatively mild gas atmosphere (pressurized air); thus these MIEC materials can be applied as the catalyst in the membrane surface at the feed side leaving the other membrane surface side to use some CO2 tolerant materials. A recent work demonstrates the use of dual-phase decoration over the GDC membrane to generalize the extendibility of the composite concept to the external shortcircuit approach.222 Fig. 11(d) displays the structure of such dualphase decoration that provides an electronic conducting pathway and additionally, higher amounts of triple phase boundaries on the surfaces of the membrane (relative to regular EC decoration in Fig. 11(c)). Besides the higher oxygen permeability, this approach provides advantages in terms of the lower amount of Ag required (relative to Ag decoration) and higher stability against CO2 (relative to MIEC decoration). Despite the very low oxygen permeation fluxes demonstrated by the best performing membrane utilizing internal and external short circuit approaches, they have very high chemical stability that allows their application in membrane reactors such as syngas production from methane. Under such reaction conditions, the oxygen concentration gradient through the


Review Article

membranes is substantially larger than that employed in our permeation experiments (for pure oxygen production), i.e., 0.21 atm (air)/10%8 to 10%16 atm (reaction) as opposed to 0.21 atm (air)/10%3 atm (sweep) (permeation); thus allowing substantially higher oxygen fluxes to be attained. 2.2.5. Recent status and advances. A Ce0.9Pr0.1O2%d(CP)– Pr0.6Sr0.4Fe0.5Co0.5O3%d (PSFC) (60 : 40) membrane was prepared via a one-pot sol–gel synthesis method and tested for its CO2 resistance.142 In situ powder X-ray diffraction (XRD) between 30 and 1000 1C in a CO2-containing atmosphere (CO2– air = 50 : 50 vol%) showed no carbonate formation on this composite and the CP phase retained its cubic structure while the PSFC phase underwent two phase transitions, e.g., from orthorhombic to rhombohedral at ca. 600 1C and from rhombohedral to cubic at ca. 800 1C. In an earlier work, cobalt-free Ce0.9Pr0.1O2%d (CP)–Pr0.6Sr0.4FeO3%d (PSF) (60 : 40) was synthesized and tested under the harsh partial oxidation of methane (POM) reaction conditions and in a CO2 atmosphere.150 Unlike CP-PSFC, CP-PSF showed no phase transitions during the in situ powder XRD test in air. No carbonate formation was observed during the in situ XRD test in a CO2-containing atmosphere (CO2–N2 = 50 : 50 vol%). Using CO2 as the sweep gas at 950 1C, the O2 flux of a 0.6 mm thick membrane reached up to 0.22 mL min%1 cm%2 and then slightly decreased and stabilized at around 0.18 mL min%1 cm%2 for more than 100 hours. At POM reaction conditions, the O2 flux of the membrane approached 4.4 mL min%1 cm%2 at 950 1C for 100 hours. The powder XRD pattern for the 100 hour-POM tested membrane indicated the retention of the original two phases in the composite, highlighting the membrane stability in a reducing atmosphere. Fang et al. developed a Ce0.85Gd0.1Cu0.05O2%d (CGC)–La0.6Ca0.4FeO3%d (LCF) (75 : 25) membrane.66 Five percent (mole) of Ce in Ce0.9Gd0.1O2%d was substituted by Cu, forming CGC with the aim to increase the ionic conductivity and to induce p-type electronic conductivity as well as to improve the sintering behavior and densification of ceria. Since Cu-doped Ce0.9Gd0.1O2%d showed larger grain size than Ce0.9Gd0.1O2%d, Cu can be considered to act as a sintering aid. Based on electron energy loss spectroscopy, it can be found that the grain boundary in between CGC grains contains La, Ca, and Fe with a negligible amount of Cu; thus ruling out significant segregation of Cu in the grain boundary. The membranes were prepared using a one-pot method and powder mixing. In terms of surface morphology, the former membranes were much more homogeneous than the latter ones which consequently led to higher oxygen permeation fluxes for the former membranes. Cu doping can also be performed in the electronic conducting perovskite phase (instead of the ionic conducting phase), for example, to partially substitute Fe.149 Partial substitution of the B-site cation with Cu was reported to enhance the oxygen permeation flux.223–225 In this context, dual-phase Sm0.5Sr0.5Cu0.2Fe0.8O3%d (SmSrCF5528)–Ce0.8Sm0.2O2%d (SDC) (40 : 60) and Sm0.3Sr0.7Cu0.2Fe0.8O3%d (SmSrCF3728)–Ce0.8Sm0.2O2%d (SDC) (40 : 60) membranes were prepared via an EDTA–citric acid complexing sol–gel process. Powder X-ray diffraction, scanning electron microscopy, and energy dispersive X-ray analysis results revealed the formation of the CuO phase on the surface of the SmSrCF5528–SDC membrane. The CuO

Chem. Soc. Rev.

View Article Online


Review Article

phase disappeared with increasing Sr content (SmSrCF3728–SDC) and the parallel increase in the unit cell of SmSrCF3728 suggested that Cu was incorporated into the perovskite structure at larger Sr contents. Relatively high oxygen fluxes of 1.12 and 1.15 mL min%1 cm%2 were obtained at 1000 1C from SmSrCF5528– SDC and SmSrCF3728–SDC membranes, respectively. At 1000 1C, both membranes demonstrated good stability against CO2. Based on the Ellingham diagram,41 the stability of SrCO3 conceptually increases with decreasing temperature at constant CO2 partial pressure. Accordingly, at lower temperatures (i.e., 950 and 900 1C), a stabilization stage occurred before the oxygen fluxes can reach their stable values. This is likely related to the tendency for Sr to react with CO2 to form SrCO3 on the sweep side. A CuO layer appeared on the surface of the feed side of the SmSrCF3728– SDC membrane which was caused by cationic counter-diffusion of Cu to the oxygen-rich side during the permeation process. Regardless of the presence of a CuO layer and a Cu-depleted layer, a steady-state oxygen permeation flux was observed; the reasons of which require further investigation. Balaguer et al. reported Ce0.8Tb0.2O2%d (CT)–NiFe2O4 (NF) (40 : 60 vol%) membranes prepared using a one pot method.132 Powder X-ray diffraction, energy dispersive X-ray spectroscopy and Raman spectroscopy demonstrated that CT–NF was stable in a CO2 environment without reactions occurring between the two phases and CO2. Furthermore, relative to the Ce0.8Gd0.2O2%d (GDC)–NiFe2O4 (NF) (40 : 60 vol%) membrane, the CT–NF membrane exhibited larger oxygen permeation flux due to the higher p-type electronic conductivity of CT enabling mixed ionic–electronic conducting behavior to occur. A similar work on Ce0.8Tb0.2O2%d (CT)–NiFe2O4 (NF) (50 : 50 vol%) displayed an O2 flux of 0.13 mL min%1 cm%2 for a 0.59 mm-thick membrane under an air/CO2 gradient.133 A prolonged stability test under 100% CO2 showed that the O2 flux progressively increased from 0.13 to 0.14 mL min%1 cm%2 during the 40 hour-test. Similar improvement of O2 flux during the long-term test under CO2 was reported by other studies.132,138,141,159 Yun et al. suggested that the possible reason is the ‘electrode activation’ effect which is widely observed in the SOFC field.159 The exact reason behind the O2 flux enhancement has not yet been clarified. For GDC–NF dual-phase membranes, Luo et al. developed Ce0.9Gd0.1O2%d (GDC)–NiFe2O4 (NF) (40 : 60) via one-pot and powder-mixing methods.138,139 No carbonate was formed during in situ powder X-ray diffraction between 30 1C and 1000 1C in an atmosphere containing 50 vol% CO2 and 50 vol% N2 (as well as 50 vol% CO2 and 50 vol% air). Another spinel structured MnCo2O4%d possessing excellent electronic conductivity and structural stability exhibited a thermal expansion coefficient (11.7 ( 10%6 K%1) close to that of Pr-doped ceria Ce0.9Pr0.1O2%d (10–11 ( 10%6 K%1).226,227 Luo et al. synthesized a dual-phase Mn1.5Co1.5O4%d (MC)–Ce0.9Pr0.1O2%d (CP) (40 : 60) membrane.154 Powder X-ray diffraction results showed that Mn1.5Co1.5O4%d comprised 57.9 wt% cubic MnCo2O4%d and 42.1 wt% tetragonal Mn2CoO4%d. The highest O2 flux of 0.48 mL min%1 cm%2 was obtained through a 0.3 mm-thick membrane at 1000 1C under an air/CO2 gradient. The membrane could operate for at least 60 hours under pure CO2 at 1000 1C without degradation; maintaining an O2 flux of 0.2 mL min%1 cm%2

Chem. Soc. Rev.

Chem Soc Rev

within this period. Zhu et al. made Ce0.8Sm0.2O1.9 (SDC)– SmMn0.5Co0.5O3 (SMC) (75 : 25) and Ce0.8Sm0.2O1.9 (SDC)– Sm0.8Ca0.2Mn0.5Co0.5O3 (SCMC) (75 : 25) and studied their stability under CO2.143 They found that SDC–SCMC showed a higher O2 flux at 940 1C yet a lower O2 flux at 850 1C (relative to SDC–SMC). The decreased O2 flux at lower temperatures was due to the stronger CO2 adsorption on the membrane surface. Ca2+ doped SmCoO3%d is envisioned to have higher CO2 stability but lower flux (relative to SmCoO3%d). A dual phase membrane comprising 75 wt% SDC and 25 wt% Sm1%xCaxCoO3%d was developed to investigate the Ca2+ doping effect.156 Two compositions, i.e., Ce0.8Sm0.2O2%d (SDC)–Sm0.8Ca0.2CoO3 (SCC82) (75 : 25) and Ce0.8Sm0.2O2%d (SDC)–Sm0.6Ca0.4CoO3 (SCC64) (75 : 25) were studied. At lower temperatures (i.e., 800–875 1C), for both membranes, the O2 fluxes were lower when pure CO2 was used as sweep gas, compared to pure He as the sweep gas case. However, the O2 flux increased under pure CO2 at higher temperatures (875–950 1C). This trend can be explained by the weaker CO2 adsorption at higher temperatures. Both the SDC–SCC82 and SDC–SCC64 membranes showed stable O2 fluxes under pure CO2 for at least 100 hours. The SDC–SCC64 and SDC–SCC82 membranes exhibited an O2 flux of 0.16 and 0.19 mL min%1 cm%2, respectively. Li et al. further compared the oxygen permeation behaviors of Ce0.8Sm0.2O2%d–Sm0.6Ca0.4CoO3%d (SDC–SCC) and Ce0.8Sm0.2O2%d–Sm0.6Ca0.4FeO3%d (SDC–SCF).76 Two different composites were made for each composite type, resulting in 4 composites, i.e., SDC–SCC (75 : 25), SDC–SCC (67 : 33), SDC–SCF (75 : 25), and SDC–SCF (67 : 33). Co containing membranes displayed significantly higher O2 flux than Fe containing membranes. Upon comparing SDC–SCC (75 : 25) to SDC–SCC (67 : 33), the O2 flux of the latter was almost 4 times that of the former. This is due to the fact that 25 wt% SCC did not provide sufficient electronic conductivity and catalytic activity as 33 wt% SCC did. Still, SDC–SCF (75 : 25) showed higher O2 permeability than SDC– SCF (67 : 33) due to the lower O2 permeability of SCF (compared to SDC). Despite the discrepancy in the oxygen fluxes observed for these four membranes, all of them showed excellent stability during 100 hour-operation using CO2 as the sweep gas. Jiang et al. doped the electronic conducting phase PrBaCo2%xFexO3%d (PBC2%xFx) into GDC to fabricate a dual-phase membrane GDC– PBC2%xFx (60 : 40).140 Earlier, it was found that the fluorite structure SDC can enhance both the oxygen permeability and structural stability of the PrBaCo2O5+d (PBC) membrane.199 In the case of GDC–PBC2%xFx, the introduction of Fe (x = 1.5) can improve the oxygen flux by B1.7 times more than GDC–PBC. For example, GDC–PBC0.5F1.5 showed an oxygen flux of 0.36 mL min%1 cm%2 at 925 1C with pure CO2 as the sweep gas. A ZnO doped Gd0.1Ce0.9O1.95%d membrane was reported recently.141 The Al0.02Ga0.02Zn0.96O1.02 (AGZ)–Gd0.1Ce0.9O1.95%d (GDC) (50 : 50 vol%) dual-phase membrane showed chemical stability against CO2. By doping a slight amount of Al and Ga, ZnO exhibits excellent electronic conductivity in the membrane without sacrificing the ionic conductivity of GDC. Moreover, ZnO is chemically stable in CO2 at high temperatures due to the decomposition of ZnCO3 at above 200 1C in air. At 860 and


View Article Online


Chem Soc Rev

940 1C, a stable O2 flux was proven for over 500 hours with pure CO2 as the sweep gas. To further enhance the oxygen ion conductivity of Ce0.8Gd0.2O2%d, Pr can be used to substitute half of the original Gd to form Pr0.1Gd0.1Ce0.8O2%d.228 A Pr0.1Gd0.1Ce0.8O2%d (PGC)–CoFe2O4 (CF) (50 : 50) membrane was fabricated to evaluate its oxygen permeation behavior in a CO2 atmosphere.162 The O2 flux was found to gradually increase within the first 100 hours before reaching its steady value of 0.36 mL min%1 cm%2 for the next 100 hours. Compared with Gd and Sm, Nd-doped ceria is reported to exhibit the highest ionic conductivity.175 Recognizing this fact, Luo et al. developed a Ce0.9Nd0.1O2%d (CN)–Nd0.6Sr0.4FeO3%d (NSF) (60 : 40) dual-phase membrane which showed excellent chemical stability under CO2.145 A stable oxygen flux of 0.48 mL min%1 cm%2 under a CO2 atmosphere was obtained with a porous La0.6Sr0.4CoO3%d coating on the 0.6 mm membrane. In situ powder X-ray diffraction between 30 and 1000 1C of the dual-phase membrane in a CO2 containing atmosphere dismissed the existence of carbonate and phase transition, suggesting that the dual-phase membrane is thermally and chemically stable in CO2. Furthermore, the incorporation of an Al dopant at the B-site was reported to increase the thermochemical stability under a reducing atmosphere.229,230 The oxygen permeability can also be enhanced by such Al incorporation at the B-site. Nd0.6Sr0.4Al0.2Fe0.8O3%d (NSAF6428)–Ce0.9Nd0.1O2%d (CN91) (40 : 60) and Nd0.5Sr0.5Al0.2Fe0.8O3%d (NSAF5528)–Ce0.8Nd0.2O2%d (CN82) (40 : 60) were developed and their oxygen permeation properties were studied under He and CO2 atmospheres.160 The higher Sr content composition (NSAF5528–CN82) exhibited higher O2 flux than the lower Sr content composition (NSAF6428– CN91) irrespective of whether He or CO2 was used as the sweep gas. The O2 flux of NSAF5528–CN82 however decreased from 0.24 to 0.19 mL min%1 cm%2 at 900 1C during the 130 hour-test under CO2. In the NSAF6428–CN91 case, a stable O2 flux of 0.15 mL min%1 cm%2 at 900 1C was maintained for at least 100 hours. These results denote the negative effect of Sr towards the stability under a CO2 atmosphere. A novel cobalt-free dualphase membrane with a composition of Ce0.8Sm0.2O2%d (SDC)– Ba0.95La0.05Fe1%xZrxO3%d (BLF1%xZx) (60 : 40) was developed as an oxygen permeable membrane for oxy-fuel combustion.146 The substitution of Fe by Zr reduced the basicity of BLF1%xZx as is evident from the X-ray photoelectron spectra; thus enhancing the stability of the dual-phase membrane under CO2. For a 1.0 mm thick membrane with 15% Zr doping (SDC–BLF0.85Z0.15), when pure CO2 was applied as the sweep gas at 925 1C, a stable oxygen flux of 0.24 mL min%1 cm%2 was obtained for more than 80 hours. The oxygen permeation behaviors under CO2 for a similar dualphase material series Ce0.8Gd0.2O2%d (GDC)–Ba0.95La0.05Fe1%xNbxO3%d (BLF1%xNx) (60 : 40) were studied by Cheng et al.148 An oxygen flux of approximately 0.19 mL min%1 cm%2 was displayed by GDC–BLF0.9N0.1 for 95 hours under the same conditions (under pure CO2 sweep gas). It is quite counterintuitive that the oxygen permeation flux of the GDC–BLF1%xNx membrane could not be fully recovered after the sweep gas was changed from CO2 to He, unlike the full recovery observed for the SDC–BLF1%xZx membrane, despite the absence of carbonate formation in both cases. For example, the oxygen flux of GDC–BLF0.85N0.15 was 0.23 mL min%1 cm%2 in pure He during the


Review Article

first 10 h, but after 30 h under CO2-containing sweep gas, the oxygen flux became 0.18 mL min%1 cm%2 when the sweep gas was changed from CO2 to He. There was no explicit explanation given for this phenomenon, particularly for the oxygen flux decrease with increasing CO2 concentration (from 15% to 100%). Very recently, a highly stable dual-phase Y0.8Ca0.2Cr0.8Co0.2O3 (YCCC)–Sm0.2Ce0.8O1.9 (SDC) (50 : 50) membrane was developed for oxygen separation under harsh operation conditions.147 The membrane was composed of Ca- and Co-doped yttrium chromite (YCCC) and samarium doped ceria (SDC). Generally, chromite-based ceramic is difficult to be densified by sintering due to the formation of thin Cr2O3 layers at the interparticle neck at the beginning of sintering. In their work, a low melting point cobalt dopant was used to lower the sintering temperature by forming a liquid phase during sintering. The relative density of the sintered YCCC–SDC was over 97% (gas-tight) after being sintered at 1400 1C, in contrast to the highly porous structure of YCC–SDC without Co-doping that featured poor sintering characteristics. The oxygen permeation flux was as high as 6.7 mL min%1 cm%2 at 950 1C for a 1.3 mm-thick membrane when the forming gas (i.e., 3 vol% H2 and 97 vol% N2) was used as the sweep gas. When 50 vol% CO2 was introduced into the forming gas, the oxygen flux decreased to 3.08 mL min%1 cm%2 with the degradation rate of less than 2% during the 350 hour test. Moreover, after the 350 hour test with the co-existence of H2 and CO2, the membrane showed no phase decomposition and carbonate formation as evidenced by powder XRD patterns. It should be noted that the very high oxygen flux (of 6.7 mL min%1 cm%2 at 950 1C) was ascribed to the 3% H2 in the sweep gas. Moreover, under reducing conditions, the electronic conductivity of SDC can be more dominant than its ionic conductivity which also contributes to such high oxygen flux.231,232 The work of Yoon and Marina showed that without H2 in the sweep gas, the oxygen flux was very low, only 0.34 mL min%1 cm%2. Kharton et al. reported that the oxygen permeation flux through GDC/LSM composites was 10 times lower than that expected due to the inter-diffusion of the constituents.188 However, another work on the Ce0.9Gd0.1O2%d (GDC)–La0.7Sm0.3MnO3)d (LSM) (80 : 20 vol%) dual-phase membrane demonstrated that the O2 flux can be increased substantially by applying surface decoration and lowering the LSM content to 20 vol%.151 Without surface decoration, the bare GDC–LSM (80 : 20 vol%) displayed an identical O2 flux relative to the GDC membrane. The presence of La0.6Sr0.4CoO3%d (LSC) as the surface decoration on the feed side led to a marginal increase in the O2 flux. However, the increase was improved by an order of magnitude when LSC decoration was present on the permeate side. The more significant oxygen flux enhancement by permeate side decoration indicates that the oxygen surface reaction in the permeate side was the rate limiting step. Furthermore, coating a LSC porous layer on both sides further increased the O2 flux of the GDC–LSM membrane, i.e., by 3 orders of magnitude at 800 1C compared with the values of the bare membrane, which can be attributed to the formation of electrical short circuit via the LSC layer and LSM.151 However, the LSCdecorated dual-phase membrane showed low stability under a CO2 atmosphere due to the decomposition of LSC to a Ruddlesden– Popper phase (i.e., (LaSr)2CoO4) and the reaction of CO2 with the segregated SrO.151 The oxygen permeation flux of the dual-phase

Chem. Soc. Rev.

