Final Report Summary - NASA-OTM (Nanostructured surface activated ultra-thin oxygen transport membrane)
Executive summary:
Oxygen transport membranes (OTMs) are able to separate oxygen from air with high efficiency particularly if heat can be provided by the targeted application. Therefore, the main goal of this project is the development of gastight oxygen separation membranes for oxygen generation or the use in catalytic membrane reactors, which are characterized by:
1. high oxygen flux of 10 ml cm-2 min-1,
2. infinite oxygen selectivity and
3. high membrane stability.
In order to address all of these goals at the same time, the use of nano-structured surface activated ultra-thin OTMs is proposed. The development was done stepwise by developing firstly 'thin film membranes' (d = 20 to 50 µm) and subsequently the targeted 'ultra-thin film membranes' (d around 1 µm).
Three different applications are in focus:
1. oxyfuel carbon capture and storage (CCS) power plants requiring pure oxygen,
2. oxidative coupling of methane (OCM) as a known reaction using available catalysts and requiring high membrane stability in reducing atmospheres and
3. hydrogen cyanide (HCN) synthesis via the Andrussow reaction (AR), where the use of a membrane reactor is rarely investigated so far and thus novel catalysts are necessary while high membrane stability in reducing and acid atmospheres is required.
In order to meet the specific requirements of the targeted applications two membrane material classes are being investigated:
1. perovskites, which possess high permeability but limited stability for pure oxygen generation (oxyfuel) and
2. fluorites, which possess low permeability but high stability particularly in reducing atmospheres for the chemical applications (OCM and AR).
Perovskite membranes were developed based on Ba0.5Sr0.5Co0.8Fe0.2O3-d (BSCF) and La0.58Sr0.4Co0.2Fe0.8O3-d (LSCF). BSCF is the material with the highest permeability known so far whereas LSCF has lower permeability but higher stability. Surface activated tape cast thin film BSCF membranes revealed that the rate limiting step was no longer the bulk diffusion or the surface exchange but the concentration polarisation in the support pores. This effect occurs when using sweep gas to remove the oxygen from the membrane. An optimisation of the support porosity to 41% enhancing the support's binary gas diffusibility increased the permeation rate further, but did not prevent rate limitation by the support. Avoiding a sweep gas at the support side led to the highest permeation rates ever reported in literature and attracted much interest in the research community, accordingly. In LSCF thin film membranes this effect was much smaller due to the lower permeation rate, but the relative permeation enhancement compared to a 1 mm thick pellet is accordingly higher. Fluorite membranes were developed based on doped ceria. Tape cast thin film membranes showed very high yield in the OCM reaction. Although not all targets could be met at the same time, e.g. partly due to geometrical reasons using planar membranes, high potential of these membranes could be shown also in other catalytic processes. In AR, however, the experiments failed mainly because of undesired by-reactions possibly catalysed by the membrane material itself. More detailed studies on the catalysis of this reaction are required.
Both material classes were successfully applied as ultra-thin films using magnetron sputtering (MS). This required the development of appropriate substrates and interlayers providing a fine structured surface. LSCF (d = 0.7 µm) and cerium gadolinium oxide (CGO, d around 1 to 2 µm) were sputtered on a zirconia substrate with a thin ceria interlayer showing sufficient gas-tightness. However, permeation measurements revealed that the microstructure of the substrate is inappropriate due to the small pores of 80 to 100 nm across the whole support thickness of 2 to 2.5 µm. This leads to severe concentration polarisation as could be expected after the project results on thin film membranes. Finally, a metallic support with higher permeability and ceramic interlayers providing sufficient surface structure could be developed and gas-tight CGO layers applied. Detailed functional analysis of this very promising architecture has yet to be done.
Summarising, the project clearly shows the high potential of perovskitic thin film membranes for pure oxygen generation revealing that a further decrease in thickness to the µm range is not reasonable. In catalytic membrane reactors the metal supported ultra-thin film CGO membrane developed in this project is very promising, but still needs further investigation. In general, more attention has to be paid on the complex transport behaviour in such multi-layered membrane assemblies by experimental and modelling efforts.
Project context and objectives:
All scenarios about future global energy requirements anticipate an increasing demand for electricity, which in 2030 is predicted to be twice the current demand. These scenarios also indicate that this increase can only be met by considerable use of the fossil fuels coal, natural gas and oil. It becomes clear that fossil fuels will contribute to more than 60 % of future electricity generation. However, fossil fuel power plants are by far the biggest point sources of carbon dioxide (CO2) and contribute more than 40% of the worldwide anthropogenic CO2 emissions. Therefore, CCS is an important strategy to reduce CO2 emissions in order to counteract global warming and thus, with regard to Kyoto Protocol and the Energy Technology Strategic Plan of the European Union (EU), makes a key contribution to allow future electricity supply to be environmentally safe and sustainable. Efficient CO2 capture is possible using the oxyfuel technology. In an oxyfuel power plant, partly recirculated flue gas is enriched in oxygen and used to combust a carbon containing fuel forming primarily CO2 and water (H2O). This makes it much easier and cheaper to capture the CO2 than by using air for combustion. The required oxygen can be produced using ceramic OTMs, which are associated with significantly lower efficiency losses compared to conventional cryogenic separation technologies. OTMs consist of gastight mixed ionic electronic conductors (MIEC), which allow oxygen diffusion through vacancies in the crystal lattice and simultaneous transport of electrons in the opposite direction, thus obviating the need for an external electrical short circuit. However, using membranes in this synthesis has not been investigated so far.
In order to decrease the required membrane area a maximisation of the oxygen permeation rate is an evident requirement. This can be done by the choice of a material with high permeability. However, it is common knowledge that a trade-off between permeability and stability exist so that this approach has only limited practicability. Therefore, thin film membranes are developed in this project since the permeation rate is inversely proportional to the membrane thickness.
The project is organised in five scientific work packages (WPs). WPs one to three are related to the (active) OTM layer, the porous substrate and catalyst surface layers, respectively. The research is supported by modelling in WP4, i.e. thermo-mechanical and ab-initio transport modelling. Application-oriented evaluation (WP5) of the new membranes takes oxyfuel technology in fossil power plants as well as highly selective catalytic membrane applications into account. This choice of applications represents three different approaches, namely pure oxygen generation in 'clean' conditions, a relatively well known catalytic chemical reaction (OCM) in close contact to the membrane surface and a hardly investigated catalytic chemical reaction in close contact to the membrane surface.
At the start of the project, first generation materials for membranes (fluorite + perovskite), substrates (metallic + ceramic) and catalysts (for each application) had to be defined acting as reference materials. WP1 and WP2 work in close cooperation on the stepwise development of different types of planar lab scale membranes: bulk membranes (0.5 to 1 mm) were fabricated for characterisation of bulk properties and development of next generation materials. In parallel, thin film asymmetric membranes (d around 20 to 50 µm) using tape casting were manufactured. These membranes were used for e.g. catalysts testing (WP3) and first application-oriented tests (WP5). The development of the desired ultra-thin film membranes (i.e. d = 1µm or below) was gradually performed during the project with increasing effort and concentrating on most promising manufacturing technologies. Different substrate materials including ceramics, metals and cermets had to be investigated in order to obtain sufficient mechanical stability. Furthermore different deposition techniques with low deposition temperature are used in order to obtain compressive stresses in the membrane layer at application conditions. For both aims thermo-mechanical modelling supported the development. The supports needed to fulfil the following requirements:
1. high mechanical strength and chemical stability to support the active, ultra-thin membrane and to withstand application and testing conditions,
2. sufficient porosity to provide gas access to the membrane layer and to allow easy infiltration of active catalyst materials and
3. an adequate surface finish on one side of the supports to allow the application of ultra-thin membrane layers. This requires the development of graded interlayers.
WP3 is dedicated to the development of highly effective and stable catalysts in the targeted applications as well as processing of surface layers by bulk and thin film membrane samples. After testing the catalysts properties itself, activated membrane assemblies are delivered to WP5 in order to test the effectiveness of the developed catalysts. Catalytic coatings should allow increasing the oxygen permeance of thin-films at 750 ºC by at least a factor of four in comparison to not activated membranes (MS3) with an intermediate target of 5 ml cm-2 min-1 oxygen permeation rate (MS6). The final goal is the development of catalytical promotion of ultra-thin film membranes, which reach oxygen permeation values of 10 ml cm-2 min-1 with a degradation of 1 % per 1 000 hours and oxygen purity higher than 99 %.
WP4 starts modelling activities from the beginning with the knowledge of first generation materials. 'Thermo-mechanical behaviour of membranes' supports WP1 and WP2 concerning thermo-mechanical stability of ultra-thin film membranes. Finite element modelling of layered structures allows the derivation of an assessment procedure for the fracture stress of the supporting substrate material on the basis of experimental results as a measure of its robustness. Furthermore, a comparison of planar and tubular membrane set-ups shall permit to conclude which arrangement is favourable with regard to the respective stress distribution. The analysis utilises finite element simulations and is based on critical failure criteria and relevant mechanical data.
'Modelling of surfaces transport properties' is working on the improvement of first generation perovskite material by atomistic modelling of transport and surface exchange properties. The first principles large scale computer simulations will focus on the relation of chemical composition and functional properties. Thermodynamic stability of perovskites has to be analysed as a function of oxygen non-stoichiometry. Detailed quantum chemical calculations of the oxygen vacancy formation, and migration energies in a series of compositions, lead to the determination of major factors and compositions enabling fast oxygen transport. This shall lead to the prediction of a material composition increasing the oxygen flux through the membrane by a factor of two to three (MS7), which has to be confirmed by experiments.
WP5 evaluates the developed membranes regarding the targeted applications, i.e. oxyfuel, OCM and HCN synthesis. Testing of specimens as well as economical assessment and ecological evaluations is performed including life cycle assessment (LCA).
'Oxyfuel power plants' focuses on the system integration of an OTM membrane based air separation unit (OTM-ASU) into a CO2-emission-free fossil oxyfuel power plant. As bench mark for the assessment of the economic competitiveness of the OTM technology the conventional air separation by cryogenic distillation (cryogenic ASU) was chosen. By simulation of the OTM process the derivation of optimal operation conditions with respect to the membrane performance, a conceptual module design and the integration need are aimed in order to perform an economic assessment. An ecological analysis of a membrane-based oxyfuel power plant was performed LCA. The environmental impacts along production, use and disposal of membrane power plants were compared to oxyfuel processes using cryogenic air separation. LCA results are used to identify those processes which contribute most to environmental impacts. By comparing the membrane with conventional technology benchmarks for further membrane development were defined.
Selected chemical applications of technical interest were in focus of OCM and HCN synthesis. For both applications membrane catalyst assemblies (MCAs) following the most favourable concepts has to be fabricated with membranes provided by WP1 using adopted OCM or dedicated developed AR catalysts. Main task however is the installation of the MCAs in specially adopted membrane reactors, the implementation in suitable laboratory scale test facilities and the test execution of oxidative methane conversions in order to evaluate the technical potentials of the membrane reactor approach.
Project results:
WP1: Membrane materials, manufacture and characterisation
The main task of WP1 was to obtain gastight membrane layers on porous supports. Membrane materials were defined in accordance with all partners in WP5 and relevant materials properties were determined or collected from partners if accessible. Perovskites as well as doped ceria were considered. Moreover, mechanical characterisation was done with regard to materials' bulk properties as well as the properties of porous supports and asymmetric membrane assemblies. Membrane materials as well as promising layered membrane assemblies were delivered to application-oriented testing (WP5).
Membrane materials
At the kick-off meeting all partners agreed on the materials considered, i.e. LSCF and CGO. Later BSCF was additionally taken into account particularly for thin film asymmetric membranes. Since all these materials are well known by the partners involved, reliable data of the relevant materials properties are accessible.
The atomistic modelling of transport properties of membrane materials were also analysed. With regard to MS7 it was recommended to increase the strontium (Sr) content in LSCF. Respective experiments showed the perspective of that approach revealing that already an increase of the Sr content from 40 to 60 % A-site occupation is sufficient for this goal. Comparison of experiments and calculations is subject to a publication which is under submission. On the other hand other properties change simultaneously and have to be considered in future membrane development, e.g. thermal expansion coefficient.
Membrane layer processing
Thin-film asymmetric membranes were intended for the evaluation of catalytic layers (WP3) and as fall-back position in the risk management. Asymmetric membranes of a porous CGO support and a thin film membrane (thickness of 20 to 30 µm) were prepared by tape casting, lamination and co-sintering. These thin film membranes were used for first tests in the chemical applications (OCM and HCN). The first generation of asymmetric CGO membranes were mechanically weak and warped and were difficult to mount into testing rigs. Major advancements in processing of these membranes (strength and flatness) were achieved during the project due to CGO support optimisation and by using optical dilatometry, which allows the observation of shape evolution of the bilayers during high temperature sintering.
The sintering study for asymmetric membranes containing cobalt oxide in both CGO layers suggests an sintering cycle with a final temperature of about 1 050 °C for achieving flat, defect free bilayers with low failure rate whereas the study on the mechanical properties of the porous CGO supports shows that higher temperatures of about 1 150°C should be chosen to improve the mechanical properties of the support. As this contradiction could not be solved easily with the existing combination of layers an additional sintering aid (Fe2O3) was introduced into the thin membrane layer. The use of Fe2O3 allowed an increase in the final sintering temperature to 1 150 °C, reducing the warpage of the bilayer due to lower sintering activity of CGO membrane layer with Fe3O4 at this temperature. Such membranes (with Fe3O4 in the membrane layer and Co3O4 in the support layer) have a good combination of flatness and mechanical strength.
The optimisation of the sintering profile for CGO bilayers by using sintering aids allowed a significant reduction in sintering temperature, resulting in improved sample flatness, a reasonable residual porosity (around 35 %) and a reduced loss of samples during co-sintering.
first generation thin film asymmetric perovskite membranes were manufactured from BSCF. The highest oxygen permeation rates ever published could be obtained meeting the targets of MS3, which is described in WP3. The membranes were manufactured via sequential tape casting, in which the membrane layer is cast first. After drying, the support is cast, which consist of the same slurry but obtaining pore former (corn starch) leading to significant porosity in the support. Finally, a co-sintering step is carried out. The final (sintered) thicknesses of the membrane layer and the support are approximately 60 µm and 800 µm, respectively. It turned out that the permeation through the membrane layer itself is that high that molecular gas diffusion through the support's pores becomes rate limiting second generation thin film asymmetric membranes were developed in two steps. Firstly, a more porous BSCF support (41 % instead of 34 % porosity) was used. In addition, thinner BSCF membrane layers (approximately 20 µm instead of 60 µm thickness) were developed.
Ultra-thin film asymmetric membranes were manufactured via physical vapour deposition (PVD), i.e. MS. Sol-gel as well as chemical vapour deposition (CVD) was also evaluated but assessed as less promising in the given time frame.
CGO layers are coated via reactive MS. A metallic target of the respective alloy Ce0.8Gd0.2 is used. During sputtering a constant oxygen gas flow is fed to the high vacuum sputtering chamber. Therefore, the oxide CGO is formed on the substrate. The substrate can be heated up to 800 °C, which results into denser layers. Moreover, a bias power up to 400 W can be applied accelerating the Ar+-ions of the sputtering plasma onto the substrate. This ion bombardment leads to a further densification of the coating due to the kinetic energy of the Ar+-ions, however, resulting in residual compressive stresses in the layer, which can lead to delamination. These phenomena were investigated in detail and a four-zone model was derived. Using appropriate parameters a gas-tight film could be achieved, i.e. a third generation ultra-thin film membrane. A respective publication is in revision.
After this principal investigation of the coating technology and the parallel development of metallic substrates with appropriate ceramic interlayers in WP2 the latter were coated using reactive MS. The strength of the substrate and particularly the interlayer is given by manufacturing and cannot easily be improved by sintering. But, based on the experience gained in the project it was finally possible to develop a crack-free CGO top layer. This is the first time that a gas-tight ultra-thin film metal supported OTM was successfully produced. A respective publication is in preparation.
For the perovskite LSCF MS was performed using a ceramic target. Nice perovskite layers of approximately 700 nm thickness could be obtained on porous zirconia (8YSZ) substrates with a porous CGO interlayer. But, an accordant oxygen permeation measurement revealed that less oxygen permeated through the membrane assembly than through a dense LSCF disc. This is due to the fact that the porous support is relatively thick (2.5 mm) and consists of very tiny pores, i.e. 80 to 100 nm. This fine microstructure is on the one hand necessary for the application of such ultra-thin layers, but on the other hand evidently leads to severe concentration polarisation in the support, which was already observed in the thin film membranes using a thinner (0.9 mm) support with much larger pores (around 1 µm).
Mechanical evaluation of ultra-thin oxygen membrane structures
After analysing the basic mechanical properties of the membrane materials such as fracture stress, elastic modulus and hardness the emphasis was laid on the mechanically dominating porous substrate variants. Creep deformation and creep rupture was additionally analysed in order to determine component lifetime. Various methods such as creep rupture time, slow crack growth sensitivity of strength data and strength-probability-time (S-P-T) diagrams have been implemented. The experimental work covered the ceramic materials CGO, BSCF and LSCF as well as the alloy FeCrAlY.
Assuming linear elastic behaviour elastic modulus and fracture stress determines the fracture strain. Particularly fracture stress is comparable for the ceramics whereas the yield strength of the FeCrAlY is 'as expected' high at room temperature and strongly decreasing at elevated temperatures. But, at high temperatures a linear elastic consideration is not valid since creep phenomena might occur especially for the metal. However, such creep phenomena make a final validation complicated because they on the one hand are able to reduce residual or applied stresses, but on the other hand creep strain can lead to failure if the respective fracture stress is exceeded.
Dense LSCF can survive the highest stresses, whereas dense CGO seems to be least tolerant against such a load. Interestingly, the tolerable constant stress at room temperature is lower for dense CGO than for porous CGO. This effect is caused by an approximately two times higher sensitivity to sub-critical crack growth in combination with a very high scatter in fracture stress leading to low reliability.
WP2: Substrate materials, manufacture and characterisation
In WP2 two types of support structures for use in asymmetric membrane structures were developed, namely thin-film asymmetric membranes consisting of a porous support and a 20 to 30 µm thin-film, active membrane layer, both basically of the same material; and porous supports with graded interlayers for deposition of ultra-thin membranes with a thickness of 1 to 2 µm. The plain supports and asymmetric thin film membrane structures were prepared for initial catalysts testing (WP3) and application-oriented tests (WP5) of the thin film membranes. The predominant objective of WP2 was to design, prepare and characterise porous supports (both ceramic and metal) which were suitable for the application of ultra-thin membrane layers in WP1. During the project the processing of the supports was optimised in order to achieve suitable properties, such as the microstructure (porosity and pore-size), mechanical properties and lifetime. For ultra-thin film application of membrane layers the surface finish of the support structure played a key role and was refined by deposition of interlayers with graded porosity and reduced surface roughness. The asymmetric membranes were fabricated from CGO or BSCF, while for the porous supports CGO or FeCrAl were used.
Process optimisation for CGO porous supports
In WP2 porous CGO support layers with a thickness between 0.2 and 0.5 mm were made by tape casting, using graphite as pore former to introduce open porosity larger than 25 % after sintering. For preparation of asymmetric CGO structures (membrane bilayers consisting of 0.5 mm porous support and 30 µm active membrane layer), additionally thin CGO layers were tape casted. The tapes were then laminated together, debindered and co-sintered to form the CGO bilayers. The fabrication of high quality CGO support structures that are suitable for ultra-thin film (PVD) and thin film membrane application (co-sintering) required several optimisation steps. Typical defects included delamination of layers, warpage during sintering, surface scratches from the sintering plates and large irregular pores from the flaky graphite used as pore-former. All of these defects make the supports mechanically weak, difficult to handle and test.
Throughout the project continuous improvements in the quality of the CGO supports was achieved by detailed analyses of the ceramic processing, including optimisation of the slurry composition, the ceramic dispersion, the pore former size and shape to get more uniform pore size distribution. Sup-ports with improved strength and reduced warpage were produced by using a single casting process to make thicker films. In addition, the sintering process was investigated in more detail5 in order to optimise the microstructure of the supports with respect to the open porosity and mechanical properties. Here the pore-former (graphite) content was varied from 8 wt% to 11 wt% and the sintering temperature from 1 050 °C to 1 250 °C. The resulting microstructures can be seen in Figure 12 along with their open porosity values. The samples sintered at 1 050 °C had very high porosity and poor mechanical proper-ties, while those sintered at the high temperature of 1 250 °C had insufficient porosity. The conditions of 11 % (high) graphite content and a processing temperature of 1 150 °C produced a good open porosity of 45 % and adequate mechanical properties (strength = 46.7 ± 2 MPa, E-modulus = 45.6 ± 7.5 MPa). The Weibull modulus (a measure of the reliability) of the latest generation of samples increased to five as a result of better processing (less defects in the samples). This value is still low, even for ceramic materials, which is potentially problematic for applications involving the CGO membranes where there is a high risk of breakage during sealing, testing and changing gas atmospheres (redoxing).
Asymmetric thin-film CGO membranes
Asymmetric membranes of a porous CGO support and a thin film membrane (thickness of 20 to 30 µm) were prepared by tape casting, lamination and co-sintering. These thin film membranes were used for first tests in the chemical applications (OCM and HCN reactions, WP5). The first generation of asymmetric CGO membranes were mechanically weak and warped and were difficult to mount into testing rigs. Major advancements in processing of these membranes (strength and flatness) were achieved during the project due to CGO support optimisation and by using optical dilatometry.
Metallic porous supports An alumina-forming ferritic steel powder of the composition Fe21Cr7Al1Mo0.5Y was specifically designed for the NASA-OTM project and its properties were extensively characterised during the project. Metal supports potentially have advantages over ceramic supports with respect to cost, module integration and superior mechanical properties. However, the lifetime of porous metal supports could be limited by corrosion and propensity for creep, both effects are more severe with temperature and porosity. Corrosion data are presented in the following section with discussion of design requirements to ensure a long lifetime for the desired applications.
It is observed that the lower porosity samples have a lower weight gain, this is due to the smaller pores being blocked by the growth of the oxide scale. In order to maintain an open structure with good gas permeability over the life of the device, the following geometrical parameters are recommended, i.e. an initial porosity of around 40% with the smallest pores being much larger than the dimensions of the oxide scale. Oxidation of chromia-forming steel was also undertaken under the same conditions and it was confirmed that the alumina-forming steel had a lifetime around five times that of the FeCr. Modelling of the porous support lifetime as a function of the operating temperature, surface area and oxidation rate determined from experiments, showed that a lifetime of over 30 000 hours could be achieved operating at 750 °C. This recommended operating temperature is also preferable with respect to the mechanical properties, where the creep rate can be reduced to the target value of 1 % per year under these conditions (see WP1).