View Article Online


Review Article

membrane made of GDC and LSM phases can be further enhanced by applying a hierarchically porous structure.152 A bilayer dualphase membrane with a dense Ce0.9Gd0.1O1.95 (GDC)– (La0.8Sr0.2)0.95MnO3%d (LSM) (60 : 40 vol%) layer and a hierarchically porous GDC–LSM support was fabricated by a combined freezedrying tape-casting and screen-printing method. The dense film layer was about 30 mm-thick to minimize the bulk diffusion resistance. A 870 mm-thick porous top layer with a porosity of 41.6% was present on top of the dense layer to enhance to the oxygen exchange rate on the feed side. The O2 flux of this bilayer membrane was improved by 81% relative to the 900 mm-thick dense GDC–LSM membrane. Such increment implied that the oxygen transport process was limited by the surface exchange reaction. By depositing a porous SDC–LSCF6428 layer on the permeate side, the oxygen permeation flux was enhanced by 600% relative to the GDC–LSM membrane without surface deposition, i.e., from 0.105 to 0.780 mL min%1 cm%2. After switching the sweep gas from He to CO2, an O2 permeation flux of 0.659 mL min%1 cm%2 could still be achieved.152 Park and Choi reported a similar observation that, using a LaCrO3 porous layer, the oxygen flux of yttria-stabilized zirconia (YSZ) was enhanced by B4 times with permeate-side surface modification and by B1.4 times with feed-side surface modification.233 Site-deficient Sr0.97Ti0.5Fe0.5O3%d, which has been reported to be highly stable against CO2,106 was decorated on the membrane surface. The oxygen flux was decreased by only 12.5% from 0.32 to 0.28 mL min%1 cm%2 over 93 hour-operation under pure CO2. The stability against CO2 was significantly enhanced although at the expense of oxygen flux. The chemical stability could be further improved using SFT–GDC (50 : 50 vol%) as the decoration material for which the O2 flux stabilized to 0.44 mL min%1 cm%2 after 93 h of operation in a CO2 atmosphere. This work provides an effective method to improve the O2 permeation flux and chemical stability of membranes under a CO2 atmosphere. Another related work is the Ce0.8Sm0.2O2%d (SDC)–PrBaCo2O5+d (PBC) (60 : 40 vol%) dual-phase membrane,200 based on which it was found that SDC–PBC (50 : 50 vol%) surface decoration can contribute towards 26% improvement of the O2 flux. A Ce0.8Sm0.2O2%d (SDC)–La0.9Sr0.1FeO3%d (LSF) (70 : 30 vol%) dual-phase membrane was also developed by Wang et al.153 They mentioned that cobalt-containing perovskites have structural instability while chromium within chromium-containing perovskites can easily evaporate at high temperatures. By utilizing LSF, the membrane showed excellent stability without any performance degradation at 900 1C with CO2 as the sweep gas for 150 hours. The ionic inter-diffusion between SDC and LSF was probed using inductively coupled plasma atomic emission spectrometry (ICPAES). The results showed that Sm, La, and Fe were co-doped into CeO2 in the dual-phase membrane. The tested CeO2 phase had a chemical composition of Ce0.774(Sm0.132La0.087Fe0.007)O2%d. Such codoped CeO2 was conceived to have enhanced conductivity which contributed positively towards the oxygen permeability of the dualphase membrane. Transition metal oxides such as Fe2O3 can also be used as an electronic conductor phase in dual-phase membranes. An O2 flux of 0.18 mL min%1 cm%2 could be achieved through a 0.5 mm-thick Fe2O3 (FO)–Ce0.9Gd0.1O2%d (GDC) (40 : 60) membrane

Chem. Soc. Rev.

Chem Soc Rev

under an air/He gradient at 1000 1C.155 After depositing a porous LSC layer on the surface of the feed side, a steady O2 flux of 0.2 mL min%1 cm%2 was detected for more than 150 h using pure CO2 as the sweep gas, indicating that the coated FO–GDC membrane was CO2 stable. 2.3.

Other materials

Another type of material which belongs to the K2NiF4 family also shows potential in terms of stability to CO2. La2NiO4+d is one of the main examples. The substantial interest in La2NiO4+d originates from its relatively low thermal expansion coefficient (11.9–13.8 ( 10%6 K%1) compared to perovskite oxides and its distinct crystal structure which consists of alternating perovskitelike and rock-salt-like layers along the c-axis (Fig. 12).40 At elevated temperatures, the oxygen transport through this compound mainly takes place via 2-dimensional interstitial oxygen diffusion in the rock-salt layers. La2NiO4+d, however, exhibits a much lower oxygen flux relative to the benchmark perovskite oxide material such as Ba0.5Sr0.5Co0.8Fe0.2O3%d. Klande et al. also found that the partial substitution of 10 mole% Ni in La2NiO4+d by various cations, i.e., Al, Co, Cu, Fe, Mg and Zr diminished the original oxygen flux (of nondoped materials).40 U-shaped K2NiF4 hollow fiber membranes based on the (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d material were

Fig. 12 Crystal structure of a typical K2NiF4 material, La2NiO4+d, consisting of alternating perovskite layers and rock-salt layers along the c-axis. The arrows indicate interstitial oxygen diffusion through the rock-salt layers. Reproduced with permission from ref. 40. Copyright 2011, Elsevier Inc.


View Article Online


Chem Soc Rev

synthesized by a phase-inversion spinning/sintering process.42,234 Ni, Cu, and Ga were incorporated into this composition to increase the CO2 resistance in accordance with the facts that NiCO3 is less stable than La2(CO3)3, La2O2CO3, and Pr2(CO3)3 (as deduced from the Ellingham diagram), CuCO3 is not stable above ambient temperature and Ga2(CO3)3 is rarely known to form (at least under the reported conditions). The fiber retained its original structure during in situ powder X-ray diffraction under a CO2 atmosphere from room temperature to 1000 1C. Compared with the (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d disk membrane, the O2 flux of this U-shaped hollow fiber was significantly higher (0.32 versus 0.9 mL min%1 cm%2 under pure CO2).235 By substituting Pr with Nd, the O2 flux was further enhanced under the same test conditions.236 Xue et al. found that the A-site deficient (Pr0.9La0.1)1.9(Ni0.74Cu0.21Ga0.05)O4+d exhibited an O2 flux of more than two times higher than (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d without any evidence of reduced CO2 resistance.237 Since Nd and Al are considered to be cheaper than Pr and Ga, (Nd0.9La0.1)2(Ni0.74Cu0.21Al0.05)O4+d was then developed which was also stable to pure CO2 up to 975 1C.238 Similarly, a sandwich structured (Pr0.9La0.1)2(Ni0.74Cu0.21Nb0.05)O4+d was also tested as a CO2 resistant oxygen permeation membrane.239 The performance details of the above materials are also listed in Table 3.40,42,235–238 2.4.

Engineering approaches

Besides the chemical and structural modification approach to enhance CO2 resistance, an engineering approach such as surface decoration can also be applied. Surface decoration has been frequently used in the solid oxide fuel cell (SOFC) field to enhance the cathode performance or stability in harsh atmospheres.240–244 In oxygen permeation membranes, the O2 flux is proportional to the inverse of the membrane thickness when the membrane is sufficiently thick so that the bulkdiffusion step limits the O2 transport according to Wagner’s equation.15 If the thickness is far below the characteristic membrane thickness (LC), the O2 flux becomes limited by the surface-exchange step.89 Therefore, LC can be used as an indicator to quantify the relative limiting step from bulkdiffusion, surface-exchange domination or both.245 Noteworthy is that LC is not a constant but dependent on the membrane material phase, morphology and operating conditions (temperature and gas atmosphere). Depositing a surface decoration layer can be utilized if surface-exchange dominates. Within the context of increasing the oxygen permeation flux by enhancing the surface exchange reaction rate, for example, La1%xSrxCoO3%d was often deposited as a porous layer on the dense membrane surface.138,141,145,151,158,246 A previous study demonstrated that the oxygen permeation flux of a 120 mm-thick BSCF5582 membrane with a 6.5 mm-thick porous layer was 3.6 times that of the BSCF5582 membrane without such a layer.247 Weight relaxation curves established higher surface exchange rate (kchem) in the presence of this porous layer coating; indicating that the larger reaction area of the porous coating promoted faster surface exchange. Other coating materials such as silver (Ag), platinum (Pt), Sm0.5Sr0.5CoO3%d, SDC, LaNiO4+d, LSCF6428, SrCo0.8Fe0.2O3%d, SrCo0.9Nb0.1O3%d, La0.5Sr0.5FeO3%d, and La1%xSrxFe1%yGayO3%d


Review Article

have also been applied as surface decoration on oxygen permeation membranes.129,131,219,220,248–255 The studies involving these aforementioned surface decorations were nonetheless focused on the oxygen permeation properties in a normal air to inert sweep gas gradient, i.e., in the absence of CO2. However, there are also studies of the effect of the surface activation layer on the CO2-resistance of membranes. Gaudillere et al. reported that the porous LSCF6428 activation layer on a hierarchically structured LSCF6428 membrane not only enhances the surface reaction rate, but also provides protection against CO2.62 Compared to the bare LSCF6428 membrane, the activated membranes displayed significantly lower O2 flux reduction when the originally inert sweep gas was changed to CO2.62 The negative effect of CO2 on the membrane performance was alleviated by the activation layer, especially at higher temperatures, i.e., Z900 1C. However, at 850 1C, the O2 flux drop became identical for the activated LSCF6428 and bare LSCF6428.62,256 More studies should be performed to reveal the mechanism behind this phenomenon. The activation layer also influences the CO2-resistance of the overall membrane, that is, the CO2-resistance depends on both the membrane and the activation layer composition. For example, the O2 flux of the Ce0.9Gd0.1O2%d–La0.7Sr0.3MnO3%d (80 : 20) dual phase membrane with a La0.6Sr0.4CoO3%d surface coating significantly decreased upon using a CO2 sweep gas.151 This was due to the instability of La0.6Sr0.4CoO3%d against CO2, which decomposed under CO2 exposure into firm SrCO3.151 However, when the activation layer was changed into CO2-resistant Sr0.97Ti0.5Fe0.5O3%d, the O2 flux remained stable for 100 h.151 Another study on the short-circuit GDC membrane showed the same effect.221 When BSCF5582 was applied as the activation layer on both feed and sweep sides, the O2 flux decreased with time when CO2-containing sweep gas was used. Such degradation behavior disappeared when silver was used as the sweep side decoration.221 In this case, although GDC is CO2-resistant, the unstable surface activation layer deteriorated the stability of the membrane. Still, surface decoration can be applied to design CO2-resistant membranes via proper choice of decoration materials that are stable to CO2. Development of CO2-protective layers on the surfaces of membranes is an effective route to obtain CO2-resistant membranes. A CO2-resistant layer can simply be coated onto the membrane surfaces. Zhang et al.’s work provides such an example.257 They made bi-layer structure Ba0.5Sr0.5Co0.8Fe0.2O3%d–Pr0.5Ce0.5O2%d (BSCF5582–PrCe) membranes by wet spraying PrCe ink (i.e., a premilled mixture of isopropyl alcohol, ethylene glycol, glycerol, and PrCe) onto a BSCF5582 membrane followed by calcination. The PrCe layer was CO2-resistant as confirmed by Fourier-Transform Infra-Red and CO2-temperature programmed desorption results. The oxygen permeation test under a CO2-containing atmosphere (10 vol% CO2 in He) showed a flux of around 1.35 mL min%1 cm%2 at 850 1C. At the same temperature, the flux was 1.6 mL min%1 cm%2 under pure helium. The retainment of a relatively large portion of oxygen flux even after CO2 was introduced to the sweep gas was most likely due to the weak adsorption of CO2 onto the PrCe surface. Another example is the coating of a SrFe0.8Nb0.2O3%d (SFN) nano-sized layer onto the BSCF5582 membrane.258 SFN has been proven as a

Chem. Soc. Rev.

Chem. Soc. Rev.

Disk (Pr0.9La0.1)1.9- EDTA– (Ni0.74Cu0.21- citric acid Ga0.05)O4+d

0.6

0.8

(Pr0.9La0.1)2- EDTA– Disk (Ni0.74Cu0.21- citric acid Ga0.05)O4+d

U-shaped (Pr0.9La0.1)2- EDTA– (Ni0.74Cu0.21- citric acid hollow Ga0.05)O4+d fiber

0.6

1

Disk (Nd0.9La0.1)2- EDTA– (Ni0.74Cu0.21- citric acid Al0.05)O4+d

EDTA– Disk citric acid

800– 0.36– 975 0.89 (He) 0.32– 0.88 (CO2)

850– 0.17– 975 0.34 (He) 0.15– 0.32 (CO2)

0.39 (CO2)

150 mL 30 mL 4–6 min%1 air min%1 CO2 150 mL 30 mL 6–7 min%1 air min%1 He 0.88 (CO2)

XRD of the spent membrane SEM of the spent membrane EDX of the spent membrane

Ref.

235

238

Excellent stability

237

No carbonate 42 formation and phase transition




Review Article

0.89 (He)

0.89 (He)

150 mL 30 mL 1–3 min%1 air min%1 CO2 30 mL 3–4 150 mL %1 min air min%1 He 0.88 (CO2)

3.417– 4.583

0.5– 3.417


180 mL 60 mL min%1 air min%1 CO2 60 mL 180 mL min%1 air min%1 He


0.94 (He)

0.9 (CO2)

30 mL 975 0.89 150 mL (He) min%1 air min%1 He

150 mL min%1 mL min%1 air

150 mL min%1 air

150 mL min%1 air

975 0.5 (He) 150 mL min%1 air

900 0.39 (He)

No carbonate 40 formation and phase transition

T (1C) Atmosphere Remarks

Other CO2 test

29 mL 0– In situ pow- 30– CO2 : air 150 mL 1000 (50 : 50 min%1 air min%1 He 6.667 der XRD vol%) 0.22 150 mL 29 mL 6.667– (CO2) min%1 air min%1 100 CO2 30 mL 975 0.5 (He) 150 mL 30 mL 0–100 In situ pow- 300– 20– min%1 He min%1 air min%1 He der XRD 950 100%CO2 (+He) 30 mL 150 mL 30 mL 0.4 100– min%1 min%1 air min%1 (CO2) 200 CO2 CO2 30 mL 200– 0.5 (He) 150 mL min%1 mL min%1 He 300 min%1 air 0.39 150 mL 30 mL 0–260 (CO2) min%1 air min%1 CO2 30 mL 975 0.32 150 mL 30 mL 0–230 XRD after CO2 %1 %1 min He (CO2) min air min%1 exposure to CO2 CO2 XRD of the 30 mL spent min%1 CO2 membrane 60 mL 0–0.5 In situ pow- RT- p(CO2) = 105 975 0.96 180 mL der XRD 1000 Pa (He) min%1 air min%1 He

Permeate Time Feed side side (h) Method

Long term CO2 test

J(O2) (mL Permeate T min%1 Feed side side (1C) cm%2)


J(O2) (mL ThickSynthesis T min%1 ness Area route Geometry (mm) (cm2) Coating (1C) cm%2)


CO2-resistance and performance of other membranes

La2NiO4+d

Materials

Table 3


View Article Online

Chem Soc Rev




0.6

150 mL min%1 air 150 mL min%1 air 150 mL min%1 air 150 mL min%1 air 150 mL min%1 air

0.48 (CO2) 0.53 (He) 0.48 (CO2) 0.53 (He) 0.48– 0.53 (CO2)



0.69 (He) 0.66 (CO2) 0.69 (He) 975 0.53 (He)


900 0.69 (He) 0.66 (CO2)

0–1

6–7

4–6

3–4

1–3

0–1

XRD of the spent membrane 30 mL 1–2.5 SEM of the %1 min spent CO2 membrane 30 mL 2.5–4 XRD after min%1 He exposure to CO2 30 mL 4–5.5 min%1 CO2 30 mL 5.5– min%1 He 6.5 30 mL 0–420 min%1 CO2

30 mL min%1 He 30 mL min%1 CO2 30 mL min%1 He 30 mL min%1 CO2 30 mL min%1 He 30 mL min%1 He

Permeate Time Feed side side (h) Method

Long term CO2 test

J(O2) (mL Permeate T min%1 Feed side side (1C) cm%2)


J(O2) (mL ThickSynthesis T min%1 ness Area route Geometry (mm) (cm2) Coating (1C) cm%2)

(continued)

(Nd0.9La0.1)2- EDTA– Disk (Ni0.74Cu0.21- citric acid Ga0.05)O4+d

Materials

Table 3


Slight change


T (1C) Atmosphere Remarks

Other CO2 test


236

Ref.

View Article Online


Chem. Soc. Rev.