Metal support tapes with a thickness of 0.3 to 0.5 mm were fabricated from the FeCrAl powder by tape-casting and sintering. A water based tape casting process was developed, avoiding many environmental and health risks associated with organic solvents. This process was optimised to achieve tapes with good flatness and homogeneity and few defects. Sintering was performed at 1250 °C in a protective atmosphere to prevent excessive corrosion. Enough oxygen was present in sintering atmosphere to provide 'pre-oxidation' of the supports during sintering.
Substrates with graded interlayers for ultra-thin film asymmetric membranes
The as-prepared CGO and metal supports required some surface finish layers to reduce significantly the larger pore sizes and surface roughness to allow the deposition of defect-free dense ultra-thin membranes. The as-sintered CGO10 supports had a sub-micron surface roughness and maximum size of surface pores and defects around 2 µm. The metal supports had an average surface roughness of 3 to 4 µm with pore sizes over 10 µm due to the large particle size (10 to 38 µm) of the metal powder. Metal supports required two interlayers, i.e. a first coarser ceramic layer, then a second fine finishing layer. The CGO support required only the final CGO finishing layer.
Development of the first interlayer for the metal supports was very challenging and various interlayer materials and fabrication techniques, e.g. spray coating, screen-printing, dip coating and electrophoretic deposition (EPD) were investigated. It proved difficult to find ceramic materials that were stable and did not react with the metal support under the sintering conditions required for the metal support. Eventually, alumina was selected as the first interlayer material, as it is chemically stable under the sintering conditions, does not react with the metal support, adheres well and maintains a fine microstructure. The EPD technique is very suitable for this deposition as high solids-loaded films can be deposited and was up-scaled during the project to fabricate large batches of 25 cm2 samples. Deposition of a final alumina interlayer (also by EPD) is being investigated as it would reduce the processing steps and time considerably. Unfortunately this process cannot be used for the CGO supports as a conducting substrate is required.
The final finishing layer of fine CGO was deposited by screen-printing on both metal and CGO supports. An alternative finishing layer for the CGO supports was developed using the sol-gel technique and dip-coating. A good surface quality was achieved when using ideal substrates, but initial tests on the porous CGO supports indicated complete penetration of the sol into the support, as the pores are still too large for this process.
WP3: Nano-structured catalytic surface layers
The focus of WP3 laid on the development of high-quality catalysts as well as processing of surface layers by bulk and thin film membrane samples as the latter are expected to be rate limited by surface exchange. These new catalytic layers allowed increasing the oxygen exchange rate and consequently the overall oxygen-ion flux through the ultra-thin membrane, and/or improving the con-version and selectivity towards olefins and HCN. Moreover, the surface chemistry characterisation, catalytic and electrochemical testing of these new catalytic layers was carried out.
Main efforts were focused on the development of new catalytic layers in order to obtain higher oxygen permeation fluxes and stability, by means surface coatings based on BSCF and LSCF. In the case of HCN synthesis, studies were focused on platinum (Pt) based catalysts preparation and the obtained performance when deposited over different supports.
The presence of the catalytic layer on the permeate side of the membrane improves the oxygen permeation flux through the membrane in the low range temperature. The porosity of the catalytic layer is a key property for the activation of the OTMs. Improving the porosity and the composition of the catalytic layer, the oxygen permeation flux was 2.2 times increased at 650 ºC.
The first generation of thin-film BSCF membranes was tested. The use of a surface catalytic layer allowed a substantial increase in flux when the limitations due to fluid dynamics (gas polarisation) and bulk transport are minor. The second generation of thin-film BSCF membranes was also tested. Three different catalytic layers were selected and studied over asymmetric BSCF membranes. These catalytic layers consist in a 15 microns BSCF porous layer, adding silver (Ag) and palladium (Pd) to promote catalytic activity. The best obtained results correspond to BSCF M30 membrane with a deposition of porous BSCF layer with 5 % Pd.
Development of the first generation of surface catalytic layers for OCM involved two strategies, namely catalysts based on material thermochemically compatible with MIEC membrane reactors and catalysts based on material non-compatible with MIEC membrane reactors.
The most promising results are obtained with the catalyst based on materials non-compatible with MIEC membrane reactors. However, the results are strongly influenced by the experimental set-up employed. This fact could originate from the limitation in the oxygen permeation flux through the thick membrane or from experimental deviations in measurements.
Different catalyst formulations based on MIEC were considered by CSIC in the development of the second generation of the catalytic coating. The catalytic membranes synthesised were used in OCM to improve the production of C2-hydrocarbons. Thirty different catalysts have been tested in OCM reaction, the membranes were tested as a function of the operating temperature, methane per cent in feed and feed flow rate in the permeate side. When 20 % methane in AR is used as sweep gas, C2-hydrocarbons productivity of 2 ml/min cm2 was achieved at 1 000 ºC using the best catalyst tested.
High performance activation layers were fabricated by impregnation of (La0.6Sr0.4)0.99CoO3-d (LSC) into CGO backbones structures, mimicking the infiltration of porous sup-ports. An optimum firing temperature of the LSC infiltrate was found to be at 600 °C, when a balance between nano-scaled microstructure and catalyst formation was reached. Other processing conditions including CGO firing temperature and LSC loading were also found to be important to achieve a high performance activation layer. The results indicate the possibility of controlling separately the CGO backbone firing, LSC loading and most importantly the LSC firing to achieve exceptionally low polarisation resistances. Degradation studies indicate stability up to 450 hours suggesting reliability of the activation layer for use in OTMs.
First experiments on the AR reaction have been carried out using ceria-based asymmetric membranes with a Pt-Rh/LSCF catalytic layer. The formation of HCN was observed when feeding CH4+NH3+He to the permeate side. In all cases, the concentrations of the produced HCN are below 0.5 % at temperature of 900 to 1 000 ºC. A new catalyst was developed: supported Pt particles of around 100 to 300 nm were found to catalyse the HCN formation.
Thickness reduction of BSCF M30 (from 60 µm to 20 µm) allowed J(O2) improvement in the high temperature region, proving membrane thickness influence on permeation controlling processes at high temperature.
Asymmetric thin-film La0.58Sr0.4Co0.2Fe0.8O3-d membranes were tested. Catalytic activation was carried out following the same strategy than in the BSCF case: a LSCF porous layer with 5 % w/w was deposited on the permeate side. Obtained results show that LSCF membranes are very promising materials for oxygen separation, due to their relatively high J(O2) values and moderate stability. An oxygen flux of 11.87 ml/min-cm2 was reached at 1 000 ºC and pO2 = 1 atm in the feed.
Once proven BSCF membranes capability to reach oxygen fluxes above 10 ml/min-cm2, a long term test was performed, showing that after 1 000 hours membrane degradation was lower than 1 %, obtaining an oxygen purity higher than 99 %.
Dense and asymmetric BSCF membranes
In the development of the first generation of the catalyst layer, CSIC had different catalyst formulations based on MIEC (perovskites). The porosity, thickness and formulation of the catalyst surface layer have been studied. The presence of a catalytic layer on the permeate side of the membrane improves the JO2 through the membrane in the low temperature range. The best results were obtained with the most porous catalytic layer. By improving the porosity of the catalytic layer, JO2 was 2.2 times increased at 650 ºC.
Temperature dependence of the JO2 indicates Arrhenius behaviour and the apparent activation energies (Ea) for the oxygen transport process change as a function of the temperature. These results indicate that the rate limiting step varies according to the temperature range investigated; at higher temperatures, the bulk oxygen-ion diffusion is the limiting step for oxygen permeation across the membrane, while at lower temperatures, this process is limited by surface exchange reactions. Therefore, the study of the values of the Ea obtained can give indications about the limiting step in each case. Similar values of Ea for bulk BSCF bare membrane were previously reported. In the high temperature range, the presence of a catalytic layer hardly affects the performance of the membrane reactor in the oxygen permeation test.
Bulk BSCF membranes and first generation of thin-film BSCF membranes were tested. It can be ascertained that the increase of the sweep flow rate is very beneficial for the oxygen permeation, especially at the highest temperatures, when the highest oxygen fluxes are reached. The presence of the porous support has a negative effect on the oxygen flux that permeates through the membrane because it produces a concentration polarisation on the air side due to the hindered removal of nitrogen molecules. When polarisation effects on the oxygen feed are negligible due to use of pure oxygen feed and high space velocities, a substantial increase in the flux is observed. Under these conditions, surface exchange becomes one of the major rate limiting steps and the use of the activation layer has a more evident impact in the oxygen flux. The oxygen permeation flux of the coated BSCF thin film membrane disk reached 4.77 ml min-1cm-2 under and air/argon at 1 023 K and 22.46 ml min-1cm-2 under O2/argon at 1 023 K. In the latter case, the use of a surface catalytic layer allowed an important increase in flux.
The second generation of thin-film BSCF were manufactured varying dense membrane thickness and increasing substrate porosity up to 41 %. Thus, two membranes belonging to second generation BSCF M30 membranes were prepared, consisting of 60 µm and 20 µm thickness top gastight layer. From the permeation experiments it be inferred that thickness reduction leads to an improvement in J(O2), specially at high temperatures, this can be related with bulk diffusion processes controlling in this region. The highest oxygen fluxes were obtained for the 20 µm M30 membrane. Moreover, surface activation on permeate side was performed by means a deposition of a 15 µm porous BSCF layer, three different catalysts have been considered: BSCF, BSCF plus 5 % Pd and BSCF plus 5 % Ag. The use of Pd-activated membrane showed the best performance amongst all the considered options. This improvement in oxygen permeating flux is even higher at low temperatures, where the surface exchange reaction is believed to control the permeation. A flux of 97.63 ml/min-cm2 at 950 °C could be achieved.
Asymmetric LSCF membranes: To study oxygen permeation on asymmetric LSCF membranes, several disk-shaped membranes were manufactured, consisting of a 30 µm thick top gastight layer supported on a 650 µm thickness LSCF porous support. It is observed that oxygen fluxes increases with increasing pO2 in feed.
OCM catalysts: The research activities carried out entail the development and optimisation of OCM catalyst by applying the following strategies:
1. Catalysts based on materials thermochemically compatible with MIEC membranes. OCM catalysts compatible with OTM materials were used as catalytic layer in MIEC membrane reactors. The first generation of catalytic layers were based on MIEC perovskites. Different membrane reactors are compared in terms of yield to C2-hydrocarbons. Although the C2-yield obtained with LaCeSFC perovskite is better than that obtained with SmSFC, similar C2-productivity was obtained with these catalytic coatings. The best results (C2-yield and C2-productivity) were obtained when LaSr/CaO is used as catalytic coating, but still far from the target of the project.
2. Supported OCM catalysts incompatible with OTM materials. Perovskite hollow-fibre membranes of the composition BaCoxFeyZrzO3-d (BCFZ, x+y+z=1) in combination with an established 2 wt% Mn/5 wt% Na2WO4 on SiO2 catalyst as packed bed around the fibre were established for the oxidative coupling of methane. The fibres were produced in a spinning process and allow controlled oxygen feeding into the reactor over its axial length. Oxygen separation from air and C2 formation could be established at 800 °C with long-term stability. The highest C2 selectivity of approximately 75 % was observed at methane conversion of 6 % with a C2H4 to C2H6 ratio of 2:1. The highest C2H4 to C2H6 ratio of 4:1 and maximum C2 yield of 17 % was obtained at 50 % C2 selectivity. Further details can be found in Czuprat et al.
The thin film CGO supported membranes were tested for the oxidative coupling of methane using a corresponding catalyst which was the established 2 wt% Mn/5 wt% Na2WO4 on SiO2 catalyst as packed bed. For both membrane types, the C2 selectivities increase with an increasing methane concentration as expected whereas the COx selectivities decrease.
In contrast to the measurements already reported, the higher the methane concentration in the sweep gas, the higher the methane conversion. This could be explained by the fact that more methane consumes more oxygen on the membrane's surface, increasing the oxygen partial pressure gradient and therefore the flux of oxygen across the membrane. At 20 % methane conversion at C2 selectivity of about 92 % has been obtained. The porous CGO supported CGO membranes with spray-coated LSC/CGO layer on top failed after approximately 24 hours of OCM operation.
Nano-structured surface layers for HCN synthesis: A new catalyst for AR has been developed. Supported Pt particles of around 100 to 300 nm were found to catalyze the HCN formation. Different preparation routes has been developed, common for all is that a thermal conditioning step is necessary to crystallise the Pt particles at about 950 °C for 5h. It has been confirmed that Pt/Rh nanoparticles are the active phase in the AR if no metallic meshs/grids are used. The metallic Pt catalysts can be prepared on different supports (but catalytic tests showed that the supports must not contain Fe).
WP4: Modelling
Thermo mechanical behaviour
Knowledge of the mechanical properties is necessary for the reliability assessment of membrane materials and potential substrates. Important mechanical properties needed in analysis and simulations are elastic modulus and for brittle materials fracture strength, both can be determined using a bi-axial ring-on-ring bending tests. However, standard analytical relationships (ASTM C 1499) are limited to simple sample geometries requiring discrete ratios of thickness to loading/support ring radii. Finite element simulations have been carried out for thin, curved bi-layered materials to develop a testing procedure that permits the extension of standard analytical relationships to thin curved bi-layered specimens with the aim to determine the properties of the mechanically dominating support material.
Due to differences in thermal expansion, residual stresses can occur for bi-layered specimens with thin substrates which will be associated with bending. There are two basic arrangements to test the mechanical properties of such specimens in a bending test, we call them concave and convex shape, where the curvature is either increased or decreased during the test. The results of the finite element simulations not given in detail here imply that the convex shape is preferable. In the concave shape the undesirable non-linear deformation behaviour increases during the test.
The optimal CGO batch
In the experimental part various CGO batches were characterised with respect to elastic behaviour and fracture strength. It was verified that the properties depended on sintering temperature and porosity. In the current finite element analysis was used to find the CGO batch which provides optimal mechanical properties in the arrangement of symmetrical CGO on CGO membrane assembly. A chemical strain of 0.2 % was taken into consideration as an applied load for the substrate on the reduced side and also as a linear gradient across the membrane thickness diminishing at the interface.
Ratio of apparent stress and fracture strength as well as the ratio of the apparent strain and critical strain were taken as parameters to find the optimum arrangement.
Fracture energy criterion
One of key questions with respect to the membrane layer is if it can survive under these conditions since the stresses acting on the membrane are very high. However, there is a critical thickness below which the membrane will not fail since the elastic energy stored in the layer will not be high enough to initiate cracking.
The work was carried out for a CGO membrane on porous CGO and a dense BSCF membrane on porous BSCF, considering application relevant chemical strains. In case of CGO an application relevant chemical strain of 0.018 % was imposed. For simplification room temperature mechanical properties were used. The finite element analysis revealed that the membrane exhibits tensile stresses in this loading situation due to the expansion of the substrate. Hence, the membrane might fail by surface cracking. However, since the applied chemical strain induces only a rather low stress and also a low energy release rate for crack propagation, it can be expected that the CGO membrane will survive under these conditions.
Investigations were also carried out for a dense BSCF membrane layer on a porous BSCF substrate, based on application of relevant chemical strain value of 0.07 %. The energy release rate and the residual stress are below their critical values; hence both criteria imply that stable operation is possible from a mechanical point of view.
Modelling of surface and transport properties
The formation and migration of oxygen vacancies in the series of (Ba, La, Sr)(Co, Fe)O3 perovskites used in oxygen permeation membranes were studied in detail by means of first principles density functional calculations. Structure distortion, charge redistribution and transition state energies during oxygen ion migration are obtained and analyzed. Both the overall chemical composition and vacancy formation energy are found to have only small impact on the migration barrier; it is rather the local cation configuration which affects the barrier. The electron charge transfer from the migrating O ion towards the transition metal ion in the transition state is much smaller in (La,Sr)(Co,Fe)O3compared to (Ba,Sr)(Co,Fe)O3 perovskites where such a charge transfer makes a significant contribution to the low migration barriers observed (in particular for high Ba and Co content).
The oxygen migration in ABO3 perovskites occurs via the vacancy mechanism. The oxygen ion passage through the 'critical triangle' formed by one B site cation and two A site cations is the bottleneck; the trajectory is slightly curved away from the B site cation. In (La,Sr)(Mn,Fe,Co)O3 perovskites, the oxygen vacancy diffusion coefficients were experimentally found to be almost independent of the cation composition. In contrast, (Ba,Sr)(Co,Fe)O3 perovskites exhibit a higher diffusivity with migration barriers as low as 0.5 eV, in particular, for the samples with high Ba and Co content. We performed detailed comparative computer modelling of the oxygen transport in BSCF and LSCF.
Calculated vacancy formation energy varies nearly linearly as a function of the average oxidation number, whereas in Co-free materials, a step between the 4+/3+ and 3+/2+ regimes is visible. This is in a good agreement with the experimental observation of a plateau for the Fe3+ oxidation state while for Co the change from 4+ to 2+ occurs smoothly. The comparison of the results with the avail-able experimental data shows that the present calculations correctly reproduce the general trend of changing Ev with the average oxidation state and transition metal, although the quantitative magnitude is different.
The fundamental question arises, why (Ba,Sr)(Co,Fe)O3 and (La,Sr)(Co,Fe)O3 perovskites behave so differently - BSCF shows a strong dependence of the migration barriers on the vacancy formation energy leading to low barriers in particular for Ba- and Co-rich compositions, while LSCF hardly shows any dependence and comparably high barriers. Geometrical parameters such as the elongation of the A*A'* and B*B** distances in the transition state, representing the medium-range lattice relaxation and similar in both groups of perovskites. Hence, purely geometrical criteria such as the critical radius (rc) or the metal-oxygen distances in the transition state cannot explain the difference between (Ba,Sr)(Co,Fe)O3 and (La,Sr)(Co,Fe)O3 as the rc values are similar. The substitution of La or Sr by Ba ions expands the lattice constant, but also requires more space itself, thus the geometrical constraints in the transition state do not become more relaxed.
We also performed calculations on how Sr content in LSCF could affect the oxygen vacancy concentration. Oxygen vacancy formation in the Sr-poor perovskites involves a reduction of the transition metals below the 3+ oxidation state. The reduction in the 3+/2+ regime (Sr-poor sample) is energetically more costly than in the 4+/3+ regime. The difference in vacancy formation enthalpy is quite large (around 0.7 eV) and practically temperature independent. This is a direct result of the difference in the transi-tion metal oxidation states, in agreement with experiments. This demonstrates how carefully experimental conditions should be analysed when comparing to theory.
An increase of the temperature from 300 K to 1 200 K leads to a drop in the free enthalpy of vacancy formation. The calculated oxygen vacancy formation energy at zero Kelvin in La0.875Sr0.125Co0.25Fe0.75O3 is close to the experimental reaction enthalpy of 3.3 eV for La0.9Sr0.1FeO3- averaging over the values for 4+/3+ and 3+/2+ valence change. For La0.5Sr0.5Co0.25Fe0.75O3 the agreement is less good (the experimental enthalpies are only 1.0-1.1 eV for La0.5Sr0.5CoO3-d and La0.5Sr0.5FeO3); some improvement might be achieved by including phonons (entropic contributions in the solid).
Summing up, the electronic structure calculations were performed to explore the oxygen vacancy formation and diffusion processes in (La,Sr)(Co,Fe)O3 and (Ba,Sr)(Co,Fe)O3 perovskite solid solutions. The analysis revealed important differences of these two families of materials, the under-standing of which is imperative for applications of these materials in energy conversion devices, such as solid oxide fuel cells and oxygen permeation membranes.
WP5: Application-oriented evaluation
Oxyfuel power plants
This work topic focussed on conducting the necessary studies to examine whether the use of OTMs to separate oxygen from air is - compared with the available cryogenic air separation by distillation - a reliable and economic competitive technology. The studies involve an intensive review on oxyfuel technologies as well as OTM technology to define the operating conditions performances and likely process configurations such as the possibility of recycling the flue gases in order to decrease the oxyfuel combustion temperature or the mode of operation to provide the necessary driving force in the OTM process.
An industrial power plant of 500 MW was modelled. It includes the combustion process with flue gas pollutant purification as start-up baseline. After that simulations were run with an integrated ASU. On one hand, the OTM-membrane system was modelled considering a second generation membrane permeability equation and with parameters provided by experimental data from the project partners. On the other hand, cryogenic air distillation was simulated to evaluate its energy consumption taking a single column and a double column processes into account. The CO2 compression stage was evaluated too. The result of the simulations was a CO2-emissions-free power plant concept. Lastly, the steam cycle simulation was incorporated to the scheme for the purpose of reviewing the power plant energy production. The result was that the electrical efficiency of an oxyfuel power plant is about 3 % points higher with an integrated OTM-ASU than with a cryogenic ASU. Meaning that with high oxygen flux membranes the membrane alternative can show under certain circumstances an advantage over the conventional cryogenic air separation.
Economic evaluations on basis of the investment costs of the OTM and the cryogenic ASU using a factored-cost method support the previous studies.
An important item was the development of a conceptual membrane module design. Taking into account several considerations as well as mass and energy balances and the use of the second generation membrane permeability equation an optimal design was found.
OCM and HCM synthesis (AR)
Selected chemical applications of technical interest aiming at the direct oxidative conversion of methane to useful products were also in focus. In the case of the OCM Na2(Mn)WO4 catalysts supported on SiO2 were used for the fabrication of MCAs. In case of the AR new catalysts has to be developed resulting in Pt or Pt/Rh suspensions with particle diameters of circa 5 nm and powder-supported (powders = MgO, GCO, perowskite) catalysts with a Pt or Pt/Rh load up to 30 %.
The designed most favourable concepts for MCAs were the following:
1. The incorporation of unsupported catalysts in the porous substructure. For this concept a nano particulate suspension of the catalyst is needed to impregnate the porous substructure. The deposition of a sup-ported catalyst on top of porous substructure requires a catalyst on a support.
2. The coating of a powder supported catalysts on the gas-tight separation layer to get a porous catalyst layer. This concept would require catalysts supported on perovskite, CGO or inactive material powders as manganese oxide (MgO).