View Article Online


Review Article

perovskite that has high chemical stability against CO2.67 The resultant membrane maintained stable oxygen flux in a CO2containing atmosphere at 850 1C. The spray pyrolysis method can also be used to make dense CO2-resistant layers.259 Spray pyrolysis is considered as a scalable and low-cost process. Furthermore, the morphology and chemical composition can be controlled via appropriate selection of solvents, precursor concentrations, droplet sizes, and deposition temperatures.259,260 The 250 nm-thick GDC protective layer on the permeate side prevented the formation of strontium carbonate on the LSCF6428 membrane; thus the O2 flux of the protected LSCF6428 showed no decrease when CO2 was introduced in the sweep gas. He et al. fabricated a sandwich structure LSCF6428 membrane containing a dense LSCF6428 layer in the center and two porous LSCF6428 layers on the opposite surfaces.261 GDC particles were coated onto the porous LSCF6428 layers by impregnating the membrane with a gadolinium nitrate (Gd(NO3)3) and cerium nitrate (Ce(NO3)3) solution mixture. After being heated at 900 1C for 1 hour to remove the nitrate ions and organics, a fluorite phase was formed. Their powder X-ray diffraction patterns confirmed the formation of GDC on the LSCF6428 surface. GDC particle modification simultaneously improved the surface exchange kinetics and enhanced the chemical resistance of LSCF6428 membranes as evidenced by the long-term stability under a CO2-containing atmosphere for 100 hours. Using the same method, a sandwich structured (Pr0.9La0.1)2(Ni0.74Cu0.21Nb0.05)O4+d (PLNCN) was modified by GDC particles on the porous layer.239 By doing such GDC modification, the O2 flux of the ultrathin (5–6 mm) PLNCN membrane reached up to 4.22 mL min%1 cm%2 at 950 1C with 100 mL min%1 He sweep gas flow. During the subsequent 200 hour-test under pure CO2 at 900 1C, this membrane demonstrated stable oxygen permeation fluxes of 3.22 mL min%1 cm%2. Unlike the dense GDC layer reported in the work of Garcıá-Torregrosa et al.,259 the GDC particles on the porous layers contributed to the higher surface exchange rate and the resultant higher oxygen permeation flux. Likewise, Zhang et al. also demonstrated this concept by coating a 100 nm-thick SDC dense layer onto the BSCF5582 surface.262 With such a CO2-resistant SDC dense layer, the membrane displayed a stable oxygen flux for at least 60 hours when 10 vol% CO2 + 90 vol% He was used as the sweep gas, in contrast to the BSCF5582 membrane without the coating that showed a continuous decrease in CO2-containing sweep gas. In addition, the protective layer can also be made of the same material as the dense layer. A research study on the SrFe0.95W0.05O3%d membrane, for example, was carried out to study the effect of surface decoration on the stability of the membrane in a CO2 atmosphere.263 The test results indicated enhancement of CO2 stability for the doped SrFe0.95W0.05O3%d by decoration of the protective layer. Moreover, by decorating the porous layer and silver particles on the permeate side, the oxygen flux under helium was increased due to the enhanced oxygen desorption process. Furthermore, the protective porous layer obstructed the circulation of CO2 in pores, preventing the complete blockage of the surface by strontium carbonate. This factor contributed to minimizing the degradation rate of oxygen fluxes. The performance details of the above engineering approach can be found in Table 4.239,257–259,261–263

Chem. Soc. Rev.

Chem Soc Rev

Several approaches have been developed to fabricate a thin film layer to achieve a composite membrane (asymmetric membrane) structure in SOFC and oxygen permeation applications. These methods include chemical vapor deposition (CVD),264 physical vapor deposition (PVD),265 dip-coating,186,266 slip-casting,265,267 screen-printing,62,73,152 spray pyrolysis,257,259 pulsed laser deposition,258 and spin-coating.268 These methods can also be applied to fabricate protection layers as discussed above. The protection layers should be made from CO2-resistant materials and should be gas-tight. In a typical CVD process, volatile precursors are injected into a chamber to react or decompose on the substrate membrane and produce the desired layers. For example, Ce(DPM)4 and Gd(DPM)3 were employed to prepare a GDC layer through CVD,264 while in a PVD process, pure GDC or LSCF6428 were directly deposited onto the substrate membrane via sputtering, plasma-spraying or electron beam evaporation.265,269,270 Dip-coating is a conventional technique to manufacture dense layers on a substrate membrane. Meng et al. used a dip-coating process to fabricate a dense BaCo0.7Fe0.2Nb0.1O3%d (BCFN) layer on a porous support.266 BCFN powders were firstly dispersed in deionized water and ball-milled for 10 h to form a BCFN suspension slurry. Then, the porous BCFN support was soaked into the slurry for 2 min, followed by drying and cosintering to obtain the final asymmetric membrane. The thickness of the dip-coated dense layer was controlled by adjusting the slurry concentration. Instead of using ceramic powder to prepare the precursor slurry, metal nitrates with polymer and organic solvent can be utilized.262,268 Spin-coating is an alternative to dip-coating. In the spin-coating process, the precursor slurry was dropped onto the substrate membrane that is fixed on a spin disk in a spin coater. The spinning disk was then rotated at high speed to spread the slurry homogeneously on the substrate.268 The layer thickness can be controlled by adjusting the spinning time. Instead of using a spin coater, the coating process can also be achieved by an air-blast spray gun, which is referred to as the spray pyrolysis process.257,259 For instance, in one report, the PrCe slurry (i.e., mixture of PrCe powder, water, isopropyl alcohol, ethylene glycol, and glycerol) was wet sprayed onto the surface of the substrate disk using a spray gun to form a protection layer.257 Gaudillere et al. coated a 30 mm-thick dense LSCF6428 layer using a screen-printing method.73 The slurry for screen-printing was obtained by mixing the LSCF powder with a binder (6 wt% ethylcellulose in terpineol) in a 50 : 50 weight ratio. Furthermore, by adding pore former into the slurry, a porous layer can also obtained via screen-printing.62 Pulsed laser deposition was used by Zhang et al. to deposit a CO2-resistant SrFe0.8Nb0.2O3%d layer on a BSCF5582 membrane.258 The thickness of the CO2-resistant layer was controlled to B45 nm using this method. 2.5.

Patents and industrial activities

The previous discussions on CO2-resistant MIEC membranes mainly feature studies from journal articles and conference papers. As a promising technology, its industrial development should also be discussed. Praxair Technology Inc. and Air



1

Ba0.5Sr0.5EDTA–citric Disk Co0.8Fe0.2O3%d acid

La0.6Sr0.4Fe0.8- Commercial Disk Co0.2O3%d 0.8

Feed side

Permeate side

Gd0.1650– 0.018– 60 mL Ce0.9O2%d 1000 0.59 (Ar) min%1 air 0.013– 60 mL 0.62 (Ar min%1 + CO2) air

9.75 mL min%1 CO2 + 55.25 mL min%1 Ar

65 mL min%1 Ar

Gd0.1750– 1.03– Ambient 100 mL Ce0.9O2%d 950 4.22 (He) air min%1 mL min%1 He

0.45 Sm0.2500– 0.25–3.7 Ambient 100 mL Ce0.8O1.9 900 (He) air min%1 He

Ambient 100 mL air min%1 He

Pr0.5700– 0.26–2.1 Ambient 100 mL Ce0.5O2%d 900 (He) air min%1 He

0.57 SrFe0.8750– 1.303– Nb0.2O3%d 900 2.721 (He)

EDTA–citric Sandwich 0.005 0.5 acid

1

EDTA–citric Disk Ba0.5Sr0.5Co0.8Fe0.2O3%d acid

(Pr0.9La0.1)2(Ni0.74Cu0.21Nb0.05)O4+d

1

Synthesis route

850 CO2

3.31 (He) 3.22 (CO2)

3.22 (CO2)

Ambient 100 mL min%1 7–60 XRD of the 900 CO2 air CO2 spent membrane Ambient 100 mL min%1 60–85 He air Ambient 100 mL min%1 85– air mL min%1 200 CO2

Ref. 257

Good chemical stability GdCu2O4 formed

259

239

262

SrFe0.8Nb0.2- 258 O3%dCO2 stable

Pr0.5Ce0.5O2%dCO2 stable

T Atmo(1C) sphere Remarks

Other CO2 test

1.35 (He Ambient 90 mL min%1 3–15 + CO2) air He + 10 mL min%1 CO2 1.6 (He) Ambient 100 mL min%1 15–18 air He 850 2.6 (He) Ambient 100 mL min%1 0–6 XRD, CO2- 850 CO2 air He TPD and FTIR %1 2.25 (He Ambient 90 mL min 6–15 + CO2) air He + 10 mL min%1 CO2 850 2.8 (He) Ambient 100 mL min%1 0–2.5 air He 2.5 (He Ambient 90 mL min%1 2.5– He + 10 mL 62.5 + CO2) air min%1 CO2 2.8 (He) Ambient 100 mL min%1 62.5– air He 80 900 3.31 Ambient 100 mL min%1 0–7 TG RT- CO2 (He) air He 1000

FTIR and CO2-TPD

Time Permeate side (h) Method

850 1.6 (He) Ambient 100 mL min%1 0–3 air He

Feed side

J(O2) (mL T min%1 (1C) cm%2)

J (O2) (mL T min%1 (1C) cm%2)

Thickness Area Geometry (mm) (cm2) Coating

Ba0.5Sr0.5EDTA–citric Disk Co0.8Fe0.2O3%d acid

Materials

Long term CO2 test



Table 4 CO2-resistance and performance of engineering approaches


View Article Online


Chem. Soc. Rev.

Chem. Soc. Rev.

Synthesis route

Permeate side

SrFe0.95W0.05O3%d

Solid-state reaction

1.48

1.48

Disk

Disk

SFW

850 0.07 (He) 200 mL min%1 air 200 mL 0.035 min%1 (He + air CO2) 200 mL 0.024 min%1 (He + air CO2) 0.0125 200 mL min%1 (He + air CO2) 200 mL 0.011 min%1 (He + air CO2) 0.0025 200 mL min%1 (CO2) air 850 0.21 (He) 200 mL min%1 air 200 mL 0.152 min%1 (He + air CO2) 0.125 200 mL (He + min%1 air CO2) 0.11 (He 200 mL + CO2) min%1 air 0.085 200 mL (He + min%1 air CO2) 200 mL 0.08 min%1 (CO2) air

Ambient air Ambient air

100 mL min%1 He 95 mL min%1 He + 5 mL min%1 CO2 Ambient 100 mL min%1 air He 105– 120

5–105

0–5


T Atmo(1C) sphere Remarks

Other CO2 test

Ref.

263

261

Review Article

4 mL min%1 CO2 + 16 mL min%1 He 8 mL min%1 CO2 + 12 mL min%1 He 12 mL min%1 CO2 + 8 mL min%1 He 16 mL min%1 CO2 + 4 mL min%1 He 20 mL min%1 CO2

20 mL min%1 He


20 mL min%1 He

3.45 (He)

900 3.45 (He) 3.08 (He + CO2)

Feed side

J(O2) (mL T min%1 (1C) cm%2)

J (O2) (mL T min%1 (1C) cm%2)

Thickness Area Geometry (mm) (cm2) Coating Feed side

Long term CO2 test



La0.6Sr0.4EDTA–citric Sandwich 0.004 0.45 Gd0.1700– 0.72– Ambient 100 mL Co0.2Fe0.8O3%d acid Ce0.9O2%d 900 3.51 (He) air min%1 He

Materials

Table 4 (continued)


View Article Online

Chem Soc Rev


View Article Online

Review Article

T Atmo(1C) sphere Remarks J(O2) (mL T min%1 (1C) cm%2) J (O2) (mL T min%1 (1C) cm%2) Thickness Area Geometry (mm) (cm2) Coating


20 mL min He


0.067 (CO2)

0.08 (He + CO2)

0.09 (He + CO2)

0.11 (He + CO2)

0.13 (He + CO2)

200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air 200 mL min%1 air SFW and 850 0.172 Ag (He) 1.48 Disk

Permeate side Feed side Materials

Synthesis route

Table 4 (continued)

%1

Feed side

Long term CO2 test Instantaneous performance Membrane architecture



Other CO2 test

Ref.

Chem Soc Rev

Products and Chemicals Inc. are two major companies that have allocated substantial investments in this field. The exploration of this technology for industrial application is dated back to the 1990s, when Burggraaf and Lin from Air Products and Chemicals Inc. developed a method to fabricate ultrathin (r0.5 mm) YSZ membranes for oxygen separation.271 Their ultrathin membrane demonstrated oxygen fluxes in the order of 10%2 mL min%1 cm%2. Praxair Technology Inc. designed an oxygen separation system to produce oxygen from air via two stages, both of which included an electrically energized oxygen permeation process through ionic or mixed conducting membranes.272 Prior to this patent, Chen and Cook from Air Products and Chemicals Inc. also developed electrically driven solid–electrolyte membranes to separate oxygen from crude argon.273 Following these early-stage explorations of ionic or mixed conducting membranes for oxygen production,274–281 or oxygen recovery from exhaust gas,282 hundreds of patents have been granted on this topic. Here, we choose several representative results for a brief discussion, especially those involving CO2-resistance. Patents from Air Products and Chemicals Inc. and Praxair Technology Inc. can be categorized into three applications, i.e., oxygen production, oxygen production integrated with power generation, and synthesis gas production. In the first category, a typical oxygen production process can be described as follows: an oxygen-containing gas mixture is compressed prior to entering a heating system; the compressed and heated gas mixture is then passed into a membrane separation unit; and the high-purity oxygen stream and the oxygen depleted stream are finally withdrawn separately from the membrane separation unit.272–275,280,283–287 The membrane separation unit becomes the key unit in the process. In the patents from Air Products and Chemicals Inc., they usually applied LaxA1%xCoyFe1%yO3%d as the oxygen ionic conducting membrane, where A is chosen from barium, strontium, and calcium. Praxair Technology Inc. also utilized mixed conducting ceramic membranes for oxygen production.272,278–280,288–290 However, in their patents, they tried a dozen different materials including La1%xSrxCo1%yFeyO3%d, SrMn1%xCoxO3%d, BaFe0.5Co0.5YO3%d, La0.2Ba0.8Co0.8Fe0.2O3%d, CaTi0.95Mg0.05O3%d, and YSZ/ noble metal. Besides oxygen production,272,280 Praxair Technology Inc. also explored the combination of an oxygen production unit with a steam stream generator to produce an O2–steam mixture stream for coal gasification.288,289 Moreover, unlike studies in journal articles where single disk membranes or hollow fibers dominate, in these patents, membranes were designed into various configurations, such as multi-layer membranes,274 planar modules,285 and tubular modules.286,291 The oxygen separation process is not accomplished as simply as in the lab experiment. It involves multi-separation stages.272,275 For example, in a twostage oxygen permeation process reported by Praxair Technology Inc,272 oxygen-containing gas is introduced to the first gas chamber and permeates through the first membrane to the second gas chamber. Then, the oxygen-depleted gas from the first chamber is delivered to the third gas chamber to permeate again through the second membrane to the fourth gas chamber. The oxygen-enriched gas from the second and fourth gas chamber is then collected for further use. As CO2 is inevitably present in the industrial gas

Chem. Soc. Rev.

View Article Online


Review Article

atmosphere, how to restore the membrane performance after detrimental interaction with CO2, H2O, or volatile hydrocarbons becomes an important issue. For example, Air Products and Chemicals Inc. developed a continuous process to restore the permeance of an ionic transport La0.2Ba0.8Co0.8Fe0.2O2.6 membrane that was degraded by adverse reaction with CO2 and H2O.292 They restored the membrane performance by increasing the process temperature to above 810 1C, which was greater than the oxygen separation temperature of 650–800 1C. An identical temperature range was also applied in another patent, where the performance of La0.2Ba0.8Co0.8Fe0.2O3%d degraded at 783 1C in a CO2-containing atmosphere.293 In another patent, a series of LaxSrx0 CoyFey0 Cuy00 O3%d materials were tested in a CO2-containing atmosphere. They found that A-site deficient (or called B-site excess) La0.2Sr0.8Co0.41Fe0.41Cu0.2O3%d displayed a higher CO2-resistance than A-site excess La0.2Sr0.8Co0.41Fe0.41Cu0.2O3%d.103 Oxygen separation using oxygen ionic conducting membranes should be operated at high temperatures above 700 1C. Thus, excess, non-utilized heat energy is available in the permeated and non-permeated gas stream. This redundant energy can be effectively utilized by compressors, combustors, hot gas turbines, steam turbines, and heat exchangers. Several patents have been granted on these applications.276,277,294–296 A typical design consists of adding an expansion turbine through which the hot non-permeate gas passes to generate shaft power to drive an electric power generator. The hot exhaust gas can be optionally cooled down by heat exchange with water to generate steam to operate additional electric power generator. The oxygen ionic conducting membranes integrated in such a process were also chosen from the previously mentioned LaxA1%xCoyFe1%yO3%d. Praxair Technology Inc. added extra fuel combustors in the oxygen separation unit to combust the oxygen-depleted gas with fuels to produce more power.290 With the presence of such a combustor, CO2 can be obtained by removing water from the combustion products.290 Several other patents reported different designs of the combination of a combustion chamber and an oxygen separation module.295,297–299 The membranes included in these modules were also selected from the above mentioned materials for oxygen separation application in Praxair Technology Inc. The third application area is synthesis gas production. Synthesis gas containing hydrogen and carbon monoxide is an important feedstock for the production of liquid hydrocarbons and oxygenated organic compounds. In the traditional commercial synthesis gas process that applies an autothermal reforming step, oxygen supplied for this process was obtained from low-temperature air distillation or pressure swing adsorption. An alternative oxygen production technology is mixed using ceramic membrane technology. Patents from Air Product and Chemicals Inc. and Praxair Technology Inc. were granted on this technology.291,300–307 In such synthesis gas production systems, a reaction zone was provided that contains an oxidant side and a reactant side separated by a ceramic mixed conducting membrane. Then, a heated oxygen-containing oxidant gas and a methane-containing reactant gas were fed into the oxidant side and reactant side, respectively. Oxygen from the

Chem. Soc. Rev.