Four different reactors have been evaluated in OCM to C2. It seems that the disc membrane reactor is not suitable for high C2 yields. As one of the major problems it turned out that in case of the OCM the residence time of the reactants is too short and not defined in membrane reactors with planar disc-membrane geometry. Another drawback was the lack of a reaction model for the OCM which is based on molecular kinetics. Therefore, based on our present knowledge, the milestone MS8 could not be achieved in its full complexity for scientific reasons. With tubular or hollow fibre membrane reactors however, higher yields and selectivities could be obtained.
As general problem for the AR it was figured out during the first reaction tests, that ammonia decomposes at metallic surfaces at reaction temperature to hydrogen and nitrogen. It could be proven that quartz or dense alumina do not show this effect up to 950 °C. This finding led to the development of special adapted membrane reactors. Because membrane materials, the catalyst supports and the Pt-Rh catalyst particles themselves promote the ammonia (NH3) decomposition too, dif-ferent materials were tested in a quartz tube with a feed of 16.8 Nml/min of a He (90%) - NH3 (10%) mixture at different temperatures. The best catalyst in terms of the lowest decomposition is Pt/Rh supported by GCO (without sinter aid).
Different membrane reactor designs were used at LUH and BASF. The reactor developed at BASF was consequently built from quartz glass. Used were simple quartz glass ground-joints, where the sealing is done by gluing the membrane disc with gold paste to the lower part of the ground-joint. The air side was open to the air atmosphere of the furnace. At LUH beside a reactor for flat membrane disks a reactor for hollow fibre membranes with complete outside cold sealing were used.
For the fabrication of membrane catalyst assemblies the powder-supported catalysts were coated as a suspension with a paint-brush on the membrane. In case of the unsupported catalyst the porous side of the membrane was several times immersed with the developed Pt/Rh particle suspension.
All reaction tests at LUH and BASF with flat BSCF (Jülich) and CGO membranes (Risoe) with different catalysts either on the porous or on the gas tight separation layer do not lead to detectable HCN concentrations.
Only during the tests of a BCFZ hollow fibre (made by Fraunhofer IGB Stuttgart) with a porous cata-lyst layer of Pt/Rh on BCFZ powder (16.8 % metal content) on the outside of the fibre a HCN yield of 8.5 % at nearly 100% NH3 conversion could be achieved at 950 °C. We have to stress that this result is far away from the performance of the practiced conventional Andrussow synthesis.
The reason for this is too long residence times of feed and products in the lab-scale membrane reactors which results in a fast decomposition of NH3 before HCN can be formed and an additional decay or further oxidation of HCN. Enforced were these side reactions by the reactor geometry especially for flat membrane discs where an intermixing of educts, products and oxygen was occurring. A partly better flow regime could be achieved in the hollow fibre reactor. This results as shown in the formation of HCN in low concentrations.
In the LCA the project specific membrane information from other WPs were combined with additional life cycle data to describe the environmental effects of the entire process chain of electricity production by oxyfuel power plants. Considered life cycle stages beside the operation of the power plant are coal supply, setup of the power plant, ASU plant and membrane module, as well as CO2-compression, transport and storage.
The results were compared to those of an oxyfuel system using conventional cryogenic ASU and set into relation to a basic power plant without CCS. Five environmental effects were considered for comparison on the basis of 1 kWh of electricity produced, e.g. global warming potential (GWP), acidification potential (AP), eutrophication potential (EP), photochemical ozone creation potential (POCP), human toxicity potential (HTP).
For each impact the effect of a power plant without CCS is compared to those of oxyfuel plants using either cryogenic or membrane-based ASU systems. The GWP, AP and POCP decrease, because direct emissions at the power plant are avoided and rest emissions are discharged with the CO2 stream. At the same time, the effects due to a higher demand on coal and the additional CO2 transport and storage increase. However, the sum of all life cycle stages is smaller than for power plants without CCS. This cannot be found for EP and HTP. The impacts for EP increase by 7 % and 14 % for membrane-based and cryogenic ASU, respectively and for HTP even by 16 % and 25 %, mostly due to the higher coal demand. The high dependency of the coal supply results in an always better performance of the membrane-based ASU, as the efficiency drop is smaller than for cryogenic ASU.
Only the POCP is also sensible to the production of the membrane (emissions of burned binder during sintering). A reduction of the assumed membrane life time from 40 to 1 year and a 10 times smaller permeability would increase the POCP considerably and the total POCP would exceed that of the cryogenic system.
Among the resources necessary to produce BSCF powder, only Cobalt is classified among the moment as a critical material for the EU. However, an estimated capacity build-up of 1 GW would demand only 0.02 % of the yearly mining capacity.
Potential impact:
Membranes developed in this project provide potential for future commercial exploitation. Major attention should be drawn to two approaches, i.e. perovskitic thin film membranes (d = 10 to 50 µm) as well as metal supported ultra-thin film ceria membranes (d around 1 µm).
Thin film perovskite membranes needs to be scaled up and modules have to be designed considering the transport limitation in the support as discovered in this project. Scaling up of membrane components as well as the development of appropriate transport models is already ongoing in different groups. One major scientific issue is the sealing of the membrane components to the metallic parts of the modules, which is also addressed already in different projects. Demonstration of this concept can be expected as of 2014. The potential impact is expected to be high in specific applications.
CGO membranes showed high potential for membrane reactors, e.g. oxidative coupling of methane as investigated in this project using thin film ceria membranes. In this respect the metal supported ultra-thin film ceria membranes are very promising although experimental validation is still missing. Further research required comprises the demonstration of reliable and reproducible membranes and the investigation of the catalysis needed for the targeted applications. Especially the results of NASA-OTM for the oxidative coupling of methane to ethane and ethylene led to new insights for the design and the operation of an OTM reactor. The impact of such a development is quite high because there is a high need for the direct conversion of methane to useful products like HCN or hydrocarbons.
Project website: http://www.nasa-otm.eu/
Dr Wilhelm A. Meulenberg, Forschungzentrum Juelich GmbH, w.a.meulenberg@fz-juelich.de
Phone: +49-246-1616323
FAX: +49-246-1612455
Oxygen transport membranes (OTMs) are able to separate oxygen from air with high efficiency particularly if heat can be provided by the targeted application. Therefore, the main goal of this project is the development of gastight oxygen separation membranes for oxygen generation or the use in catalytic membrane reactors, which are characterized by:
1. high oxygen flux of 10 ml cm-2 min-1,
2. infinite oxygen selectivity and
3. high membrane stability.
In order to address all of these goals at the same time, the use of nano-structured surface activated ultra-thin OTMs is proposed. The development was done stepwise by developing firstly 'thin film membranes' (d = 20 to 50 µm) and subsequently the targeted 'ultra-thin film membranes' (d around 1 µm).
Three different applications are in focus:
1. oxyfuel carbon capture and storage (CCS) power plants requiring pure oxygen,
2. oxidative coupling of methane (OCM) as a known reaction using available catalysts and requiring high membrane stability in reducing atmospheres and
3. hydrogen cyanide (HCN) synthesis via the Andrussow reaction (AR), where the use of a membrane reactor is rarely investigated so far and thus novel catalysts are necessary while high membrane stability in reducing and acid atmospheres is required.
In order to meet the specific requirements of the targeted applications two membrane material classes are being investigated:
1. perovskites, which possess high permeability but limited stability for pure oxygen generation (oxyfuel) and
2. fluorites, which possess low permeability but high stability particularly in reducing atmospheres for the chemical applications (OCM and AR).
Perovskite membranes were developed based on Ba0.5Sr0.5Co0.8Fe0.2O3-d (BSCF) and La0.58Sr0.4Co0.2Fe0.8O3-d (LSCF). BSCF is the material with the highest permeability known so far whereas LSCF has lower permeability but higher stability. Surface activated tape cast thin film BSCF membranes revealed that the rate limiting step was no longer the bulk diffusion or the surface exchange but the concentration polarisation in the support pores. This effect occurs when using sweep gas to remove the oxygen from the membrane. An optimisation of the support porosity to 41% enhancing the support's binary gas diffusibility increased the permeation rate further, but did not prevent rate limitation by the support. Avoiding a sweep gas at the support side led to the highest permeation rates ever reported in literature and attracted much interest in the research community, accordingly. In LSCF thin film membranes this effect was much smaller due to the lower permeation rate, but the relative permeation enhancement compared to a 1 mm thick pellet is accordingly higher. Fluorite membranes were developed based on doped ceria. Tape cast thin film membranes showed very high yield in the OCM reaction. Although not all targets could be met at the same time, e.g. partly due to geometrical reasons using planar membranes, high potential of these membranes could be shown also in other catalytic processes. In AR, however, the experiments failed mainly because of undesired by-reactions possibly catalysed by the membrane material itself. More detailed studies on the catalysis of this reaction are required.
Both material classes were successfully applied as ultra-thin films using magnetron sputtering (MS). This required the development of appropriate substrates and interlayers providing a fine structured surface. LSCF (d = 0.7 µm) and cerium gadolinium oxide (CGO, d around 1 to 2 µm) were sputtered on a zirconia substrate with a thin ceria interlayer showing sufficient gas-tightness. However, permeation measurements revealed that the microstructure of the substrate is inappropriate due to the small pores of 80 to 100 nm across the whole support thickness of 2 to 2.5 µm. This leads to severe concentration polarisation as could be expected after the project results on thin film membranes. Finally, a metallic support with higher permeability and ceramic interlayers providing sufficient surface structure could be developed and gas-tight CGO layers applied. Detailed functional analysis of this very promising architecture has yet to be done.
Summarising, the project clearly shows the high potential of perovskitic thin film membranes for pure oxygen generation revealing that a further decrease in thickness to the µm range is not reasonable. In catalytic membrane reactors the metal supported ultra-thin film CGO membrane developed in this project is very promising, but still needs further investigation. In general, more attention has to be paid on the complex transport behaviour in such multi-layered membrane assemblies by experimental and modelling efforts.
Project context and objectives:
All scenarios about future global energy requirements anticipate an increasing demand for electricity, which in 2030 is predicted to be twice the current demand. These scenarios also indicate that this increase can only be met by considerable use of the fossil fuels coal, natural gas and oil. It becomes clear that fossil fuels will contribute to more than 60 % of future electricity generation. However, fossil fuel power plants are by far the biggest point sources of carbon dioxide (CO2) and contribute more than 40% of the worldwide anthropogenic CO2 emissions. Therefore, CCS is an important strategy to reduce CO2 emissions in order to counteract global warming and thus, with regard to Kyoto Protocol and the Energy Technology Strategic Plan of the European Union (EU), makes a key contribution to allow future electricity supply to be environmentally safe and sustainable. Efficient CO2 capture is possible using the oxyfuel technology. In an oxyfuel power plant, partly recirculated flue gas is enriched in oxygen and used to combust a carbon containing fuel forming primarily CO2 and water (H2O). This makes it much easier and cheaper to capture the CO2 than by using air for combustion. The required oxygen can be produced using ceramic OTMs, which are associated with significantly lower efficiency losses compared to conventional cryogenic separation technologies. OTMs consist of gastight mixed ionic electronic conductors (MIEC), which allow oxygen diffusion through vacancies in the crystal lattice and simultaneous transport of electrons in the opposite direction, thus obviating the need for an external electrical short circuit. However, using membranes in this synthesis has not been investigated so far.
In order to decrease the required membrane area a maximisation of the oxygen permeation rate is an evident requirement. This can be done by the choice of a material with high permeability. However, it is common knowledge that a trade-off between permeability and stability exist so that this approach has only limited practicability. Therefore, thin film membranes are developed in this project since the permeation rate is inversely proportional to the membrane thickness.