Chem Soc Rev

oxidant side permeated through the membrane to react with methane, producing the product gas that contains hydrogen and carbon monoxide. This is a typical membrane reactor design to combine separation and reaction in one unit. The prerequisite is that the membrane applied in such a system should be CO2-, CH4-, H2-, and CO-resistant. In one patent, Air Products and Chemicals Inc. developed (La0.8Ca0.2)1.01FeO3%d and (La0.85Ca0.15)1.01FeO3%d membranes for synthesis gas production.303 Both membranes remained stable for at least 200 h of operation. Such a membrane reactor system had less than half the power requirement relative to the conventional partial oxidation process.300 Planar ceramic membranes and staged membrane reaction systems were also further developed to improve the production line.308–310 It should be noted that several patents from Praxair Technology Inc. exist, which feature composite ceramic membranes.268,281,311–317 The composite membrane consisted of a thin dense matrix layer and at least one porous coating or support layer. Thin dense films can produce higher oxygen fluxes, reduced membrane area, and lower operating temperatures (relative to the thick ones).311 In one patent, composite ceramic membranes were fabricated by depositing a thin film on a porous support layer.268 For instance, a B0.5 mm-thick porous La0.05Sr0.95CoO3 support was fabricated by mixing La0.05Sr0.95CoO3 powder with 20 wt% Ag powder, followed by pressing in a die under a pressure of 10.4 kpsi and sintering at 950 1C. A dense La0.05Sr0.95CoO3 thin film was deposited on a porous support by spin-coating for several or dozens of cycles. The oxygen permeation test showed more than five times flux enhancement for the 2 mm-thick thin film with a porous support compared to the 1 mm-thick dense membrane. In addition to spin coating, other coating methods such as dip coating, spray coating, slip coating, and thermal/ plasma spray were also deployed in the invention. The support layer can also be formed by inert materials which do not transport oxygen ions, e.g., metal and metallic alloys.316,317 However, the metal support layer exhibits creep at high temperatures and failure due to oxidation. Another problem is the diffusion resistance caused by the interconnected pores in the porous support layer. Increasing porosity to higher than 40% was proposed as a solution, but such high porosity made the support weak and brittle. Cylindrical pores or conical pores as disclosed in US Pat. 7,229,537 B2 can solve this concern.311 In this patent, the support layer was fabricated from an oxidedispersion strengthened metal alloy, metal-reinforced intermetallic alloys, or ceramic materials such as YSZ.311 In one example of this patent, the dense layer was made from the mixture of 40 wt% LSCF, 40 wt% GDC, and 20 wt% silver. With MA956 as the support layer, the resultant membrane showed an excellent stability for permeation under 90% CO/10% CO2, and 85% H2/15% CO2 gas mixtures for over 2500 h of operation. In addition to the support layer, an extra active layer on the dense thin film can further enhance the oxygen permeability of composite membranes. Virkar et al. from Praxair Technology Inc. discussed the effect of distribution of pore radii of the porous active layer on the oxygen flux.281 The active layer can be made from a mixture of a mixed conductor and an ionic


View Article Online


Chem Soc Rev

conductor, e.g., GDC/LSCF mixture.315 In some cases, an intermediate porous layer can also be added between the dense layer and the support layer.311–314,318 Such an intermediate layer can be made from the same materials as the dense layer and acts as a distributor of permeated oxygen and/or reactant. Several patents from Air Products and Chemicals Inc. disclosed composite membranes as well.274,284 While oxygen-selective MIEC membranes showed great potential for oxygen production, oxyfuel combustion, clean coal energy delivery, catalytic membrane reactors, and solid oxide fuel cells, the greatest potential for commercialization in the future appears to lie in the oxygen production application.319 Air Products and Chemicals Inc. has played an important role in the commercialization of MIEC for oxygen production. The company started research and development in the ionic transport membrane (ITM) project from 1988. They currently hold more than 90 US patents in this field. From 1998 to 2008, the US Department of Energy and Air Products and Chemicals Inc. invested over $150 million in ITM technology.320,321 On collaboration with other companies and universities, they performed a multiphase plan on the commercialization of MIEC membranes for oxygen production. Phases 1 and 2 were completed in 2001 and 2006, respectively, achieving the goal of producing 5 tons O2 per day for over 1000 days. The plan is currently in phases 3–5 simultaneously. Phase 3 is designed to achieve an O2 production capability of 100–150 tons per day. The goal of phase 4 is to develop MIEC membrane reactors for syngas production, while phase 5 is aiming to supply 2000 tons of pure O2 per day via wafer-like planar ceramic membrane modules.322 Phase 4 shows potential for 30% reduced capital cost in the synthesis gas generation process and over 20% capital cost saving for the overall synthesis gas/H2/CO2 production process.322 In addition to Air Products and Chemicals Inc., Aachen University also made a pilot-plant investigation into the combination of MIEC technology with oxyfuel for power generation and CO2 capture.323,324 They built a large pilot module with an oxygen production capacity of 0.61 tons per day. More details on these industrial activities can be found in Zhu and Yang’s book.319

3. Performance overview Single-phase perovskite-based membranes, especially barium and cobalt containing perovskites, generally have higher O2 permeation flux than dual phase membranes. In Table 1, most of Ba and Co-containing perovskite compositions display fluxes higher than 1.0 mL min%1 cm%2 in an inert gas-containing sweep gas atmosphere. The highest O2 flux was exhibited by a BaBi0.05Co0.8Ta0.15O3%d membrane.72 Its O2 flux could reach up to 3.57 mL min%1 cm%2 for a 0.8 mm-thick BaBi0.05Co0.8Ta0.15O3%d membrane at 950 1C under an air/He oxygen gradient. BaCoxFeyNb(Zr)O3%d perovskites also displayed a high O2 flux of around 1 mL min%1 cm%2 for a 1 mm-thick disk membrane at 900 1C using He as the sweep gas.30,39,70 O2 flux decreased to as low as 0.12 mL min%1 cm%2 at 900 1C for the 0.85 mm thick


Review Article

BaFe0.65Nb0.35O3%d membrane at 900 1C using He as the sweep gas, which highlights the fact that the high redox activity of Co promotes high O2 permeability.30 However, when pure CO2 was used as the sweep gas, O2 flux degraded to zero due to the formation of BaCO3 that fully covered the membrane surface, especially at lower temperatures.30,325 Complete replacement of Ba by Sr or La in addition to reduction of Co content effectively enhanced CO2 resistance. For example, the SrFe0.8Nb0.2O3%d membrane could maintain 0.31 mL min%1 cm%2 O2 flux at 900 1C for at least 210 hours using pure CO2 as the sweep gas67 whereas for the BaCo0.6Fe0.2Nb0.2O3%d membrane under the same conditions, the O2 flux was reduced to an almost negligible value.30 Likewise, LSCF6428 and Pr0.6Sr0.4Co0.2Fe0.8O3%d membranes could maintain a stable O2 flux for more than 90 hours while for the BSCF6428 membrane, the O2 flux started to degrade immediately after CO2 was introduced.27,62,75 Although appropriate selection of A and B-site cation components can improve CO2 resistance, performance deterioration is inevitable in pure CO2 atmospheres.53,60,124 An attractive alternative to perovskite that has been widely studied is K2NiF4based materials. Examples of these materials are La2NiO4+d and doped La2NiO4+d such as (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d and (Nd0.9La0.1)2(Ni0.74Cu0.21Al0.05)O4+d membranes. They have been proven to be completely CO2-resistant during long-term operation in a CO2 atmosphere.40,42,235–238 K2NiF4-based materials have better stability but lower O2 permeability than perovskite materials. For example, the O2 flux of a 0.66 mm-thick (Pr0.9La0.1)1.9(Ni0.74Cu0.21Ga0.05)O4+d membrane was only 0.66 mL min%1 cm%2 at 900 1C under an air/CO2 oxygen gradient.237 Dual-phase membranes generally have better stability than perovskite membranes. Table 2(a) and (b) summarize the performances of CO2-resistant dual-phase membranes. After CO2 exposure, these membranes did not show evidence of phase change and carbonate formation. These dual-phase membranes normally displayed less than 1 mL min%1 cm%2 at 950 1C when CO2 was used as the sweep gas. The main reason for the low O2 flux is the large amount of fluorite phase such as Ce0.8Tb0.2O2%d, Ce0.9Pr0.1O2%d, Ce0.8Sm0.2O2%d, Ce0.9Gd0.1O2%d, Y0.08Zr0.92O2%d, and Ce0.8Nd0.2O2%d; all of which have very low O2 permeability due to the limited electronic conductivity. It is worth repeating that when an electronic conducting phase with low or negligible ionic conductivity is used, the oxygen ionic conducting path is blocked. This leads to lower oxygen flux relative to the case using the mixed ionic–electronic conducting phase. To take NiFeO4 (NF)–Ce0.8Tb0.2O2%d (CT) (60 : 40 vol%) as an example on which a NF electronic conductor was used, its O2 flux at 900 1C swept by CO2 was only 0.14 mL min%1 cm%2.132 In the Ce0.9Pr0.1O2%d (CP)– Pr0.6Sr0.4Fe0.5Co0.5O3%d (PSFC) (60 : 40 wt%) case which includes a MIEC PSFC phase, on the other hand, its O2 flux was 0.7 mL min%1 cm%2 at 950 1C.142 Such a value is among the highest fluxes observed for dual-phase membranes.

4. Forward directions In retrospect, there is an obvious trade-off between CO2 resistance and O2 permeation flux which makes it difficult to

Chem. Soc. Rev.

View Article Online


Review Article

achieve simultaneously high CO2 resistance and high O2 permeation flux within a single-phase material. To this end, it is more realistic to enhance the O2 permeation flux of CO2resistant materials with low flux (rather than the other way around, i.e., enhancing the CO2 resistance of unstable materials with high flux) via the application of internal and/or external short-circuit concepts and/or engineering approaches, resulting in what we called ‘‘hybrid material-engineering optimized membranes’’. Another equally important factor that cannot be overlooked is that such optimized membranes should have low material and production costs to realize their commercialization. Despite the major improvements demonstrated by numerous studies within the last decade, there are still many avenues and opportunities to be pursued. The two most promising major directions can be delineated as follows: (1) New membrane materials and configurations with high CO2 resistance and high O2 permeation flux This direction covers the development of new perovskite and K2NiF4-type compositions as well as new dual-phase combinations. For perovskite materials, the incorporation of Ba in the A-site and Co in the B-site almost always provides high oxygen permeability such as in the case of BaBi0.05Co0.8Ta0.15O3%d and BaCoxFeyNb(Zr)O3%d.30,39,70,72 The longest stability in CO2 nonetheless has thus far been obtained using perovskites that do not contain Ba and Co as represented by the SrFe0.8Nb0.2O3%d hollow fiber which was stable for 600 hours in a CO2 atmosphere.68 Caro’s group has published extensively on new CO2-resistant dual-phase combinations.66,138,139,142,145,149,150,154,160 A survey of the existing literature highlighted the three most promising phase combinations, i.e., fluorite–perovskite, fluorite–spinel, and fluorite–metal oxide. Despite their lower fluxes relative to perovskite membranes, rapid advances in the oxygen permeability of dual-phase membranes over the past half-decade means that attaining an economic flux of above 1 mL min%1 cm%2 via such composite technology in the near future is almost a certainty.326 Modification of the membrane structure and configuration can also be performed to further improve the oxygen permeability. This is exemplified by the formation of a thin dual-phase membrane with a porous coating and a bilayer dual-phase membrane with a hierarchically porous structure.151,152 The amount of studies addressing membrane configuration is relatively sparse in comparison to those focusing on developing new perovskite compositions. There are still many opportunities ahead as well as a large improvement margin associated with the former aspect. (2) Beneficial phase reactions and conductive phase generation along grain boundaries in dual-phase membranes An innovative direction emerged recently that exploits beneficial ‘‘in situ’’ phase reactions on certain dual-phase combinations to create conductive new phases along the grain boundaries that enhance the oxygen permeability of the resultant membrane (relative to the mechanically mixed dual-phase mixture). Traditionally, chemical compatibility between the ionic conducting phase and the electronic conducting phase was the main consideration to ensure that these two phases did not react to form new phases that have low ionic and/or electronic conductivities. Such non-desirable

Chem. Soc. Rev.

Chem Soc Rev

reactions, for example, occurred in the case of the GDC–LSM composite which displayed a decreasing oxygen permeation flux with time.188 Therefore, phases with low kinetic stability, i.e., high cation diffusion tendency generally are the last choice for the composite component. The phase reactions nonetheless can also create new phases that have higher ionic and electronic conductivities (relative to the original individual phases). Lin et al., for example, detected such new phase formation in a Ce0.8Gd0.2O2%d–CoFe2O4 dual-phase membrane using scanning transmission electron microscopy-high angle annular dark-field (STEM-HAADF), STEM-energy dispersive X-ray spectroscopy (STEM-EDX), STEM-electron energy loss spectroscopy (STEMEELS), and selected area electron diffraction (SAED).196 Gdand Fe-rich phases containing Gd, Fe, Ce, Co, and O (denoted as the GFCCO phase) formed initially at the GDC–GDC boundaries. This GFCCO phase had a perovskite structure and high concentration of oxygen vacancies that promotes the oxygen ionic conductivity. The in situ production of the GFCCO phase also decreased Gd accumulation at GDC–GDC boundaries which facilitates the oxygen transport within these boundaries. The work of Zhang et al. represents another example of this direction.197 SDC, SrCO3, and Co3O4 were used as the precursors to prepare dual-phase membranes. A new tetragonal perovskite phase, i.e., SmxSr1%xCoO3%d formed during sintering. This phase contributed towards higher permeability for the sintered membrane. Despite the apparent potential this direction provided, there are no clear guidelines on the design of the new phase. This is associated with the difficulties to predict what reactions will occur as well as the formed phase properties. Additional potential offered by this direction is the modification of the mechanical strength. Future studies that evaluate systematically the mechanical strength before and after beneficial phase reactions as well as the origins behind the change are warranted given their current absence. Breakthrough in the performance of dual-phase membranes is likely to come from pursuing this direction which also may provide insights into ionic and electronic transport processes around the grain boundaries.

5. Degradation behavior and mechanism probing The simultaneous use of several spectroscopy techniques to probe the degradation processes during membrane exposure to CO2 is a powerful method as demonstrated by Yi et al.30 In their work on BaCo0.4Fe0.4Nb0.2O3%d (BCFN), powder X-ray diffraction (XRD), scanning electron microscopy (SEM, including back scattered electron (BSE) imaging), energy dispersive X-ray (EDX) analysis (including elemental mapping) and scanning transmission electron microscopy (STEM, including electron energy loss spectroscopy (EELS), high-angle annular dark field (HAADF) imaging, and selected area electron diffraction (SAED) pattern) were combined. Powder XRD for BCFN annealed at 900 1C in CO2 showed the formation of BaCO3 and CoO. SEM and STEM (Fig. 13(a)) revealed the formation of a top carbonate


View Article Online


Chem Soc Rev

layer mainly consisting of BaCO3 followed by a porous layer containing a BSCF5582 layer which was depleted of Co and enriched in Fe and Nb as well as CoO on top of the unaffected bulk perovskite layer. SAED (Fig. 13(a)) showed major reflections from the main cubic phase of Co-depleted BCFN phase together with minor reflections from BaCO3 (indicated as b) and CoO (indicated as c). The gold particles which were deposited on the surface of BCFN prior to CO2 exposure became embedded between the carbonate layer and the porous layer which indicates the outward growth of BaCO3 at the surface (Fig. 13(b)). BSE imaging and elemental mapping of BCFN annealed at 900 1C in CO2 for 10 days (Fig. 13(c) and (d)) confirmed that the previous morphology was retained (with larger thickness for each layer) at longer exposure time. The variation of the cation composition with the depth of the membrane (from the boundary of the carbonate layer and porous layer to the bulk) was obtained using SEM-EDX (Fig. 13(e)). Despite the variation in Co, Fe and Nb, the ratio between A and B cations remained constant. The authors also found that the thickness of both layers, i.e., carbonate and porous increased with CO2 annealing time following a simple parabolic rate law (Fig. 13(f)) which suggests that the process is a diffusion-controlled one. They proposed that during the decomposition, initially BaCO3 formed on the surface (and grew outward) while the layer below became depleted in Ba due to diffusion of Ba2+ from the bulk layer to the surface. This resulted in the precipitation of CoO to form the Co-depleted porous layer. At the outer surface, Ba2+ reacted with CO2 and O2

Review Article

in the presence of electrons to form BaCO3 (instead of reacting within the porous layer which requires O2 and CO2 to diffuse into the layer; as summarized in eqn (14)). aCO2 + bO2 + BaCo0.4Fe0.4Nb0.2O3%d - aBaCO3 + cCoO + (1 % a)BaCo(0.4%y)Fe(0.4+z)Nb(0.2+y%z)O3%d 0

(14)

In this work, Yi et al. also studied the effect of increasing the content of Fe or Nb on BaCo1%x%yFexNbyO3%d (x = 0.2–0.8, y = 0.2–0.5). Increasing Fe or Nb resulted in higher CO2 resistance and reduced oxygen permeation flux, i.e., the trade-off problem we mentioned earlier. Notably, incorporating cations with a high oxidation state affects the oxygen non-stoichiometry, i.e., decreasing the oxygen non-stoichiometry according to a charge neutrality criterion and strengthening the metal–oxygen bonding; both of which inhibit the formation of oxygen vacancies and therefore, high oxygen ionic conductivity. Zhou et al. also investigated the degradation and recovery mechanisms of BCFN perovskites.325 They observed the same phase and microstructure change of the BCFN membranes after annealing under pure CO2 at 850 1C for 10 hours. They further tested the recovery of BCFN membranes under permeation conditions by using pure CO2 as the sweep gas, followed by switching the sweep gas to 5 vol% CO2 in He. After applying 5 vol% CO2 in He for 60 hours, the fully covered BaCO3 top layer broke; generating pores and cracks on the surface. The O2 flux could be partially recovered at the end of the 60 hour-permeation test. The main contributor to the carbonate layer decomposition and

Fig. 13 Characterization of BaCo0.4Fe0.4Nb0.2O3%d membranes exposed to CO2 at 900 1C; (a) scanning transmission electron microscopy high-angle annular dark field micrograph of the porous layer (left) and electron diffraction pattern (right) along334 the zone axis of the perovskite phase (b and c represent the BaCO3 and CoO phase, respectively) – after 2 hours; (b) back-scattered electron microscopy (BSEM) image of the cross-section of the membrane initially deposited with gold particles on its top surface – after 90 hours; (c) BSEM image and (d) elemental mapping of the cross-section of the membrane – after 240 hours; (e) cation composition as a function of sample depth from scanning electron microscopy-energy dispersive X-ray analysis – after 240 hours; and (f) the plot of the square of the thickness of the decomposed layer versus the exposure time (linear plot indicates the adequacy of the parabolic rate law to describe the decomposition process). Reproduced with permission from ref. 30. Copyright 2010, American Chemical Society.


Chem. Soc. Rev.