The project is organised in five scientific work packages (WPs). WPs one to three are related to the (active) OTM layer, the porous substrate and catalyst surface layers, respectively. The research is supported by modelling in WP4, i.e. thermo-mechanical and ab-initio transport modelling. Application-oriented evaluation (WP5) of the new membranes takes oxyfuel technology in fossil power plants as well as highly selective catalytic membrane applications into account. This choice of applications represents three different approaches, namely pure oxygen generation in 'clean' conditions, a relatively well known catalytic chemical reaction (OCM) in close contact to the membrane surface and a hardly investigated catalytic chemical reaction in close contact to the membrane surface.
At the start of the project, first generation materials for membranes (fluorite + perovskite), substrates (metallic + ceramic) and catalysts (for each application) had to be defined acting as reference materials. WP1 and WP2 work in close cooperation on the stepwise development of different types of planar lab scale membranes: bulk membranes (0.5 to 1 mm) were fabricated for characterisation of bulk properties and development of next generation materials. In parallel, thin film asymmetric membranes (d around 20 to 50 µm) using tape casting were manufactured. These membranes were used for e.g. catalysts testing (WP3) and first application-oriented tests (WP5). The development of the desired ultra-thin film membranes (i.e. d = 1µm or below) was gradually performed during the project with increasing effort and concentrating on most promising manufacturing technologies. Different substrate materials including ceramics, metals and cermets had to be investigated in order to obtain sufficient mechanical stability. Furthermore different deposition techniques with low deposition temperature are used in order to obtain compressive stresses in the membrane layer at application conditions. For both aims thermo-mechanical modelling supported the development. The supports needed to fulfil the following requirements:
1. high mechanical strength and chemical stability to support the active, ultra-thin membrane and to withstand application and testing conditions,
2. sufficient porosity to provide gas access to the membrane layer and to allow easy infiltration of active catalyst materials and
3. an adequate surface finish on one side of the supports to allow the application of ultra-thin membrane layers. This requires the development of graded interlayers.
WP3 is dedicated to the development of highly effective and stable catalysts in the targeted applications as well as processing of surface layers by bulk and thin film membrane samples. After testing the catalysts properties itself, activated membrane assemblies are delivered to WP5 in order to test the effectiveness of the developed catalysts. Catalytic coatings should allow increasing the oxygen permeance of thin-films at 750 ºC by at least a factor of four in comparison to not activated membranes (MS3) with an intermediate target of 5 ml cm-2 min-1 oxygen permeation rate (MS6). The final goal is the development of catalytical promotion of ultra-thin film membranes, which reach oxygen permeation values of 10 ml cm-2 min-1 with a degradation of 1 % per 1 000 hours and oxygen purity higher than 99 %.
WP4 starts modelling activities from the beginning with the knowledge of first generation materials. 'Thermo-mechanical behaviour of membranes' supports WP1 and WP2 concerning thermo-mechanical stability of ultra-thin film membranes. Finite element modelling of layered structures allows the derivation of an assessment procedure for the fracture stress of the supporting substrate material on the basis of experimental results as a measure of its robustness. Furthermore, a comparison of planar and tubular membrane set-ups shall permit to conclude which arrangement is favourable with regard to the respective stress distribution. The analysis utilises finite element simulations and is based on critical failure criteria and relevant mechanical data.
'Modelling of surfaces transport properties' is working on the improvement of first generation perovskite material by atomistic modelling of transport and surface exchange properties. The first principles large scale computer simulations will focus on the relation of chemical composition and functional properties. Thermodynamic stability of perovskites has to be analysed as a function of oxygen non-stoichiometry. Detailed quantum chemical calculations of the oxygen vacancy formation, and migration energies in a series of compositions, lead to the determination of major factors and compositions enabling fast oxygen transport. This shall lead to the prediction of a material composition increasing the oxygen flux through the membrane by a factor of two to three (MS7), which has to be confirmed by experiments.
WP5 evaluates the developed membranes regarding the targeted applications, i.e. oxyfuel, OCM and HCN synthesis. Testing of specimens as well as economical assessment and ecological evaluations is performed including life cycle assessment (LCA).
'Oxyfuel power plants' focuses on the system integration of an OTM membrane based air separation unit (OTM-ASU) into a CO2-emission-free fossil oxyfuel power plant. As bench mark for the assessment of the economic competitiveness of the OTM technology the conventional air separation by cryogenic distillation (cryogenic ASU) was chosen. By simulation of the OTM process the derivation of optimal operation conditions with respect to the membrane performance, a conceptual module design and the integration need are aimed in order to perform an economic assessment. An ecological analysis of a membrane-based oxyfuel power plant was performed LCA. The environmental impacts along production, use and disposal of membrane power plants were compared to oxyfuel processes using cryogenic air separation. LCA results are used to identify those processes which contribute most to environmental impacts. By comparing the membrane with conventional technology benchmarks for further membrane development were defined.
Selected chemical applications of technical interest were in focus of OCM and HCN synthesis. For both applications membrane catalyst assemblies (MCAs) following the most favourable concepts has to be fabricated with membranes provided by WP1 using adopted OCM or dedicated developed AR catalysts. Main task however is the installation of the MCAs in specially adopted membrane reactors, the implementation in suitable laboratory scale test facilities and the test execution of oxidative methane conversions in order to evaluate the technical potentials of the membrane reactor approach.
Project results:
WP1: Membrane materials, manufacture and characterisation
The main task of WP1 was to obtain gastight membrane layers on porous supports. Membrane materials were defined in accordance with all partners in WP5 and relevant materials properties were determined or collected from partners if accessible. Perovskites as well as doped ceria were considered. Moreover, mechanical characterisation was done with regard to materials' bulk properties as well as the properties of porous supports and asymmetric membrane assemblies. Membrane materials as well as promising layered membrane assemblies were delivered to application-oriented testing (WP5).
Membrane materials
At the kick-off meeting all partners agreed on the materials considered, i.e. LSCF and CGO. Later BSCF was additionally taken into account particularly for thin film asymmetric membranes. Since all these materials are well known by the partners involved, reliable data of the relevant materials properties are accessible.
The atomistic modelling of transport properties of membrane materials were also analysed. With regard to MS7 it was recommended to increase the strontium (Sr) content in LSCF. Respective experiments showed the perspective of that approach revealing that already an increase of the Sr content from 40 to 60 % A-site occupation is sufficient for this goal. Comparison of experiments and calculations is subject to a publication which is under submission. On the other hand other properties change simultaneously and have to be considered in future membrane development, e.g. thermal expansion coefficient.
Membrane layer processing
Thin-film asymmetric membranes were intended for the evaluation of catalytic layers (WP3) and as fall-back position in the risk management. Asymmetric membranes of a porous CGO support and a thin film membrane (thickness of 20 to 30 µm) were prepared by tape casting, lamination and co-sintering. These thin film membranes were used for first tests in the chemical applications (OCM and HCN). The first generation of asymmetric CGO membranes were mechanically weak and warped and were difficult to mount into testing rigs. Major advancements in processing of these membranes (strength and flatness) were achieved during the project due to CGO support optimisation and by using optical dilatometry, which allows the observation of shape evolution of the bilayers during high temperature sintering.
The sintering study for asymmetric membranes containing cobalt oxide in both CGO layers suggests an sintering cycle with a final temperature of about 1 050 °C for achieving flat, defect free bilayers with low failure rate whereas the study on the mechanical properties of the porous CGO supports shows that higher temperatures of about 1 150°C should be chosen to improve the mechanical properties of the support. As this contradiction could not be solved easily with the existing combination of layers an additional sintering aid (Fe2O3) was introduced into the thin membrane layer. The use of Fe2O3 allowed an increase in the final sintering temperature to 1 150 °C, reducing the warpage of the bilayer due to lower sintering activity of CGO membrane layer with Fe3O4 at this temperature. Such membranes (with Fe3O4 in the membrane layer and Co3O4 in the support layer) have a good combination of flatness and mechanical strength.
The optimisation of the sintering profile for CGO bilayers by using sintering aids allowed a significant reduction in sintering temperature, resulting in improved sample flatness, a reasonable residual porosity (around 35 %) and a reduced loss of samples during co-sintering.
first generation thin film asymmetric perovskite membranes were manufactured from BSCF. The highest oxygen permeation rates ever published could be obtained meeting the targets of MS3, which is described in WP3. The membranes were manufactured via sequential tape casting, in which the membrane layer is cast first. After drying, the support is cast, which consist of the same slurry but obtaining pore former (corn starch) leading to significant porosity in the support. Finally, a co-sintering step is carried out. The final (sintered) thicknesses of the membrane layer and the support are approximately 60 µm and 800 µm, respectively. It turned out that the permeation through the membrane layer itself is that high that molecular gas diffusion through the support's pores becomes rate limiting second generation thin film asymmetric membranes were developed in two steps. Firstly, a more porous BSCF support (41 % instead of 34 % porosity) was used. In addition, thinner BSCF membrane layers (approximately 20 µm instead of 60 µm thickness) were developed.
Ultra-thin film asymmetric membranes were manufactured via physical vapour deposition (PVD), i.e. MS. Sol-gel as well as chemical vapour deposition (CVD) was also evaluated but assessed as less promising in the given time frame.
CGO layers are coated via reactive MS. A metallic target of the respective alloy Ce0.8Gd0.2 is used. During sputtering a constant oxygen gas flow is fed to the high vacuum sputtering chamber. Therefore, the oxide CGO is formed on the substrate. The substrate can be heated up to 800 °C, which results into denser layers. Moreover, a bias power up to 400 W can be applied accelerating the Ar+-ions of the sputtering plasma onto the substrate. This ion bombardment leads to a further densification of the coating due to the kinetic energy of the Ar+-ions, however, resulting in residual compressive stresses in the layer, which can lead to delamination. These phenomena were investigated in detail and a four-zone model was derived. Using appropriate parameters a gas-tight film could be achieved, i.e. a third generation ultra-thin film membrane. A respective publication is in revision.
After this principal investigation of the coating technology and the parallel development of metallic substrates with appropriate ceramic interlayers in WP2 the latter were coated using reactive MS. The strength of the substrate and particularly the interlayer is given by manufacturing and cannot easily be improved by sintering. But, based on the experience gained in the project it was finally possible to develop a crack-free CGO top layer. This is the first time that a gas-tight ultra-thin film metal supported OTM was successfully produced. A respective publication is in preparation.
For the perovskite LSCF MS was performed using a ceramic target. Nice perovskite layers of approximately 700 nm thickness could be obtained on porous zirconia (8YSZ) substrates with a porous CGO interlayer. But, an accordant oxygen permeation measurement revealed that less oxygen permeated through the membrane assembly than through a dense LSCF disc. This is due to the fact that the porous support is relatively thick (2.5 mm) and consists of very tiny pores, i.e. 80 to 100 nm. This fine microstructure is on the one hand necessary for the application of such ultra-thin layers, but on the other hand evidently leads to severe concentration polarisation in the support, which was already observed in the thin film membranes using a thinner (0.9 mm) support with much larger pores (around 1 µm).