View Article Online


Review Article

the partial O2 flux recovery is the fast exchange of oxygen ions between BCFN and carbonate, which possibly leads to a dynamical equilibrium of carbonate decomposition and formation. The presence of O2 in the sweep gas restrained carbonate formation through competitive adsorption of O2 and CO2 on the BCFN surface and stabilized the cubic perovskite structure. Another example of excellent work using STEM in combination with EDX mapping and EELS is the work of Ravkina et al.44 Fig. 14(a)–(d) shows the characterization results on the SrCo0.8Fe0.2O3%d membrane side subjected to CO2 exposure at 900 1C for 30 hours. HAADF imaging (Fig. 14(a)) displays regions with different phase compositions. Energy-filtered transmission electron microscopy (EFTEM) image and EDX mapping (Fig. 14(b) and (c)) clearly reveal different phases; regions dominated by CoO (red region),

Chem Soc Rev

SrCO3 (blue region) and perovskite phase (green region). EELS onto these different phases (Fig. 14(d)) verify the presence of peaks characteristic of Fe–L2,3 (710 and 723 eV) for the perovskite phase, Co–L2,3 (782 and 796 eV) for the CoO phase which are absent for the carbonate phase. Similar results on the La0.2Sr0.8Co0.8Fe0.2O3%d membrane side subjected to CO2 exposure at 900 1C for 200 hours which feature HAADF imaging, EDX mapping, and the resultant decomposition scheme are included as Fig. 14(e)–(j). Areas with different contrast appear in these figures, particularly the carbonate phase (‘‘c’’ in Fig. 14(e) – which shows a lack of signals from La, Sr, Co and Fe – Fig. 14(f)– (i)) and CoO rich phase (Fig. 14(g) – which also shows a lack of signals from La, Sr and Fe – Fig. 14(f), (h) and (i)). Another set of results for the LSCF6482 membrane side (also subjected to CO2

Fig. 14 Characterization of the sweep side of the SrCo0.8Fe0.2O3%d membrane after exposure to CO2 at 900 1C for 30 hours; (a) scanning transmission electron microscopy (STEM) image; (b) energy-filtered transmission electron microscopy (EFTEM) image; (c) energy-dispersive X-ray spectroscopy (EDXS) maps (orange dashed line represents the border of SrCO3); (d) electron energy loss spectra (EELS) for the perovskite phase, cobalt oxide and strontium carbonate; characterization of the sweep side of the LSCF6482 membrane after exposure to CO2 at 900 1C for 200 hours; (e) high-angle annular dark field STEM image; (f–i) EDXS maps; (j) decomposition scheme; characterization of the sweep side of the LSCF6482 membrane after exposure to CO2 at 900 1C for 200 hours; (k) high-angle annular dark field STEM image and EDXS maps; and (l) EELS for the perovskite phase and cobalt oxide. Reproduced with permission from ref. 44. Copyright 2015, Elsevier Inc.

Chem. Soc. Rev.


View Article Online

Chem Soc Rev

Benson et al. studied the degradation mechanism of LSCF6428 in an aggressive atmosphere containing carbon dioxide and water (H2O/CO2/O2 in the ratio of 2 : 1 : 1).327 They characterized the degradation evolution using powder XRD, Raman spectroscopy, isotope exchange depth profiling, and secondary ion mass spectrometry (Fig. 15). The powder X-ray diffraction (XRD) pattern of the sample aged for 1 week at 1 atm


exposure at 900 1C for 200 hours – Fig. 14(k) and (l)) show the formation of a CoO rich phase (‘‘1’’ in Fig. 14(k)) and a perovskite phase (‘‘2’’ in Fig. 14(k)) without any indication of the carbonate phase. These microscopy imaging and analysis results are consistent with the oxygen permeation testing results; both of which substantiate the fact that LSCF6482 is stable to CO2 at 900 1C while SrCo0.8Fe0.2O3%d and LSCF2882 are not.

Review Article

Fig. 15 (a) Powder X-ray diffraction patterns of aged samples: (A) untreated standard LSCF6428, (B) aged for 1 week at 1 atm total pressure, (C) aged for 1 week at 5 atm, (D) aged for 6 weeks at 5 atm; (b) Raman spectra of the bulk material of the aged samples: (A) standard untreated LSCF, (B) aged for 1 day at 1 atm total pressure, (C) aged for 3 days at 1 atm total pressure, (D) aged for 1 week at 1 atm total pressure, (E) aged for 1 week at 5 atm total pressure, (F) aged for 6 weeks at 5 atm total pressure; (c) the effect of the duration of the ageing annealing on the 18O tracer diffusivity D* (m) and surface exchange coefficient k (K); and (d) 18O diffusion profiles obtained for unaged LSCF (A) and LSCF exposed to the ageing atmosphere for 3 days (B). Reproduced with permission from ref. 327. Copyright 1999, The Electrochemical Society.


Chem. Soc. Rev.

View Article Online


Review Article

pressure showed no discernible differences relative to the untreated sample (Fig. 15(a)). The pattern for the sample aged for 1 week at 5 atm pressure nonetheless revealed the presence of secondary phases of SrCO3 and mixed transition metal oxides. A La2O2CO3 phase and A-site deficient LSCF6428 emerged when the aging time was extended to 6 weeks (Fig. 15(a)). To this end, another study showed that B-site cations may precipitate into oxides when the A-site deficiency reached a certain level, resulting in the presence of a substantial amount of transition metal oxides detected in the powder XRD pattern.328 Raman spectra further confirmed the decomposition of the LSCF6428 surface (Fig. 15(b)). An increase in the aging annealing time and H2O/CO2/O2 pressure led to the generation of the distinct peaks at 474, 620, and 695 cm%1 in addition to the main LSCF6428 peak at 570 cm%1. The peaks at 474 and 620 cm%1 were assigned to Fe2O3 and that at 695 was assigned to the (Co, Fe)3O4 phase.329,330 In addition, the main LSCF6428 peak which was originally located at 570 cm%1 for the untreated material shifted to 590 cm%1 for the sample aged for 6 weeks. Benson et al. attributed this to the gradual transformation of cation-stoichiometric LSCF6428 to modified perovskite composition. Powder XRD and Raman spectra consistently showed the surface composition changes upon exposure to a H2O/CO2/O2 atmosphere. Isotope exchange depth profiling and secondary ion mass spectrometry were used to determine the oxygen transport properties of oxide ion conducting ceramics. Two main parameters from the two characterization methods were the tracer diffusion coefficient D* and the surface exchange coefficient k. D* represents the bulk diffusion rate of oxygen ions through the oxide and k represents the exchange rate between oxygen in the gas atmosphere and that in the oxide. The related theory details can be found elsewhere.327 D* remained almost constant for the materials exposed to a H2O/CO2/O2 atmosphere ranging from 3.75 hours to 3 days at 750 1C before isotope exchange, while k decreased dramatically after 4 hours of treatment in a H2O/CO2/ O2 atmosphere (Fig. 15(c)). Oxygen exchange furthermore appears to be inhibited by surface corrosion by the H2O/CO2/ O2 atmosphere (Fig. 15(d)). The exposure to H2O and CO2 led to significant erosion of the surface layers during aging. The erosion essentially modified the surface by generating A-site deficient perovskite and Co or Fe oxides which inhibited the surface oxygen exchange. We have presented the discussions on MIEC membrane degradation mechanisms via the utilization of different spectroscopic techniques in the membrane surface characterization. Here, we give a summary of these techniques as a brief guide to researchers in this field. The characterization of MIEC membranes, including spectroscopic techniques have been discussed in a previous comprehensive review.15 Therefore, we focus on the spectroscopic techniques for carbonate detection or CO2-resistant properties. Powder XRD is a widely-used characterization technique to analyze the phase composition of membranes. Within the context of CO2-resistant membranes, powder XRD is usually applied to determine the phase change and carbonate formation on the membrane surface after the permeation process.

Chem. Soc. Rev.

Chem Soc Rev

Yi et al. detected BaCO3 formation using powder XRD on BCFN powder after annealing in a CO2-atmosphere.30 Researchers often analyze directly the spent membrane surface (instead of crushing it into a powder form), especially the sweep side, after a permeation test. Compared to conventional powder XRD, in situ powder XRD is a more effective way of characterizing the surface phase change under a certain gas mixture in a certain temperature range. For example, in Efimov et al.’s work, in situ powder XRD tests were conducted between room temperature and 1000 1C in a 50 vol% CO2/50 vol% air and 50 vol% CO2/50 vol% N2 atmosphere.41 At each temperature step, the temperature was held for 30 min for data collection. From such in situ powder XRD spectroscopy results, we can observe the surface phase change (i.e., carbonate formation) process. We have summarized the test results and listed them in Tables 1–4. The detection limit is normally up to B2% for a certain compound within the mixture. Therefore, in most cases, carbonate formation cannot be reliably detected using powder XRD. For instance, in one report from Zhu et al., powder XRD was used to determine if carbonates were formed on SrNb0.1Co0.9%xFexO3%d (SNCFx) after long-term exposure to pure CO2. The carbonates were absent in their powder XRD results.50 Their Fourier Transform Infrared Spectroscopy (FTIR) results however revealed the presence of SrCO3 for CO2-exposed SNCFx (x = 0.1, 0.2 and 0.3). FTIR spectroscopy is a characterization technique to generate an infrared spectrum of the emission or absorption of materials, which is sensitive to detect the presence of CO32%.331 The FTIR spectrum is frequently obtained to probe carbonate formation on a MIEC membrane surface. Typical peaks representing carbonates appear in the FTIR spectrum if carbonates are present on the surface. In one example, after heat treatment at 850 1C for 10 h in a pure CO2 atmosphere, SrFeO3%d showed three characteristic bands at 860, 1445 and 1771 cm%1, which signifies the appearance of CO32% from SrCO3.124 Raman spectroscopy is used to probe the surface chemistry of a membrane with the ability to provide a fingerprint of surface molecules. Raman spectroscopy can detect subtle changes in the near-surface region (tens of nanometers in depth), which XRD is insensitive to.327 Within the CO2resistance context, this technique allows the observation of the surface phase change after the MIEC membrane is exposed to CO2, CO, or other gases. For instance, carbonate species were observed on Raman spectra for BSCF5582 and SrCoFe0.5Ox after exposure to CO and CO2.332 Moreover, using Raman spectroscopy, the change of the surface phase of samples exposed to CO2 for different time durations can be determined to reveal the degradation mechanism.327 EDX in combination with SEM (SEM-EDX) is an effective approach to characterize the membrane surface after an oxygen permeation test, especially when CO2 is present in the sweep gas. It provides the elemental analysis or the elemental distribution (EDX mapping) on the selected region on the tested membrane surface. The application of SEM-EDX analysis is illustrated in the paper by Xue et al., in which the EDX mapping of a (Pr0.9La0.1)1.9(Ni0.74Cu0.21Ga0.05)O4+d membrane was obtained to confirm the stability against CO2.237 In addition, STEM-EDX and


View Article Online


Chem Soc Rev

STEM-EELS can also be used to perform elemental analysis on a smaller sampling region on the membrane surface.30,44 STEM-EDX provides elemental distribution information in a sub-micro-sized region.44 Using this technique, carbonate particles as small as 200– 300 nanometers can be identified.44 The EDX technique is more sensitive to heavier elements. It cannot detect light elements. STEM coupled with EELS (STEM-EELS) is a complementary technique to STEM-EDX. It can be used to reveal the atomic and electronic structure of light elements at high resolution, which makes it applicable for carbon analysis.44,196 Lin et al. provided a comprehensive example of STEM-EELS application to analyze the grain boundary of dual-phase membranes.196 Ravkina et al. also used STEM-EELS to reveal carbonate formation on LSCF2882 membranes.44 XPS also plays an important role in the chemical analysis of materials. The CO2-resistance of perovskite oxide can be estimated by measuring the basicity of the perovskite oxide. The O1s binding energy obtained from XPS reflects the basicity of perovskites. Details of the use of XPS techniques are described in Section 2.1.2.2.

6. SO2-resistance and other poisoning elements We have mainly discussed the CO2-resistance of oxygen selective MIEC membranes. SO2 may be present as gas impurities from the combustion of sulfur-containing coals or other biomasses. In this section, we present a short discussion on MIEC membranes tested in a SO2-containing atmosphere. Engels et al. were the first to report the effect of SO2 on the oxygen permeation membranes.333 In their study, the performance of Sr0.5Ca0.5Mn0.8Fe0.2O3%d (SCMF) and La2NiO4+d (LN) was tested in CO2 and CO2–SO2 mixtures. For the SO2 poisoning test, they applied a mixture of CO2 and 360 ppm SO2 which mimics the expected SO2 content in the recirculated flue gas of a power plant process. The oxygen test was initially performed using pure CO2 as the sweep gas. At 900 1C, a stable O2 flux of 0.125 mL min%1 cm%2 was obtained. The change of the sweep gas to a CO2–SO2 mixture however caused an immediate stagnation of O2 permeation. Upon shifting the sweep gas to pure CO2, after a few hours, the O2 flux could not be recovered to its initial value. The EDX results on the membrane after the oxygen permeation test indicated the presence of the S element from SrSO4 as verified by powder XRD results. SrSO4 was most likely formed through the reaction between SrMnO3 and SO2. An identical oxygen permeation test was performed on the LN membrane. At 900 1C, the O2 flux dropped to below 0.01 mL min%1 cm%2 immediately but did not become zero. However, at 800 and 850 1C, the O2 flux immediately deteriorated to zero, similar to the SCMF case. SEM and EDX indicated the formation of sulfur-containing reaction products on the surface. They however did not probe in detail the effect of SO2 on the oxygen permeation. A more detailed study on the effects of SO2 on oxygen permeation can be found in the work of Wei et al. on a


Review Article

K2NiF4-type (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d (PLNCG) hollow fiber.42 When helium containing SO2 (84, 206 and 383 ppm) was used as the sweep gas, the oxygen permeation flux decreased with the test prolongation and SO2 concentration increment. In the case of 383 ppm of SO2, the O2 flux dropped by 88% within 60 min. According to Zhang et al., the SO2 poisoning on perovskite LaFe0.8Cu0.2O3 had two distinct mechanisms depending on the SO2 concentration.334 The first mechanism is prevalent at low SO2 concentrations where the poisoning is caused by the competitive adsorption of SO2 and reactants as well as the occupancy of active sites by sulfite and sulfate. The other mechanism is the severe sulphuration that results in the breakdown of the material’s structure and the generation of sulfate phases at high SO2 concentrations.334 In the PLNCG hollow fiber case, the membrane surface underwent the following reaction at high SO2 concentrations: (Pr0.9La0.1)2(Ni0.74Cu0.21Ga0.05)O4+d + SO2 + (1.025 % d)/2O2 = 0.1La2O2SO4 + 0.9Pr2O2SO4 + 0.74NiO + 0.21CuO + 0.025Ga2O3

(15)

The formation of La2O2SO4, Pr2O2SO4, and NiO was identified from the powder XRD analyses of the tested PLNCG hollow fiber. The SEM images showed a highly porous corroded region at 150 mm from the sweep surface which explained the destruction of the PLNCG structure. Moreover, La2O2SO4 and Pr2O2SO4 from the decomposition of lanthanum and praseodymium sulfates were stable between 900–1200 1C. This hampered their decomposition during the oxygen permeation process.335,336 To this end, PLNCG was confirmed as the SO2-sensitive material given the segregation of La and Pr. Tan and co-workers studied the SO2 poisoning effect on the oxygen permeation of LSCF6428 hollow fibers.337 The O2 flux decreased significantly after exposure to 1750 ppm SO2. For example, at 1000 1C, the O2 flux decreased by 80%, from 3.8 to 0.78 mL min%1 cm%2. Unlike CO2, SO2 caused irreversible damage following operation in a SO2-containing atmosphere. Via SEM, EDX, and powder XRD analyses, they attributed the poisoning effect of SO2 to the decomposition of a thin layer (i.e., several to tens of micrometres) on the hollow fiber sweep side, leading to the formation of SrSO4. In addition to these studies of the SO2 poisoning on singlephase membranes, several studies tested SO2 poisoning on dual-phase membranes. Balaguer et al. studied the simultaneous CO2 and SO2 resistance of NiFe2O4–Ce0.8Tb0.2O2%d (60 : 40 vol%) dual-phase membranes.132 The membrane materials were subjected to powder XRD analysis, EDX analysis, and Raman spectroscopy after exposure to a continuous flow of CO2 containing 5% O2 and 1% SO2 at 800 1C for 170 h. Their results showed the absence of a reaction between the gas species and the membrane materials, indicating this material has potential to be used in an oxyfuel power plant. Still, no oxygen permeation test in a SO2 (only)-containing atmosphere has been reported in their work. Another report on the SO2-resistant dual-phase membrane is the work of Cheng et al.141 In their work, dualphase Al0.02Ga0.02Zn0.96O1.02–Gd0.1Ce0.9O1.95%d (50 : 50 vol%) was

Chem. Soc. Rev.

View Article Online


Review Article

annealed in a SO2-rich atmosphere at 850 1C for 2 h. Following such exposure, sulfate did not form. They also did not perform an oxygen permeation test in an SO2 (only)-containing atmosphere. From these limited amounts of reports of SO2 poisoning effects on MIEC membranes, we can deduce that the absence of alkaline earth metal elements, e.g., La and Pr in the membrane materials, is necessary to achieve SO2-resistance. Dual-phase membranes therefore become a better option relative to singlephase membranes. In this context, the thermodynamic data showing the decomposition temperature of the sulfates or sulfites may become important for the design of SO2-resistant MIEC membranes. Materials displaying excellent CO2 and SO2 resistance are definitely of interest for application in an oxyfuel combustion power plant that involves sulfur-containing coals and biomass.3,338 Besides CO2 and SO2, other poisoning solid elements are also present that may affect the structure and chemical stability and oxygen permeation of MIEC membranes. One such common poisoning substance is silica (Si). Liu et al. reported that the O2 permeation flux of BaCe0.1Co0.4Fe0.5O3%d at 600 1C decreased by 62% after operation for 477 h.339 They attributed the flux degradation to the presence of silicon impurities from the chemicals used and/or from the preparation process of the membranes. Their SEM and EDX analyses revealed that the surface of the membrane sweep side had a higher silicon content than that on the feed side. A 25 nm-thick layer containing silicon was detected by HRTEM on the sweep side. The aggregation of silicon on the surface appeared to diffuse out of the bulk membrane. They also demonstrated that the presence of a porous surface coating made from Sm0.5Sr0.5CoO3%d could accommodate the silicon impurities and significantly improved the stability at 600 1C. Silicon can also come from the siloxanecontaining grease that was used during the permeation testing setup.340 When grease-free valves were used in the setup, the O2 permeation performance of LSCF6428 improved significantly, resulting in a stable O2 flux of 0.7 mL min%1 cm%2 that was substantially higher relative to 0.4 mL min%1 cm%2 achieved in the old setup.340 Similar to Si, sulfur (S) is another poisoning element that degrades the oxygen permeation of MIEC membranes especially during long-term operation at low temperatures.250 In contrast to SO2 sourcing from the gas phase, S generally comes as impurities (SO42%) from the raw materials. As such, it may migrate slowly from the bulk membrane to the surface during long-term permeation.250 Another typical poisoning element is chromium (Cr). Cr typically exists in the steel alloys used as construction materials in coal-fired power plants.333 Under the high temperature conditions required for oxygen permeation, the volatilization of Cr may damage the membranes. Such damage becomes more severe particularly in a water-vapor containing atmosphere.341 A study on BaCo0.7Fe0.2Nb0.1O3%d showed that its O2 flux decreased by 2.4% after 100 h operation at 850 1C when a chromia-forming steel component was present on the feed side. Moreover, when humidified air (i.e., 4% H2O) was introduced, the O2 flux decreased by 10%. BaCrO4 and BaCr2O4 were formed on the feed side while only BaCrO4 was present on the permeate side. Cr from steel alloys caused a serious poisoning effect to the oxygen permeation MIEC

Chem. Soc. Rev.

Chem Soc Rev

membranes, identical to its poisoning effect in SOFCs.342–344 Other impurities such as Na+, K+, Mg,2+ Ca2+, Mnx+, and Cl% may also appear to contribute to the deterioration of the structure and/or the performance of MIEC membranes.345 Further systematic work to evaluate the poisoning effect of these impurities is warranted.

7. Conclusions Our literature survey into the vast bodies of work on the development of CO2-resistant oxygen-selective mixed ionic–electronic conducting (MIEC) membranes revealed that only two material candidates display the highest potential, i.e., perovskite and fluorite-based materials. Perovskite-based MIEC materials generally offer high O2 permeability and low CO2 resistance especially those that contain alkaline earth metals such Ba and Sr. From the thermodynamics perspective, a stability diagram that correlates different phases as a function of the activity ratio of different metal oxide pairs can be used to quantify the CO2 resistance of perovskite oxides. A more popular alternative is the Ellingham diagram which expresses the relative tendency for a certain carbonate formation at different temperatures and CO2 partial pressures. In the Thermodynamics perspective section, we have provided the essential steps required to construct an Ellingham diagram. From the Lewis acid–base perspective, metal (or perovskite) oxides are viewed as a base whereas CO2 is viewed as an acid. Accordingly, lower basicity represents higher CO2 resistance. Sanderson’s electronegativity and the oxidation state of the cations can be utilized to assess the relative basicity of the metal oxides. An indispensable resource we included in the Chemistry perspective section is Fig. 6 that displays the relative basicity for 101 metal oxides. In the absence of ready tobe-used data, the basicity of the metal (or perovskite) oxide can also be experimentally determined from the O1s binding energy (BE) value of X-ray photoelectron spectra. Within this context, a higher O1s binding energy reflects a lower electron density (or higher oxidation state) and therefore, lower basicity. Another parameter that indicates the basicity of the metal (or perovskite) oxide is the average metal–oxygen bond energy (ABE). We can generalize three major directions that have been undertaken to enhance the CO2 resistance of perovskite materials, i.e., (1) adopting a rare earth metal cation as the A-site cation; (2) reduction in the Co content in the B-site; and (3) partial substitution of the original transition metal components in the B-site with more acidic (less basic) or more stable transition metal components. Nevertheless, since the ABE reflects the basicity of the perovskite oxide while the oxygen sorption from the lattice is also determined by how strong the oxygen is bonded to the metal and thus, the ABE (i.e., lower ABE provides high sorption tendency and high oxygen permeability), it becomes apparent that a compromise between oxygen permeability and CO resistance exists where the increase in CO2 resistance leads to the decrease in O2 permeability. Such a trade-off effect is not encountered in the fluorite material that is practically CO2-resistant but has low O2 permeability.


View Article Online


Chem Soc Rev

The electronic conductivity limitation origin of such low permeability can be addressed using internal and/or external electronic short-circuit concepts. These concepts represent different ways of adding the second electronic conducting (EC) or MIEC phase. The literature trend showed that incorporating the MIEC second phase generally leads to higher O2 permeability than that obtained from incorporating the EC second phase. For industrial applications that prioritize consistent performance for hundreds or even thousands of hours, dual-phase materials appear to be a more promising candidate than perovskite materials. We envision major performance breakthroughs for the dual-phase membrane technology given the rapid advances over the past 5 years in addition to the emergence of a new direction that exploits in situ phase reactions to create a conductive new phase within the composite along the boundary area. Despite the long way ahead towards the commercialization of CO2-resistant oxide-based membrane technology, this field will certainly benefit from having the guidelines we provided here.

Acknowledgements C. Zhang gratefully acknowledges the scholarship provided by Curtin University. S. Liu gratefully acknowledges the financial support provided by the Australian Research Council through the Future Fellow Program (FT120100178).

References 1 J. D. Figueroa, T. Fout, S. Plasynski, H. McIlvried and R. D. Srivastava, Int. J. Greenhouse Gas Control, 2008, 2, 9–20. 2 M. A. Habib, H. M. Badr, S. F. Ahmed, R. Ben-Mansour, K. Mezghani, S. Imashuku, G. J. la O’, Y. Shao-Horn, N. D. Mancini, A. Mitsos, P. Kirchen and A. F. Ghoneim, Int. J. Energy Res., 2011, 35, 741–764. ¨rster, D. Christ, C. Broeckmann, 3 R. Kneer, D. Toporov, M. Fo E. Pfaff, M. Zwick, S. Engels and M. Modigell, Energy Environ. Sci., 2010, 3, 198–207. 4 P. A. Armstrong, D. L. Bennett, E. P. (Ted) Foster and V. E. Stein, ITM oxygen for gasification, Air Products and Chemicals Inc., Washington, DC, 2004. 5 S. Engels, F. Beggel, M. Modigell and H. Stadler, J. Membr. Sci., 2010, 359, 93–101. 6 Y. Wei, W. Yang, J. Caro and H. Wang, Chem. Eng. J., 2013, 220, 185–203. 7 X. Dong, W. Jin, N. Xu and K. Li, Chem. Commun., 2011, 47, 10886–10902. 8 A. Thursfield and I. S. Metcalfe, J. Mater. Chem., 2004, 14, 2475–2485. 9 Y. Liu, X. Tan and K. Li, Catal. Rev.: Sci. Eng., 2006, 48, 145–198. 10 W. Yang, H. Wang, X. Zhu and L. Lin, Top. Catal., 2005, 35, 155–167. 11 I. Riess, J. Power Sources, 2008, 175, 325–337. 12 M. Yano, A. Tomita, M. Sano and T. Hibino, Solid State Ionics, 2007, 177, 3351–3359.


Review Article

13 T. Hibino, A. Hashimoto, T. Inoue, J.-I. Tokuno, S.-I. Yoshida and M. Sano, Science, 2000, 288, 2031–2033. 14 Y. Hao, Z. Shao, J. Mederos, W. Lai, D. Goodwin and S. Haile, Solid State Ionics, 2006, 177, 2013–2021. 15 J. Sunarso, S. Baumann, J. M. Serra, W. A. Meulenberg, S. Liu, Y. Lin and J. C. Diniz da Costa, J. Membr. Sci., 2008, 320, 13–41. 16 K. Zhang, J. Sunarso, Z. Shao, W. Zhou, C. Sun, S. Wang and S. Liu, RSC Adv., 2011, 1, 1661–1676. 17 A. Leo, S. Liu and J. C. Diniz da Costa, Int. J. Greenhouse Gas Control, 2009, 3, 357–367. 18 S. Smart, C. X. C. Lin, L. Ding, K. Thambimuthu and J. C. Diniz da Costa, Energy Environ. Sci., 2010, 3, 268–278. 19 D. D. Athayde, D. F. Souza, A. M. A. Silva, D. Vasconcelos, E. H. M. Nunes, J. C. Diniz da Costa and W. L. Vasconcelos, Ceram. Int., 2016, 42, 6555–6571. 20 A. S. Bhalla, R. Guo and R. Roy, Mater. Res. Innovations, 2000, 4, 3–26. 21 H. D. Megaw, Acta Crystallogr., Sect. A: Found. Adv., 1974, A30, 605–606. 22 V. M. Goldschmidt, Naturwissenschaften, 1926, 14, 477–485. 23 R. D. Shannon, Acta Crystallogr., Sect. A: Found. Adv., 1976, A32, 751–767. 24 A. Feldhoff, J. Martynczuk, M. Arnold, M. Myndyk, ´k, W. Gruner, U. Vogt, A. Ha ¨hnel I. Bergmann, V. ˇ Sepela and J. Woltersdorf, J. Solid State Chem., 2009, 182, 2961–2971. 25 M. Arnold, Q. Xu, F. D. Tichelaar and A. Feldhoff, Chem. Mater., 2009, 21, 635–640. 26 T. Ishihara, H. Matsuda and Y. Takita, J. Am. Chem. Soc., 1994, 116, 3801–3803. 27 M. Arnold, H. Wang and A. Feldhoff, J. Membr. Sci., 2007, 293, 44–52. ¨mpfer, J. Membr. Sci., 2011, 28 M. Schulz, R. Kriegel and A. Ka 378, 10–17. 29 J. Yi and M. Schroeder, J. Membr. Sci., 2011, 378, 163–170. 30 J. Yi, M. Schroeder, T. Weirich and J. Mayer, Chem. Mater., 2010, 22, 6246–6253. õ, J. F. Monteiro, A. V. Kovalevsky, D. P. Fagg, 31 A. Branda V. V. Kharton and J. R. Frade, Solid State Ionics, 2011, 192, 16–20. 32 H. Yokokawa, T. Kawada and M. Dokiya, J. Am. Ceram. Soc., 1989, 72, 2104–2110. 33 H. Yokokawa, N. Sakai, T. Kawada and M. Dokiya, J. Solid State Chem., 1991, 94, 106–120. 34 H. Yokokawa, N. Sakai, T. Kawada and M. Dokiya, Solid State Ionics, 1992, 52, 43–56. 35 H. Yokokawa, T. Kawada and M. Dokiya, J. Am. Ceram. Soc., 1989, 72, 152–153. 36 H. Yokokawa, N. Sakai, T. Kawada and M. Dokiya, J. Am. Ceram. Soc., 1990, 73, 649–658. 37 J. Martynczuk, K. Efimov, L. Robben and A. Feldhoff, J. Membr. Sci., 2009, 344, 62–70. 38 T. Klande, O. Ravkina and A. Feldhoff, J. Membr. Sci., 2013, 437, 122–130.

Chem. Soc. Rev.

View Article Online


Review Article

39 O. Czuprat, M. Arnold, S. Schirrmeister, T. Schiestel and J. Caro, J. Membr. Sci., 2010, 364, 132–137. 40 T. Klande, K. Efimov, S. Cusenza, K.-D. Becker and A. Feldhoff, J. Solid State Chem., 2011, 184, 3310–3318. 41 K. Efimov, T. Klande, N. Juditzki and A. Feldhoff, J. Membr. Sci., 2012, 389, 205–215. 42 Y. Wei, O. Ravkina, T. Klande, H. Wang and A. Feldhoff, J. Membr. Sci., 2013, 429, 147–154. 43 I. Barin, Thermochemical data of pure substances, WileyVCH Verlag GmbH, Weinheim, Germany, 3 edn, 1995. 44 O. Ravkina, T. Klande and A. Feldhoff, J. Membr. Sci., 2015, 480, 31–38. 45 Q. Zeng, Y. Zuo, C. Fan and C. Chen, J. Membr. Sci., 2009, 335, 140–144. 46 W. Chen, C. Chen and L. Winnubst, Solid State Ionics, 2011, 196, 30–33. 47 W. Chen, C. Chen, H. J. M. Bouwmeester, A. Nijmeijer and L. Winnubst, J. Membr. Sci., 2014, 463, 166–172. 48 N. C. Jeong, J. S. Lee, E. L. Tae, Y. J. Lee and K. B. Yoon, Angew. Chem., Int. Ed., 2008, 47, 10128–10132. 49 H. Noller, J. A. Lercher and H. Vinek, Mater. Chem. Phys., 1988, 18, 577–593. 50 Y. Zhu, J. Sunarso, W. Zhou and Z. Shao, Appl. Catal., B, 2015, 172–173, 52–57. 51 R. L. Cook and A. F. Sammells, Solid State Ionics, 1991, 45, 311–321. 52 E. Y. Konysheva, X. Xu and J. T. Irvine, Adv. Mater., 2012, 24, 528–532. 53 X. Tan, N. Liu, B. Meng, J. Sunarso, K. Zhang and S. Liu, J. Membr. Sci., 2012, 389, 216–222. 54 G. Z. Cao, J. Appl. Electrochem., 1994, 24, 1222–1227. 55 Y.-S. Lin, W. Wang and J. Han, AIChE J., 1994, 40, 786–798. ´sz, M. Ga ´l, K. So ´lymos, 56 Z. Homonnay, K. Nomura, G. Juha ´rtes, Chem. Mater., S. Hamakawa, T. Hayakawa and A. Ve 2002, 14, 1127–1135. 57 K. Nornura, Y. Ujihira, T. Hayakawa and K. Takehira, Appl. Catal., A, 1996, 137, 25–36. 58 A. Yan, B. Liu, Y. Dong, Z. Tian, D. Wang and M. Cheng, Appl. Catal., B, 2008, 80, 24–31. 59 W. Fang, F. Steinbach, C. Chen and A. Feldhoff, Chem. Mater., 2015, 27, 7820–7826. 60 J. M. Serra, J. Garcia-Fayos, S. Baumann, F. Schulze¨ppers and W. A. Meulenberg, J. Membr. Sci., 2013, 447, Ku 297–305. ´stico, 61 S. Baumann, J. M. Serra, M. P. Lobera, S. Escola ¨ppers and W. A. Meulenberg, J. Membr. Sci., F. Schulze-Ku 2011, 377, 198–205. 62 C. Gaudillere, J. Garcia-Fayos and J. M. Serra, ChemPlusChem, 2014, 79, 1720–1725. 63 K. Mezghani and A. Hamza, J. Membr. Sci., 2016, 518, 254–262. ¨ppers, S. Baumann, W. A. Meulenberg, 64 F. Schulze-Ku ¨ver and H. P. Buchkremer, J. Membr. Sci., 2013, D. Sto 433, 121–125. 65 J. Yi, S. Feng, Y. Zuo, W. Liu and C. Chen, Chem. Mater., 2005, 17, 5856–5861.

Chem. Soc. Rev.

Chem Soc Rev

66 W. Fang, F. Liang, Z. Cao, F. Steinbach and A. Feldhoff, Angew. Chem., Int. Ed., 2015, 54, 4847–4850. 67 J. Yi, M. Schroeder and M. Martin, Chem. Mater., 2013, 25, 815–817. 68 J. Zhu, S. Guo, Z. Zhang, X. Jiang, Z. Liu and W. Jin, J. Membr. Sci., 2015, 485, 79–86. 69 Z. Zhang, D. Chen, F. Dong and Z. Shao, J. Membr. Sci., 2015, 495, 187–197. 70 W. Yao, H. Cheng, H. Zhao, X. Lu, X. Zou, S. Li and C. Li, J. Membr. Sci., 2016, 504, 251–262. 71 M. P. Popov, I. A. Starkov, S. F. Bychkov and A. P. Nemudry, J. Membr. Sci., 2014, 469, 88–94. 72 Q. Liao, Y. Wang, Y. Chen, Y. Wei and H. Wang, J. Membr. Sci., 2014, 468, 184–191. 73 C. Gaudillere, J. Garcia-Fayos and J. M. Serra, J. Mater. Chem. A, 2014, 2, 3828–3833. 74 J. Zhu, S. Guo, Z. Chu and W. Jin, J. Mater. Chem. A, 2015, 3, 22564–22573. ¨scher and J. Caro, 75 K. Partovi, B. Geppert, F. Liang, C. H. Ru Chem. Mater., 2015, 27, 2911–2919. 76 H. Li, X. Zhu, Y. Liu, W. Wang and W. Yang, J. Membr. Sci., 2014, 462, 170–177. 77 Z. Shao, W. Yang, Y. Cong, H. Dong, J. Tong and G. Xiong, J. Membr. Sci., 2000, 172, 177–188. 78 H. Wang, Y. Cong and W. Yang, J. Membr. Sci., 2002, 210, 259–271. 79 S. Liu and G. R. Gavalas, J. Membr. Sci., 2005, 246, 103–108. 80 T. Klande, O. Ravkina and A. Feldhoff, J. Eur. Ceram. Soc., 2013, 33, 1129–1136. 81 M. A. Alaee, M. M. Movahednia and T. Mohammadi, J. Chem. Eng. Data, 2009, 54, 3082–3091. 82 S. Liu, X. Tan, Z. Shao and J. C. Diniz da Costa, AIChE J., 2006, 52, 3452–3461. 83 A. V. Kovalevsky, A. A. Yaremchenko, V. A. Kolotygin, F. M. M. Snijkers, V. V. Kharton, A. Buekenhoudt and J. J. Luyten, Solid State Ionics, 2011, 192, 677–681. 84 D. Han, X. Tan, Z. Yan, Q. Li and S. Liu, Ceram. Int., 2013, 39, 431–437. 85 X. Li, T. Kerstiens and T. Markus, J. Membr. Sci., 2013, 438, 83–89. ¨ppers, S. Roitsch, M. Betz, 86 S. Baumann, F. Schulze-Ku M. Zwick, E. M. Pfaff, W. A. Meulenberg, J. Mayer and ¨ver, J. Membr. Sci., 2010, 359, 102–109. D. Sto 87 C. Buysse, A. Kovalevsky, F. Snijkers, A. Buekenhoudt, S. Mullens, J. Luyten, J. Kretzschmar and S. Lenaerts, J. Membr. Sci., 2011, 372, 239–248. 88 A. V. Kovalevsky, A. A. Yaremchenko, V. A. Kolotygin, A. L. Shaula, V. V. Kharton, F. M. M. Snijkers, A. Buekenhoudt, J. R. Frade and E. N. Naumovich, J. Membr. Sci., 2011, 380, 68–80. 89 W. K. Hong and G. M. Choi, J. Membr. Sci., 2010, 346, 353–360. 90 J. H. Park, J. P. Kim and S. H. Son, Energy Procedia, 2009, 1, 369–374. 91 A. Yan, M. Cheng, Y. Dong, W. Yang, V. Maragou, S. Song and P. Tsiakaras, Appl. Catal., B, 2006, 66, 64–71.