Mechanical evaluation of ultra-thin oxygen membrane structures
After analysing the basic mechanical properties of the membrane materials such as fracture stress, elastic modulus and hardness the emphasis was laid on the mechanically dominating porous substrate variants. Creep deformation and creep rupture was additionally analysed in order to determine component lifetime. Various methods such as creep rupture time, slow crack growth sensitivity of strength data and strength-probability-time (S-P-T) diagrams have been implemented. The experimental work covered the ceramic materials CGO, BSCF and LSCF as well as the alloy FeCrAlY.
Assuming linear elastic behaviour elastic modulus and fracture stress determines the fracture strain. Particularly fracture stress is comparable for the ceramics whereas the yield strength of the FeCrAlY is 'as expected' high at room temperature and strongly decreasing at elevated temperatures. But, at high temperatures a linear elastic consideration is not valid since creep phenomena might occur especially for the metal. However, such creep phenomena make a final validation complicated because they on the one hand are able to reduce residual or applied stresses, but on the other hand creep strain can lead to failure if the respective fracture stress is exceeded.
Dense LSCF can survive the highest stresses, whereas dense CGO seems to be least tolerant against such a load. Interestingly, the tolerable constant stress at room temperature is lower for dense CGO than for porous CGO. This effect is caused by an approximately two times higher sensitivity to sub-critical crack growth in combination with a very high scatter in fracture stress leading to low reliability.
WP2: Substrate materials, manufacture and characterisation
In WP2 two types of support structures for use in asymmetric membrane structures were developed, namely thin-film asymmetric membranes consisting of a porous support and a 20 to 30 µm thin-film, active membrane layer, both basically of the same material; and porous supports with graded interlayers for deposition of ultra-thin membranes with a thickness of 1 to 2 µm. The plain supports and asymmetric thin film membrane structures were prepared for initial catalysts testing (WP3) and application-oriented tests (WP5) of the thin film membranes. The predominant objective of WP2 was to design, prepare and characterise porous supports (both ceramic and metal) which were suitable for the application of ultra-thin membrane layers in WP1. During the project the processing of the supports was optimised in order to achieve suitable properties, such as the microstructure (porosity and pore-size), mechanical properties and lifetime. For ultra-thin film application of membrane layers the surface finish of the support structure played a key role and was refined by deposition of interlayers with graded porosity and reduced surface roughness. The asymmetric membranes were fabricated from CGO or BSCF, while for the porous supports CGO or FeCrAl were used.
Process optimisation for CGO porous supports
In WP2 porous CGO support layers with a thickness between 0.2 and 0.5 mm were made by tape casting, using graphite as pore former to introduce open porosity larger than 25 % after sintering. For preparation of asymmetric CGO structures (membrane bilayers consisting of 0.5 mm porous support and 30 µm active membrane layer), additionally thin CGO layers were tape casted. The tapes were then laminated together, debindered and co-sintered to form the CGO bilayers. The fabrication of high quality CGO support structures that are suitable for ultra-thin film (PVD) and thin film membrane application (co-sintering) required several optimisation steps. Typical defects included delamination of layers, warpage during sintering, surface scratches from the sintering plates and large irregular pores from the flaky graphite used as pore-former. All of these defects make the supports mechanically weak, difficult to handle and test.
Throughout the project continuous improvements in the quality of the CGO supports was achieved by detailed analyses of the ceramic processing, including optimisation of the slurry composition, the ceramic dispersion, the pore former size and shape to get more uniform pore size distribution. Sup-ports with improved strength and reduced warpage were produced by using a single casting process to make thicker films. In addition, the sintering process was investigated in more detail5 in order to optimise the microstructure of the supports with respect to the open porosity and mechanical properties. Here the pore-former (graphite) content was varied from 8 wt% to 11 wt% and the sintering temperature from 1 050 °C to 1 250 °C. The resulting microstructures can be seen in Figure 12 along with their open porosity values. The samples sintered at 1 050 °C had very high porosity and poor mechanical proper-ties, while those sintered at the high temperature of 1 250 °C had insufficient porosity. The conditions of 11 % (high) graphite content and a processing temperature of 1 150 °C produced a good open porosity of 45 % and adequate mechanical properties (strength = 46.7 ± 2 MPa, E-modulus = 45.6 ± 7.5 MPa). The Weibull modulus (a measure of the reliability) of the latest generation of samples increased to five as a result of better processing (less defects in the samples). This value is still low, even for ceramic materials, which is potentially problematic for applications involving the CGO membranes where there is a high risk of breakage during sealing, testing and changing gas atmospheres (redoxing).
Asymmetric thin-film CGO membranes
Asymmetric membranes of a porous CGO support and a thin film membrane (thickness of 20 to 30 µm) were prepared by tape casting, lamination and co-sintering. These thin film membranes were used for first tests in the chemical applications (OCM and HCN reactions, WP5). The first generation of asymmetric CGO membranes were mechanically weak and warped and were difficult to mount into testing rigs. Major advancements in processing of these membranes (strength and flatness) were achieved during the project due to CGO support optimisation and by using optical dilatometry.
Metallic porous supports An alumina-forming ferritic steel powder of the composition Fe21Cr7Al1Mo0.5Y was specifically designed for the NASA-OTM project and its properties were extensively characterised during the project. Metal supports potentially have advantages over ceramic supports with respect to cost, module integration and superior mechanical properties. However, the lifetime of porous metal supports could be limited by corrosion and propensity for creep, both effects are more severe with temperature and porosity. Corrosion data are presented in the following section with discussion of design requirements to ensure a long lifetime for the desired applications.
It is observed that the lower porosity samples have a lower weight gain, this is due to the smaller pores being blocked by the growth of the oxide scale. In order to maintain an open structure with good gas permeability over the life of the device, the following geometrical parameters are recommended, i.e. an initial porosity of around 40% with the smallest pores being much larger than the dimensions of the oxide scale. Oxidation of chromia-forming steel was also undertaken under the same conditions and it was confirmed that the alumina-forming steel had a lifetime around five times that of the FeCr. Modelling of the porous support lifetime as a function of the operating temperature, surface area and oxidation rate determined from experiments, showed that a lifetime of over 30 000 hours could be achieved operating at 750 °C. This recommended operating temperature is also preferable with respect to the mechanical properties, where the creep rate can be reduced to the target value of 1 % per year under these conditions (see WP1).
Metal support tapes with a thickness of 0.3 to 0.5 mm were fabricated from the FeCrAl powder by tape-casting and sintering. A water based tape casting process was developed, avoiding many environmental and health risks associated with organic solvents. This process was optimised to achieve tapes with good flatness and homogeneity and few defects. Sintering was performed at 1250 °C in a protective atmosphere to prevent excessive corrosion. Enough oxygen was present in sintering atmosphere to provide 'pre-oxidation' of the supports during sintering.
Substrates with graded interlayers for ultra-thin film asymmetric membranes
The as-prepared CGO and metal supports required some surface finish layers to reduce significantly the larger pore sizes and surface roughness to allow the deposition of defect-free dense ultra-thin membranes. The as-sintered CGO10 supports had a sub-micron surface roughness and maximum size of surface pores and defects around 2 µm. The metal supports had an average surface roughness of 3 to 4 µm with pore sizes over 10 µm due to the large particle size (10 to 38 µm) of the metal powder. Metal supports required two interlayers, i.e. a first coarser ceramic layer, then a second fine finishing layer. The CGO support required only the final CGO finishing layer.
Development of the first interlayer for the metal supports was very challenging and various interlayer materials and fabrication techniques, e.g. spray coating, screen-printing, dip coating and electrophoretic deposition (EPD) were investigated. It proved difficult to find ceramic materials that were stable and did not react with the metal support under the sintering conditions required for the metal support. Eventually, alumina was selected as the first interlayer material, as it is chemically stable under the sintering conditions, does not react with the metal support, adheres well and maintains a fine microstructure. The EPD technique is very suitable for this deposition as high solids-loaded films can be deposited and was up-scaled during the project to fabricate large batches of 25 cm2 samples. Deposition of a final alumina interlayer (also by EPD) is being investigated as it would reduce the processing steps and time considerably. Unfortunately this process cannot be used for the CGO supports as a conducting substrate is required.
The final finishing layer of fine CGO was deposited by screen-printing on both metal and CGO supports. An alternative finishing layer for the CGO supports was developed using the sol-gel technique and dip-coating. A good surface quality was achieved when using ideal substrates, but initial tests on the porous CGO supports indicated complete penetration of the sol into the support, as the pores are still too large for this process.
WP3: Nano-structured catalytic surface layers
The focus of WP3 laid on the development of high-quality catalysts as well as processing of surface layers by bulk and thin film membrane samples as the latter are expected to be rate limited by surface exchange. These new catalytic layers allowed increasing the oxygen exchange rate and consequently the overall oxygen-ion flux through the ultra-thin membrane, and/or improving the con-version and selectivity towards olefins and HCN. Moreover, the surface chemistry characterisation, catalytic and electrochemical testing of these new catalytic layers was carried out.
Main efforts were focused on the development of new catalytic layers in order to obtain higher oxygen permeation fluxes and stability, by means surface coatings based on BSCF and LSCF. In the case of HCN synthesis, studies were focused on platinum (Pt) based catalysts preparation and the obtained performance when deposited over different supports.
The presence of the catalytic layer on the permeate side of the membrane improves the oxygen permeation flux through the membrane in the low range temperature. The porosity of the catalytic layer is a key property for the activation of the OTMs. Improving the porosity and the composition of the catalytic layer, the oxygen permeation flux was 2.2 times increased at 650 ºC.
The first generation of thin-film BSCF membranes was tested. The use of a surface catalytic layer allowed a substantial increase in flux when the limitations due to fluid dynamics (gas polarisation) and bulk transport are minor. The second generation of thin-film BSCF membranes was also tested. Three different catalytic layers were selected and studied over asymmetric BSCF membranes. These catalytic layers consist in a 15 microns BSCF porous layer, adding silver (Ag) and palladium (Pd) to promote catalytic activity. The best obtained results correspond to BSCF M30 membrane with a deposition of porous BSCF layer with 5 % Pd.
Development of the first generation of surface catalytic layers for OCM involved two strategies, namely catalysts based on material thermochemically compatible with MIEC membrane reactors and catalysts based on material non-compatible with MIEC membrane reactors.
The most promising results are obtained with the catalyst based on materials non-compatible with MIEC membrane reactors. However, the results are strongly influenced by the experimental set-up employed. This fact could originate from the limitation in the oxygen permeation flux through the thick membrane or from experimental deviations in measurements.
Different catalyst formulations based on MIEC were considered by CSIC in the development of the second generation of the catalytic coating. The catalytic membranes synthesised were used in OCM to improve the production of C2-hydrocarbons. Thirty different catalysts have been tested in OCM reaction, the membranes were tested as a function of the operating temperature, methane per cent in feed and feed flow rate in the permeate side. When 20 % methane in AR is used as sweep gas, C2-hydrocarbons productivity of 2 ml/min cm2 was achieved at 1 000 ºC using the best catalyst tested.
High performance activation layers were fabricated by impregnation of (La0.6Sr0.4)0.99CoO3-d (LSC) into CGO backbones structures, mimicking the infiltration of porous sup-ports. An optimum firing temperature of the LSC infiltrate was found to be at 600 °C, when a balance between nano-scaled microstructure and catalyst formation was reached. Other processing conditions including CGO firing temperature and LSC loading were also found to be important to achieve a high performance activation layer. The results indicate the possibility of controlling separately the CGO backbone firing, LSC loading and most importantly the LSC firing to achieve exceptionally low polarisation resistances. Degradation studies indicate stability up to 450 hours suggesting reliability of the activation layer for use in OTMs.