View Article Online


Chem Soc Rev

92 E. Bucher, A. Egger, G. B. Caraman and W. Sitte, J. Electrochem. Soc., 2008, 155, B1218–B1224. ńg, A. S. Harvey and L. J. Gauckler, Scr. Mater., 2009, 93 Z. Ya 61, 1083–1086. 94 A. Yan, M. Yang, Z. Hou, Y. Dong and M. Cheng, J. Power Sources, 2008, 185, 76–84. 95 A. Feldhoff, M. Arnold, J. Martynczuk, T. M. Gesing and H. Wang, Solid State Sci., 2008, 10, 689–701. 96 A. Feldhoff, J. Martynczuk and H. Wang, Prog. Solid State Chem., 2007, 35, 339–353. 97 A. M. Kalinkin, Russ. J. Phys. Chem. A, 2008, 82, 263–268. 98 K. Partovi, F. Liang, O. Ravkina and J. Caro, ACS Appl. Mater. Interfaces, 2014, 6, 10274–10282. ¨bius and M. Mu ¨ller, J. Membr. Sci., 99 A. Waindich, A. Mo 2009, 337, 182–187. 100 P. Zeng, R. Ran, Z. Chen, H. Gu, Z. Shao, J. Dacosta and S. Liu, J. Membr. Sci., 2007, 302, 171–179. 101 D. Han, J. Wu, Z. Yan, K. Zhang, J. Liu and S. Liu, RSC Adv., 2014, 4, 19999–20004. 102 D. Schlehuber, E. Wessel, L. Singheiser and T. Markus, J. Membr. Sci., 2010, 351, 16–20. 103 M. F. Carolan, P. N. Dyer, S. A. Motika and P. B. Alba, US Pat., US5712220 A, 1998. 104 M. J. Shin and J. H. Yu, J. Membr. Sci., 2012, 401–402, 40–47. 105 S. K. Kim, M. J. Shin, J. Rufner, K. van Benthem, J. H. Yu and S. Kim, J. Membr. Sci., 2014, 462, 153–159. 106 V. Kharton, A. Kovalevsky, E. Tsipis, A. Viskup, E. Naumovich, J. Jurado and J. Frade, J. Solid State Electrochem., 2002, 7, 30–36. 107 Y. He, X. Zhu and W. Yang, AIChE J., 2011, 57, 87–95. 108 Z. Shao, G. Xiong, H. Dong, W. Yang and L. Lin, Sep. Purif. Technol., 2001, 25, 97–116. 109 L. Ge, W. Zhou, R. Ran, S. Liu, Z. Shao, W. Jin and N. Xu, J. Membr. Sci., 2007, 306, 318–328. 110 Y. Zhu, Z. G. Chen, W. Zhou, S. Jiang, J. Zou and Z. Shao, ChemSusChem, 2013, 6, 2249–2254. 111 A. Yaremchenko, V. Kharton, A. Valente, F. Snijkers, J. Cooymans, J. Luyten and F. Marques, J. Membr. Sci., 2008, 319, 141–148. 112 J. Liu, H. Cheng, B. Jiang, X. Lu and W. Ding, Int. J. Hydrogen Energy, 2013, 38, 11090–11096. 113 Y. Cheng, H. Zhao, D. Teng, F. Li, X. Lu and W. Ding, J. Membr. Sci., 2008, 322, 484–490. 114 H. Luo, Y. Wei, H. Jiang, W. Yuan, Y. Lv, J. Caro and H. Wang, J. Membr. Sci., 2010, 350, 154–160. 115 J. Tong, W. Yang, R. Cai, B. Zhu and L. Lin, Catal. Lett., 2002, 78, 129–137. 116 M. Harada, K. Domen, M. Hara and T. Tatsumi, Chem. Lett., 2006, 35, 1326–1327. 117 Q. Li, X. Zhu, Y. He and W. Yang, Catal. Today, 2010, 149, 185–190. 118 H. Luo, B. Tian, Y. Wei, H. Wang, H. Jiang and J. Caro, AIChE J., 2010, 56, 604–610. 119 T. Nagai, W. Ito and T. Sakon, Solid State Ionics, 2007, 177, 3433–3444.


Review Article

120 Z. Zhang, Y. Chen, M. O. Tade, Y. Hao, S. Liu and Z. Shao, J. Mater. Chem. A, 2014, 2, 9666–9674. 121 F. Yang, H. Zhao, J. Yang, M. Fang, Y. Lu, Z. Du, K. ´ Swierczek and K. Zheng, J. Membr. Sci., 2015, 492, 559–567. 122 J. Tong, W. Yang, R. Cai, B. Zhu and L. Lin, Mater. Lett., 2002, 56, 958–962. 123 Z. Shao, G. Xiong, Y. Cong and W. Yang, J. Membr. Sci., 2000, 164, 167–176. 124 C. Zhang, H. Tian, D. Yang, J. Sunarso, J. Liu and S. Liu, ChemSusChem, 2016, 9, 505–512. 125 Z. Zhang, D. Chen, F. Dong, X. Xu, Y. Hao and Z. Shao, J. Membr. Sci., 2016, 519, 11–21. 126 X. Tan, Y. Liu and K. Li, AIChE J., 2005, 51, 1991–2000. 127 X. Tan, Z. Pang and K. Li, J. Membr. Sci., 2008, 310, 550–556. 128 Z. Wang, N. Yang, B. Meng and X. Tan, Ind. Eng. Chem. Res., 2009, 48, 510–516. 129 X. Tan, Z. Wang, H. Liu and S. Liu, J. Membr. Sci., 2008, 324, 128–135. 130 X. Tan, N. Liu, B. Meng and S. Liu, J. Membr. Sci., 2011, 378, 308–318. 131 H. Pan, L. Li, X. Deng, B. Meng, X. Tan and K. Li, J. Membr. Sci., 2013, 428, 198–204. 132 M. Balaguer, J. Garcıá-Fayos, C. Solı´s and J. M. Serra, Chem. Mater., 2013, 25, 4986–4993. 133 J. Garcia-Fayos, M. Balaguer and J. M. Serra, ChemSusChem, 2015, 8, 4242–4249. 134 M. P. Lobera, S. Escolastico, J. Garcia-Fayos and J. M. Serra, ChemSusChem, 2012, 5, 1587–1596. 135 E. Bucher and W. Sitte, Solid State Ionics, 2011, 192, 480–482. 136 X. Zhu and W. Yang, in Membrane Science and Technology, ed. S. T. Oyama and S. M. Stagg-Williams, Elsevier, 2011, vol. 14, ch. 12, pp. 275–293. 137 J. Xue, Q. Liao, Y. Wei, Z. Li and H. Wang, J. Membr. Sci., 2013, 443, 124–130. 138 H. Luo, K. Efimov, H. Jiang, A. Feldhoff, H. Wang and J. Caro, Angew. Chem., Int. Ed., 2011, 50, 759–763. 139 H. Luo, H. Jiang, K. Efimov, J. Caro and H. Wang, AIChE J., 2011, 57, 2738–2745. 140 B. Jiang, H. Cheng, L. Luo, X. Lu and Z. Zhou, J. Mater. Sci. Technol., 2014, 30, 1174–1180. 141 S. Cheng, M. Søgaard, L. Han, W. Zhang, M. Chen, A. Kaiser and P. V. Hendriksen, Chem. Commun., 2015, 51, 7140–7143. 142 F. Liang, H. Luo, K. Partovi, O. Ravkina, Z. Cao, Y. Liu and J. Caro, Chem. Commun., 2014, 50, 2451–2454. 143 X. Zhu, H. Liu, Y. Cong and W. Yang, Chem. Commun., 2012, 48, 251–253. 144 X. Zhu, Q. Li, Y. He, Y. Cong and W. Yang, J. Membr. Sci., 2010, 360, 454–460. 145 H. Luo, T. Klande, Z. Cao, F. Liang, H. Wang and J. Caro, J. Mater. Chem. A, 2014, 2, 7780–7787. 146 H. Cheng, L. Luo, W. Yao, X. Lu, X. Zou and Z. Zhou, J. Membr. Sci., 2015, 492, 220–229. 147 K. J. Yoon and O. A. Marina, J. Membr. Sci., 2016, 499, 301–306.

Chem. Soc. Rev.

View Article Online


Review Article

148 H. Cheng, N. Zhang, X. Xiong, X. Lu, H. Zhao, S. Li and Z. Zhou, ACS Sustainable Chem. Eng., 2015, 3, 1982–1992. ¨scher, F. Steinbach and J. Caro, 149 K. Partovi, C. H. Ru J. Membr. Sci., 2016, 503, 158–165. 150 H. Luo, H. Jiang, T. Klande, Z. Cao, F. Liang, H. Wang and J. Caro, Chem. Mater., 2012, 24, 2148–2154. 151 J. H. Joo, K. S. Yun, J. H. Kim, Y. Lee, C. Y. Yoo and J. H. Yu, ACS Appl. Mater. Interfaces, 2015, 7, 14699–14707. 152 T. Liu, Y. Chen, S. Fang, L. Lei, Y. Wang, C. Ren and F. Chen, J. Membr. Sci., 2016, 520, 354–363. 153 Z. Wang, W. Sun, Z. Zhu, T. Liu and W. Liu, ACS Appl. Mater. Interfaces, 2013, 5, 11038–11043. 154 H. Luo, H. Jiang, T. Klande, F. Liang, Z. Cao, H. Wang and J. Caro, J. Membr. Sci., 2012, 423–424, 450–458. 155 H. Luo, H. Jiang, K. Efimov, F. Liang, H. Wang and J. Caro, Ind. Eng. Chem. Res., 2011, 50, 13508–13517. 156 H. Li, Y. Liu, X. Zhu, Y. Cong, S. Xu, W. Xu and W. Yang, Sep. Purif. Technol., 2013, 114, 31–37. 157 S. Guo, Z. Liu, J. Zhu, X. Jiang, Z. Song and W. Jin, Fuel Process. Technol., 2016, 154, 19–26. 158 J. H. Joo, K. S. Yun, C.-Y. Yoo and J. H. Yu, J. Mater. Chem. A, 2014, 2, 8174–8178. 159 K. S. Yun, C.-Y. Yoo, S.-G. Yoon, J. H. Yu and J. H. Joo, J. Membr. Sci., 2015, 486, 222–228. 160 K. Partovi, M. Bittner and J. Caro, J. Mater. Chem. A, 2015, 3, 24008–24015. 161 X. Zhu, Y. Liu, Y. Cong and W. Yang, Solid State Ionics, 2013, 253, 57–63. 162 X. Bi, X. Meng, P. Liu, N. Yang, Z. Zhu, R. Ran and S. Liu, J. Membr. Sci., 2017, 522, 91–99. 163 P. Aldebert and J.-P. Traverse, J. Am. Ceram. Soc., 1985, 68, 34–40. 164 H. Inaba and H. Tagawa, Solid State Ionics, 1996, 83, 1–16. 165 S. Kuharuangrong, J. Power Sources, 2007, 171, 506–510. 166 M. Mogensen, N. M. Sammes and G. A. Tompsett, Solid State Ionics, 2000, 129, 63–94. 167 K. Eguchi, T. Setoguchi, T. Inoue and H. Arai, Solid State Ionics, 1992, 52, 165–172. 168 T. H. Etsell and S. N. Flengas, Chem. Rev., 1970, 70, 339–376. 169 M. Balaguer, C. Solı´s and J. M. Serra, J. Phys. Chem. C, 2012, 116, 7975–7982. 170 H. Yahiro, K. Eguchi and H. Arai, Solid State Ionics, 1989, 36, 71–75. 171 T. Mori, J. Drennan, J.-H. Lee, J.-G. Li and T. Ikegami, Solid State Ionics, 2002, 154–155, 461–466. 172 S. K. Tadokoro and E. N. S. Muccillo, J. Eur. Ceram. Soc., 2007, 27, 4261–4264. 173 S. Maschio, E. Aneggi, A. Trovarelli and V. Sergo, Ceram. Int., 2008, 34, 1327–1333. 174 D. P. Fagg, S. Garcıá-Martin, V. V. Kharton and J. R. Frade, Chem. Mater., 2009, 21, 381–391. 175 S. Omar, E. D. Wachsman, J. L. Jones and J. C. Nino, J. Am. Ceram. Soc., 2009, 92, 2674–2681. 176 M. Balaguer, C.-Y. Yoo, H. J. M. Bouwmeester and J. M. Serra, J. Mater. Chem. A, 2013, 1, 10234.

Chem. Soc. Rev.

Chem Soc Rev

177 E. Jud, C. B. Huwiler and L. J. Gauckler, J. Am. Ceram. Soc., 2005, 88, 3013–3019. 178 D. P. Fagg, A. L. Shaula, V. V. Kharton and J. R. Frade, J. Membr. Sci., 2007, 299, 1–7. 179 M. Balaguer, C. Solı´s and J. M. Serra, Chem. Mater., 2011, 23, 2333–2343. 180 Y. Xia, Y. Bai, X. Wu, D. Zhou, Z. Wang, X. Liu and J. Meng, Solid State Sci., 2012, 14, 805–808. ¨fe and F. Aldinger, 181 N. M. Sammes, G. A. Tompsett, H. Na J. Eur. Ceram. Soc., 1999, 19, 1801–1826. 182 A. L. Shaula, V. V. Kharton and F. M. B. Marques, J. Eur. Ceram. Soc., 2004, 24, 2631–2639. 183 C. S. Chen, B. A. Boukamp, H. J. M. Bouwmeester, G. Z. Cao, H. Kruidhof, A. J. A. Winnubst and A. J. Burggraaf, Solid State Ionics, 1995, 76, 23–28. 184 C. S. Chen, H. Kruidhof, H. J. M. Bouwmeester, H. Verweij and A. J. Burggraaf, Solid State Ionics, 1997, 99, 215–219. 185 J. Kim and Y. S. Lin, J. Membr. Sci., 2000, 167, 123–133. 186 J. Kim and Y. S. Lin, AIChE J., 2000, 46, 1521–1529. 187 K. Wu, S. Xie, G. S. Jiang, W. Liu and C. S. Chen, J. Membr. Sci., 2001, 188, 189–193. 188 V. V. Kharton, A. V. Kovalevsky, A. P. Viskup, F. M. Figueiredo, A. A. Yaremchenko, E. N. Naumovich and F. M. B. Marques, J. Electrochem. Soc., 2000, 147, 2814–2821. 189 B. Wang, J. Yi, L. Winnubst and C. Chen, J. Membr. Sci., 2006, 286, 22–25. 190 J. Yi, Y. Zuo, W. Liu, L. Winnubst and C. Chen, J. Membr. Sci., 2006, 280, 849–855. 191 C. Yang, Q. Xu, C. Liu, J. Liu, C. Chen and W. Liu, Mater. Lett., 2011, 65, 3365–3367. 192 W. Li, J. J. Liu and C. S. Chen, J. Membr. Sci., 2009, 340, 266–271. 193 H. Takamura, H. Sugai, M. Watanabe, T. Kasahara, A. Kamegawa and M. Okada, J. Electroceram., 2006, 17, 741–748. ¨ppers, 194 M. Ramasamy, S. Baumann, J. Palisaitis, F. Schulze-Ku M. Balaguer, D. Kim, W. A. Meulenberg, J. Mayer, R. Bhave, O. Guillon, M. Bram and J. Stevenson, J. Am. Ceram. Soc., 2016, 99, 349–355. 195 A. L. Shaula, V. V. Kharton, F. M. B. Marques, A. V. Kovalevsky, A. P. Viskup and E. N. Naumovich, J. Solid State Electrochem., 2006, 10, 28–40. 196 Y. Lin, S. Fang, D. Su, K. S. Brinkman and F. Chen, Nat. Commun., 2015, 6, 6824. 197 Z. Zhang, W. Zhou, Y. Chen, D. Chen, J. Chen, S. Liu, W. Jin and Z. Shao, ACS Appl. Mater. Interfaces, 2015, 7, 22918–22926. 198 T. Chen, H. Zhao, N. Xu, Y. Li, X. Lu, W. Ding and F. Li, J. Membr. Sci., 2011, 370, 158–165. 199 T. Chen, H. Zhao, Z. Xie, L. Feng, X. Lu, W. Ding and F. Li, Int. J. Hydrogen Energy, 2012, 37, 5277–5285. 200 T. Chen, H. Zhao, Z. Xie, Y. Lu and N. Xu, Int. J. Hydrogen Energy, 2012, 37, 19133–19137. 201 M. B. Choi, S. Y. Jeon, H. J. Hwang, J. Y. Park and S. J. Song, Solid State Ionics, 2010, 181, 1680–1684. 202 Q. Li, X. Zhu, Y. He, Y. Cong and W. Yang, J. Membr. Sci., 2011, 367, 134–140.


View Article Online


Chem Soc Rev

203 X. Zhu, M. Li, H. Liu, T. Zhang, Y. Cong and W. Yang, J. Membr. Sci., 2012, 394–395, 120–130. 204 X. Zhu, H. Liu, Q. Li, Y. Cong and W. Yang, Solid State Ionics, 2011, 185, 27–31. 205 S. B. Adler, Chem. Rev., 2004, 104, 4791–4844. 206 P. Zeng, Z. Chen, W. Zhou, H. Gu, Z. Shao and S. Liu, J. Membr. Sci., 2007, 291, 148–156. 207 Z. Duan, M. Yang, A. Yan, Z. Hou, Y. Dong, Y. Chong, M. Cheng and W. Yang, J. Power Sources, 2006, 160, 57–64. 208 J. Xue, Q. Zheng, Y. Wei, K. Yuan, Z. Li and H. Wang, Ind. Eng. Chem. Res., 2012, 51, 4703–4709. 209 S. Cheng, H. Huang, S. Ovtar, S. B. Simonsen, M. Chen, W. Zhang, M. Søgaard, A. Kaiser, P. V. Hendriksen and C. Chen, ACS Appl. Mater. Interfaces, 2016, 8, 4548–4560. 210 J. E. ten Elshof, H. J. M. Bouwmeester and H. Verweij, Solid State Ionics, 1995, 81, 97–109. 211 J. E. ten Elshof, H. J. M. Bouwmeester and H. Verweij, Solid State Ionics, 1996, 89, 81–92. 212 M. Søgaard, P. Vang Hendriksen and M. Mogensen, J. Solid State Chem., 2007, 180, 1489–1503. 213 X. Zhu, H. Wang and W. Yang, J. Membr. Sci., 2008, 309, 120–127. 214 G. Zhu, X. Fang, C. Xia and X. Liu, Ceram. Int., 2005, 31, 115–119. 215 T. Chen, H. Zhao, Z. Xie, J. Wang, Y. Lu and N. Xu, J. Power Sources, 2013, 223, 289–292. 216 K. Zhang, L. H. Liu, Z. P. Shao, R. Xu, J. C. D. da Costa, S. B. Wang and S. M. Liu, J. Mater. Chem. A, 2013, 1, 9150–9156. 217 Y. Ji, J. A. Kilner and M. F. Carolan, Solid State Ionics, 2005, 176, 937–943. 218 K. Zhang, Z. Shao, C. Li and S. Liu, Energy Environ. Sci., 2012, 5, 5257–5264. 219 S. Imashuku, L. Wang, K. Mezghani, M. A. Habib and Y. Shao-Horn, J. Electrochem. Soc., 2013, 160, E148–E153. 220 L. Wang, S. Imashuku, A. Grimaud, D. Lee, K. Mezghani, M. A. Habib and Y. Shao-Horn, ECS Electrochem. Lett., 2013, 2, F77–F81. 221 C. Zhang, R. Ran, G. H. Pham, K. Zhang, J. Liu and S. Liu, RSC Adv., 2015, 5, 5379–5386. 222 C. Zhang, X. Meng, J. Sunarso, L. Liu, R. Xu, Z. Shao and S. Liu, J. Mater. Chem. A, 2015, 3, 19033–19041. 223 Y. Teraoka, T. Nobunaga and N. Yamazoe, Chem. Lett., 1988, 503–506. 224 V. V. Kharton, V. N. Tikhonovich, L. Shuangbao, E. N. Naumovich, A. V. Kovalevsky, A. P. Viskup, I. A. Bashmakov and A. A. Yaremchenko, J. Electrochem. Soc., 1998, 145, 1363–1373. 225 Y. Teraoka, T. Nobunaga, K. Okamoto, N. Miura and N. Yamazoe, Solid State Ionics, 1991, 48, 207–212. 226 Z. Yang, G. Xia, X. Li and J. Stevenson, Int. J. Hydrogen Energy, 2007, 32, 3648–3654. 227 Ch. Ftikos, M. Nauer and B. C. H. Steele, J. Eur. Ceram. Soc., 1993, 12, 267–270. 228 K. Ahn, D. S. Yoo, D. H. Prasad, H.-W. Lee, Y.-C. Chung and J.-H. Lee, Chem. Mater., 2012, 24, 4261–4267.


Review Article

229 Z. Wu, W. Jin and N. Xu, J. Membr. Sci., 2006, 279, 320–327. ´k and 230 J. Martynczuk, F. Liang, M. Arnold, V. ´ Sepela A. Feldhoff, Chem. Mater., 2009, 21, 1586–1594. 231 H. L. Tuller and A. S. Nowick, J. Phys. Chem. Solids, 1977, 38, 859–867. 232 W. Lai and S. M. Haile, J. Am. Ceram. Soc., 2005, 88, 2979–2997. 233 H. J. Park and G. M. Choi, J. Electroceram., 2006, 17, 781–786. 234 Y. Wei, J. Tang, L. Zhou, J. Xue, Z. Li and H. Wang, AIChE J., 2012, 58, 2856–2864. 235 J. Tang, Y. Wei, L. Zhou, Z. Li and H. Wang, AIChE J., 2012, 58, 2473–2478. 236 Q. Zheng, J. Xue, Q. Liao, Y. Wei, Z. Li and H. Wang, Chem. Eng. Sci., 2013, 101, 240–247. 237 J. Xue, Q. Liao, W. Chen, H. J. M. Bouwmeester, H. Wang and A. Feldhoff, J. Mater. Chem. A, 2015, 3, 19107–19114. 238 Y. Chen, Q. Liao, Y. Wei, Z. Li and H. Wang, Ind. Eng. Chem. Res., 2013, 52, 8571–8578. 239 B. He, G. Li, L. Gui, Y. Ling and L. Zhao, J. Mater. Chem. A, 2016, 4, 4003–4008. 240 D. Ding, X. Li, S. Y. Lai, K. Gerdes and M. Liu, Energy Environ. Sci., 2014, 7, 552–575. 241 Y. Gong, D. Palacio, X. Song, R. L. Patel, X. Liang, X. Zhao, J. B. Goodenough and K. Huang, Nano Lett., 2013, 13, 4340–4345. 242 M. Risch, K. A. Stoerzinger, S. Maruyama, W. T. Hong, I. Takeuchi and Y. Shao-Horn, J. Am. Chem. Soc., 2014, 136, 5229–5232. 243 E. O. Oh, C. M. Whang, Y. R. Lee, S. Y. Park, D. H. Prasad, K. J. Yoon, J. W. Son, J. H. Lee and H. W. Lee, Adv. Mater., 2012, 24, 3373–3377. 244 W. Zhou, F. Liang, Z. Shao and Z. Zhu, Sci. Rep., 2012, 2, 327. 245 H. J. M. Bouwmeester, H. Kruidhof and A. J. Burggraaf, Solid State Ionics, 1994, 72, 185–194. 246 J. H. Joo, K. S. Yun, Y. Lee, J. Jung, C.-Y. Yoo and J. H. Yu, Chem. Mater., 2014, 26, 4387–4394. 247 Y. Hayamizu, M. Kato and H. Takamura, J. Membr. Sci., 2014, 462, 147–152. 248 A. Leo, S. Liu and J. C. Diniz da Costa, J. Membr. Sci., 2009, 340, 148–153. 249 S. Murphy, D. Slade, K. Nordheden and S. Staggwilliams, J. Membr. Sci., 2006, 277, 94–98. 250 Y. Liu, X. Zhu, M. Li, H. Liu, Y. Cong and W. Yang, Angew. Chem., Int. Ed., 2013, 52, 3232–3236. 251 T. Hong, L. Zhang, F. Chen and C. Xia, J. Power Sources, 2012, 218, 254–260. 252 Y. Teraoka, Y. Honbe, J. Ishii, H. Furukawa and I. Moriguchi, Solid State Ionics, 2002, 152–153, 681–687. 253 W. Ito, T. Nagai and T. Sakon, Solid State Ionics, 2007, 178, 809–816. 254 A. V. Kovalevsky, V. V. Kharton, F. Maxim, A. L. Shaula and J. R. Frade, J. Membr. Sci., 2006, 278, 162–172. 255 A. Vivet, P. M. Geffroy, E. Thune, C. Bonhomme, F. Rossignol, N. Richet and T. Chartier, J. Membr. Sci., 2014, 454, 97–108.

Chem. Soc. Rev.

View Article Online


Review Article

256 J. Garcia-Fayos, V. B. Vert, M. Balaguer, C. Solı´s, C. Gaudillere and J. M. Serra, Catal. Today, 2015, 257, 221–228. 257 Z. Zhang, D. Chen, Y. Chen, Y. Hao, M. O. Tade and Z. Shao, J. Membr. Sci., 2014, 472, 10–18. 258 Z. Zhang, D. Chen, Y. Gao, G. Yang, F. Dong, C. Chen, F. Ciucci and Z. Shao, RSC Adv., 2014, 4, 25924–25932. 259 I. Garcıá-Torregrosa, M. P. Lobera, C. Solı´s, P. Atienzar and J. M. Serra, Adv. Energy Mater., 2011, 1, 618–625. 260 G. L. Messing, S.-C. Zhang and G. V. Jayanthi, J. Am. Ceram. Soc., 1993, 76, 2707–2726. 261 B. He, K. Zhang, Y. Ling, J. Xu and L. Zhao, J. Membr. Sci., 2014, 464, 55–60. 262 K. Zhang, C. Zhang, L. Zhao, B. Meng, J. Liu and S. Liu, Energy Fuels, 2016, 30, 1829–1834. 263 O. A. Savinskaya and A. P. Nemudry, J. Membr. Sci., 2014, 459, 45–51. 264 G. Y. Meng, H. Z. Song, H. B. Wang, C. R. Xia and D. K. Peng, Thin Solid Films, 2002, 409, 105–111. 265 S. Uhlenbruck, N. Jordan, D. Sebold, H. P. Buchkremer, ¨ver, Thin Solid Films, 2007, V. A. C. Haanappel and D. Sto 515, 4053–4060. 266 X. Meng, W. Ding, R. Jin, H. Wang, Y. Gai, F. Ji, Y. Ge and D. Xie, J. Membr. Sci., 2014, 450, 291–298. 267 O. Buchler, J. Serra, W. Meulenberg, D. Sebold and H. Buchkremer, Solid State Ionics, 2007, 178, 91–99. 268 C.-C. Chen and R. Prasad, US Pat., 6165553, 2000. 269 D. Marcano, G. Mauer, Y. J. Sohn, R. Vaßen, J. Garcia-Fayos and J. M. Serra, J. Membr. Sci., 2016, 503, 1–7. 270 N. Zotov, S. Baumann, W. A. Meulenberg and R. Vaßen, J. Membr. Sci., 2013, 442, 119–123. 271 A. J. Burggraaf and Y.-S. Lin, US Pat., 5160618, 1992. 272 R. Prasad and C. F. Gottzmann, US Pat., 5547494, 1996. 273 M. S. Chen and P. J. Cook, US Pat., 5035726, 1991. 274 R. M. Thorogood, R. Srinivasan, T. F. Yee and M. P. Drake, US Pat., 5240480, 1993. 275 T. F. Yee, R. S. Srinivasan and R. M. Thorogood, US Pat., 5447555, 1995. 276 D. Kang, R. M. Thorogood, R. J. Allam, A. K. J. Topham and S. L. Russek, US Pat., 5565017, 1996. 277 D. Kang, S. L. Russek, R. Agrawal, D. D. Brengel and E. P. Foster, US Pat., 5562754, 1996. 278 C. F. Gottzman and R. Prasad, US Pat., 5820654, 1998. 279 R. Prasad, C. F. Gottzmann and R. F. Drnevich, US Pat., 5852925, 1998. 280 R. Prasad and C. F. Gottzmann, US Pat., 5733435, 1998. 281 A. V. Virkar, C. F. Gottzmann, R. Prasad and B. A. van Hassel, US Pat., US 6368383 B1, 2002. 282 M. S. Chen, W. P. Hegarty and W. A. Steyert, US Pat., 5174866, 1992. 283 D. Kang, R. S. Srinivasan, R. M. Thorogood and E. P. Foster, US Pat., 5516359, 1996. 284 M. F. Carolan and P. N. Dyer, US Pat., 5534471, 1996. 285 D. M. Taylor, J. D. Bright, M. F. Carolan, R. A. Cutler, P. N. Dyer, E. Minford, D. W. Prouse, R. E. Richards, S. L. Russek and M. A. Wilson, US Pat., 5681373, 1997.

Chem. Soc. Rev.

Chem Soc Rev

286 P. N. Dyer, J. D. Bright, M. F. Carolan, E. Minford, R. E. Richards, S. L. Russek, D. M. Taylor and M. A. Wilson, US Pat., 5599383, 1997. 287 S. L. Russek, J. A. Knopf and D. M. Taylor, US Pat., 5753007, 1998. 288 N. R. Keskar, R. Prasad and C. F. Gottzmann, US Pat., 5964922, 1999. 289 R. Prasad and C. F. Gottzmann, US Pat., 5935298, 1999. 290 R. Prasad and C. F. Gottzmann, US Pat., 5976223, 1999. 291 C. F. Gottzmann, R. Prasad, J. M. Schwartz, V. E. Bergten, J. E. White, T. J. Mazanec, T. L. Cable and J. C. Fagley, US Pat., 6139810, 2000. 292 M. F. Carolan, P. N. Dyer, J. M. LaBar and R. M. Thorogood, US Pat., 5240473, 1993. 293 M. F. Carolan, P. N. Dyer, S. M. Fine, J. M. Labar and R. M. Thorogood, US Pat., 5269822, 1993. 294 D. Kang, R. M. Thorogood, R. J. Allam and A. K. J. Topham, US Pat., 5657624, 1997. 295 R. F. Drnevich, M. M. Shah and C. F. Gottzmann, US Pat., US6293084 B1, 2001. 296 C. F. Gottzmann, US Pat., US6562105 B2, 2003. 297 R. Prasad and C. F. Gottzmann, US Pat., US6296686 B1, 2001. 298 L. E. Bool III, H. Junction, H. Kobayashi and P. Valley, US Pat., US6382958 B1, 2002. 299 L. E. Bool III, H. Junction and H. Kobayashi, US Pat., US6394043 B1, 2002. 300 S. Nataraj, R. B. Moore and S. L. Russek, US Pat., 6048472, 2000. 301 S. Nataraj, S. L. Russek and P. N. Dyer, US Pat., 6114400, 2000. 302 S. Nataraj and S. L. Russek, US Pat., US6214066 B1, 2001. 303 P. N. Dyer, M. F. Carolan, D. P. Butt, R. H. E. V. Doorn and R. A. Cutler, US Pat., US6492290 B1, 2002. 304 R. Prasad, J. M. Schwartz, E. T. Robinson and C. F. Gottzmann, US Pat., US6695983 B2, 2004. 305 E. H. Shreiber, E. H. xvan Hassel and R. Prasad, US Pat., US7118612 B2, 2006. 306 S. M. Kelly, B. R. Kromer, M. M. Litwin, L. J. Rosen, G. M. Christie, J. R. Wilson, L. W. Kosowski and C. Robinson, US Pat., US9.238201 B2, 2016. 307 J. M. Repasky, US Pat., US9028720 B1, 2015. 308 M. F. Carolan, P. N. Dyer, K. B. Dyer, M. A. Wilson, T. R. Ohm, K. E. Kneidel, D. Peterson, C. M. Chen and K. G. Rackers, US Pat., US7279027 B2, 2007. 309 J. M. Repasky, M. F. Carolan, V. E. Stein and C. C. MingPoh, US Pat., US2008/0302013 A1, 2008. 310 M. F. Carolan and C. F. Miller, US Pat., US 2008/0168901A1, 2008. 311 J. C. Chen, C. J. Besecker, H. Chen and E. T. Robinson, US Pat., US7229537 B2, 2007. 312 N. Nagabhushana, J. A. Lane, G. M. Christie and B. A. van Hassel, US Pat., US7556676 B2, 2009. 313 G. M. Christie and J. A. Lane, US Pat., US 8795417 B2, 2014. 314 H. Chen, J. C. Chen and W. Wang, US Pat., US7338624 B2, 2008.


View Article Online


Chem Soc Rev

315 J. D. Sirman and J. C. Chen, US Pat., US6514314 B2, 2003. 316 C.-C. Chen, R. Prasad and C. F. Gottzmann, US Pat., 5938822, 1999. 317 M. S. Kleefisch, C. A. Udovich, J. G. Masin and T. P. Kobylinski, US Pat., US 6200541 B1, 2001. 318 G. M. Christie, J. R. Wilson and B. A. V. Hassel, US Pat., US8323463 B2, 2012. 319 X. Zhu and W. Yang, Mixed conducting ceramic membranes, Springer-Verlag Berlin Heidelberg, Berlin, Germany, 1st edn, 2017. 320 L. L. Anderson, P. A. Armstrong, R. R. Broekhuis, M. F. Carolan, J. Chen, M. D. Hutcheon, C. A. Lewinsohn, C. F. Miller, J. M. Repasky, D. M. Taylor and C. M. Woods, Solid State Ionics, 2016, 288, 331–337. 321 T. Foster, Air Separation Technology-Ion Transport Membrane (ITM), Air products and Chemicals Inc., Allentown, 2008. 322 J. Repasky, D. McCarthy, P. Armstrong and M. Carolan, Workshop on Technology Pathways Forward for Carbon Capture & Storage on Natural Gas Power Systems, Washington DC, 2014. 323 A. Kaletsch, E. M. Pfaff and C. Broeckmann, M. Modifell, N. Nauels, Pilot module for oxygen separation with BSCF membranes. In: Proceedings of the 2nd ICEPE, Frankfurt am Main, 2011. 324 E. M. Pfaff, A. Kaletsch and C. Broeckmann, Chem. Eng. Technol., 2012, 35, 455–463. 325 J. Zhou, X. Tang, C. Wu, H. Wang, Y. Zhang, W. Ding, Y. Jin and C. Sun, J. Phys. Chem. C, 2015, 119, 24229–24237. 326 B. C. H. Steele, Mater. Sci. Eng., B, 1992, 13, 79–87. 327 S. J. Benson, D. Waller and J. A. Kilner, J. Electrochem. Soc., 1999, 146, 1305–1309. 328 D. Waller, J. A. Lane, J. A. Kilner and B. C. H. Steele, Mater. Lett., 1996, 27, 225–228. 329 J. S. Ledford, Y.-M. Kim, M. Houalla, A. Proctor and D. M. Hercules, Analyst, 1992, 117, 323–327.


Review Article

330 D. Renusch, B. Veal, K. Natesan and M. Grimsditch, Oxid. Met., 1996, 46, 365–381. 331 P. R. Griffiths and J. A. de Haseth, Fourier Transform Infrared Spectrometry, John Wiley & Sons, Inc., Hoboken, New Jersey, 2nd edn, 2007. 332 S. Faraji, K. J. Nordheden and S. M. Stagg-Williams, Appl. Catal., B, 2010, 99, 118–126. 333 S. Engels, T. Markus, M. Modigell and L. Singheiser, J. Membr. Sci., 2011, 370, 58–69. 334 R. Zhang, H. Alamdari and S. Kaliaguine, Appl. Catal., A, 2008, 340, 140–151. 335 K. H. Stern and E. L. Weise, High temperature properties and decomposition of inorganic salts, Part1. Sulfates, NSRDS, Washington, 1966. 336 R. L. Wilfong, L. P. Domingues and L. R. Furlong, J. Am. Ceram. Soc., 1964, 47, 240–241. 337 J. Gao, L. Li, Z. Yin, J. Zhang, S. Lu and X. Tan, J. Membr. Sci., 2014, 455, 341–348. 338 H. Stadler, F. Beggel, M. Habermehl, B. Persigehl, R. Kneer, M. Modigell and P. Jeschke, Int. J. Greenhouse Gas Control, 2011, 5, 7–15. 339 Y. Liu, X. Zhu, M. Li, W. Li and W. Yang, J. Membr. Sci., 2015, 492, 173–180. 340 M. M. Viitanen, R. G. V. Welzenis, H. H. Brongersma and F. P. F. van Berkel, Solid State Ionics, 2002, 150, 223–228. 341 K. Hilpert, D. Das, M. Miller, D. H. Peck and R. Weiß, J. Electrochem. Soc., 1996, 143, 3642–3647. 342 Y. M. Kim, X. Chen, S. P. Jiang and J. Bae, Electrochem. Solid-State Lett., 2011, 14, B41. ¨dl, E. Bucher, A. Egger, P. Kreiml, C. Teichert, 343 N. Schro ¨schen and W. Sitte, Solid State Ionics, 2015, 276, T. Ho 62–71. 344 F. Shen and K. Lu, Electrochim. Acta, 2016, 211, 445–452. 345 Y. Chen, B. Qian, S. Li, Y. Jiao, M. O. Tade and Z. Shao, J. Membr. Sci., 2014, 449, 86–96.

Chem. Soc. Rev.