First experiments on the AR reaction have been carried out using ceria-based asymmetric membranes with a Pt-Rh/LSCF catalytic layer. The formation of HCN was observed when feeding CH4+NH3+He to the permeate side. In all cases, the concentrations of the produced HCN are below 0.5 % at temperature of 900 to 1 000 ºC. A new catalyst was developed: supported Pt particles of around 100 to 300 nm were found to catalyse the HCN formation.
Thickness reduction of BSCF M30 (from 60 µm to 20 µm) allowed J(O2) improvement in the high temperature region, proving membrane thickness influence on permeation controlling processes at high temperature.
Asymmetric thin-film La0.58Sr0.4Co0.2Fe0.8O3-d membranes were tested. Catalytic activation was carried out following the same strategy than in the BSCF case: a LSCF porous layer with 5 % w/w was deposited on the permeate side. Obtained results show that LSCF membranes are very promising materials for oxygen separation, due to their relatively high J(O2) values and moderate stability. An oxygen flux of 11.87 ml/min-cm2 was reached at 1 000 ºC and pO2 = 1 atm in the feed.
Once proven BSCF membranes capability to reach oxygen fluxes above 10 ml/min-cm2, a long term test was performed, showing that after 1 000 hours membrane degradation was lower than 1 %, obtaining an oxygen purity higher than 99 %.
Dense and asymmetric BSCF membranes
In the development of the first generation of the catalyst layer, CSIC had different catalyst formulations based on MIEC (perovskites). The porosity, thickness and formulation of the catalyst surface layer have been studied. The presence of a catalytic layer on the permeate side of the membrane improves the JO2 through the membrane in the low temperature range. The best results were obtained with the most porous catalytic layer. By improving the porosity of the catalytic layer, JO2 was 2.2 times increased at 650 ºC.
Temperature dependence of the JO2 indicates Arrhenius behaviour and the apparent activation energies (Ea) for the oxygen transport process change as a function of the temperature. These results indicate that the rate limiting step varies according to the temperature range investigated; at higher temperatures, the bulk oxygen-ion diffusion is the limiting step for oxygen permeation across the membrane, while at lower temperatures, this process is limited by surface exchange reactions. Therefore, the study of the values of the Ea obtained can give indications about the limiting step in each case. Similar values of Ea for bulk BSCF bare membrane were previously reported. In the high temperature range, the presence of a catalytic layer hardly affects the performance of the membrane reactor in the oxygen permeation test.
Bulk BSCF membranes and first generation of thin-film BSCF membranes were tested. It can be ascertained that the increase of the sweep flow rate is very beneficial for the oxygen permeation, especially at the highest temperatures, when the highest oxygen fluxes are reached. The presence of the porous support has a negative effect on the oxygen flux that permeates through the membrane because it produces a concentration polarisation on the air side due to the hindered removal of nitrogen molecules. When polarisation effects on the oxygen feed are negligible due to use of pure oxygen feed and high space velocities, a substantial increase in the flux is observed. Under these conditions, surface exchange becomes one of the major rate limiting steps and the use of the activation layer has a more evident impact in the oxygen flux. The oxygen permeation flux of the coated BSCF thin film membrane disk reached 4.77 ml min-1cm-2 under and air/argon at 1 023 K and 22.46 ml min-1cm-2 under O2/argon at 1 023 K. In the latter case, the use of a surface catalytic layer allowed an important increase in flux.
The second generation of thin-film BSCF were manufactured varying dense membrane thickness and increasing substrate porosity up to 41 %. Thus, two membranes belonging to second generation BSCF M30 membranes were prepared, consisting of 60 µm and 20 µm thickness top gastight layer. From the permeation experiments it be inferred that thickness reduction leads to an improvement in J(O2), specially at high temperatures, this can be related with bulk diffusion processes controlling in this region. The highest oxygen fluxes were obtained for the 20 µm M30 membrane. Moreover, surface activation on permeate side was performed by means a deposition of a 15 µm porous BSCF layer, three different catalysts have been considered: BSCF, BSCF plus 5 % Pd and BSCF plus 5 % Ag. The use of Pd-activated membrane showed the best performance amongst all the considered options. This improvement in oxygen permeating flux is even higher at low temperatures, where the surface exchange reaction is believed to control the permeation. A flux of 97.63 ml/min-cm2 at 950 °C could be achieved.
Asymmetric LSCF membranes: To study oxygen permeation on asymmetric LSCF membranes, several disk-shaped membranes were manufactured, consisting of a 30 µm thick top gastight layer supported on a 650 µm thickness LSCF porous support. It is observed that oxygen fluxes increases with increasing pO2 in feed.
OCM catalysts: The research activities carried out entail the development and optimisation of OCM catalyst by applying the following strategies:
1. Catalysts based on materials thermochemically compatible with MIEC membranes. OCM catalysts compatible with OTM materials were used as catalytic layer in MIEC membrane reactors. The first generation of catalytic layers were based on MIEC perovskites. Different membrane reactors are compared in terms of yield to C2-hydrocarbons. Although the C2-yield obtained with LaCeSFC perovskite is better than that obtained with SmSFC, similar C2-productivity was obtained with these catalytic coatings. The best results (C2-yield and C2-productivity) were obtained when LaSr/CaO is used as catalytic coating, but still far from the target of the project.
2. Supported OCM catalysts incompatible with OTM materials. Perovskite hollow-fibre membranes of the composition BaCoxFeyZrzO3-d (BCFZ, x+y+z=1) in combination with an established 2 wt% Mn/5 wt% Na2WO4 on SiO2 catalyst as packed bed around the fibre were established for the oxidative coupling of methane. The fibres were produced in a spinning process and allow controlled oxygen feeding into the reactor over its axial length. Oxygen separation from air and C2 formation could be established at 800 °C with long-term stability. The highest C2 selectivity of approximately 75 % was observed at methane conversion of 6 % with a C2H4 to C2H6 ratio of 2:1. The highest C2H4 to C2H6 ratio of 4:1 and maximum C2 yield of 17 % was obtained at 50 % C2 selectivity. Further details can be found in Czuprat et al.
The thin film CGO supported membranes were tested for the oxidative coupling of methane using a corresponding catalyst which was the established 2 wt% Mn/5 wt% Na2WO4 on SiO2 catalyst as packed bed. For both membrane types, the C2 selectivities increase with an increasing methane concentration as expected whereas the COx selectivities decrease.
In contrast to the measurements already reported, the higher the methane concentration in the sweep gas, the higher the methane conversion. This could be explained by the fact that more methane consumes more oxygen on the membrane's surface, increasing the oxygen partial pressure gradient and therefore the flux of oxygen across the membrane. At 20 % methane conversion at C2 selectivity of about 92 % has been obtained. The porous CGO supported CGO membranes with spray-coated LSC/CGO layer on top failed after approximately 24 hours of OCM operation.
Nano-structured surface layers for HCN synthesis: A new catalyst for AR has been developed. Supported Pt particles of around 100 to 300 nm were found to catalyze the HCN formation. Different preparation routes has been developed, common for all is that a thermal conditioning step is necessary to crystallise the Pt particles at about 950 °C for 5h. It has been confirmed that Pt/Rh nanoparticles are the active phase in the AR if no metallic meshs/grids are used. The metallic Pt catalysts can be prepared on different supports (but catalytic tests showed that the supports must not contain Fe).
WP4: Modelling
Thermo mechanical behaviour
Knowledge of the mechanical properties is necessary for the reliability assessment of membrane materials and potential substrates. Important mechanical properties needed in analysis and simulations are elastic modulus and for brittle materials fracture strength, both can be determined using a bi-axial ring-on-ring bending tests. However, standard analytical relationships (ASTM C 1499) are limited to simple sample geometries requiring discrete ratios of thickness to loading/support ring radii. Finite element simulations have been carried out for thin, curved bi-layered materials to develop a testing procedure that permits the extension of standard analytical relationships to thin curved bi-layered specimens with the aim to determine the properties of the mechanically dominating support material.
Due to differences in thermal expansion, residual stresses can occur for bi-layered specimens with thin substrates which will be associated with bending. There are two basic arrangements to test the mechanical properties of such specimens in a bending test, we call them concave and convex shape, where the curvature is either increased or decreased during the test. The results of the finite element simulations not given in detail here imply that the convex shape is preferable. In the concave shape the undesirable non-linear deformation behaviour increases during the test.
The optimal CGO batch
In the experimental part various CGO batches were characterised with respect to elastic behaviour and fracture strength. It was verified that the properties depended on sintering temperature and porosity. In the current finite element analysis was used to find the CGO batch which provides optimal mechanical properties in the arrangement of symmetrical CGO on CGO membrane assembly. A chemical strain of 0.2 % was taken into consideration as an applied load for the substrate on the reduced side and also as a linear gradient across the membrane thickness diminishing at the interface.
Ratio of apparent stress and fracture strength as well as the ratio of the apparent strain and critical strain were taken as parameters to find the optimum arrangement.
Fracture energy criterion
One of key questions with respect to the membrane layer is if it can survive under these conditions since the stresses acting on the membrane are very high. However, there is a critical thickness below which the membrane will not fail since the elastic energy stored in the layer will not be high enough to initiate cracking.
The work was carried out for a CGO membrane on porous CGO and a dense BSCF membrane on porous BSCF, considering application relevant chemical strains. In case of CGO an application relevant chemical strain of 0.018 % was imposed. For simplification room temperature mechanical properties were used. The finite element analysis revealed that the membrane exhibits tensile stresses in this loading situation due to the expansion of the substrate. Hence, the membrane might fail by surface cracking. However, since the applied chemical strain induces only a rather low stress and also a low energy release rate for crack propagation, it can be expected that the CGO membrane will survive under these conditions.
Investigations were also carried out for a dense BSCF membrane layer on a porous BSCF substrate, based on application of relevant chemical strain value of 0.07 %. The energy release rate and the residual stress are below their critical values; hence both criteria imply that stable operation is possible from a mechanical point of view.
Modelling of surface and transport properties
The formation and migration of oxygen vacancies in the series of (Ba, La, Sr)(Co, Fe)O3 perovskites used in oxygen permeation membranes were studied in detail by means of first principles density functional calculations. Structure distortion, charge redistribution and transition state energies during oxygen ion migration are obtained and analyzed. Both the overall chemical composition and vacancy formation energy are found to have only small impact on the migration barrier; it is rather the local cation configuration which affects the barrier. The electron charge transfer from the migrating O ion towards the transition metal ion in the transition state is much smaller in (La,Sr)(Co,Fe)O3compared to (Ba,Sr)(Co,Fe)O3 perovskites where such a charge transfer makes a significant contribution to the low migration barriers observed (in particular for high Ba and Co content).
The oxygen migration in ABO3 perovskites occurs via the vacancy mechanism. The oxygen ion passage through the 'critical triangle' formed by one B site cation and two A site cations is the bottleneck; the trajectory is slightly curved away from the B site cation. In (La,Sr)(Mn,Fe,Co)O3 perovskites, the oxygen vacancy diffusion coefficients were experimentally found to be almost independent of the cation composition. In contrast, (Ba,Sr)(Co,Fe)O3 perovskites exhibit a higher diffusivity with migration barriers as low as 0.5 eV, in particular, for the samples with high Ba and Co content. We performed detailed comparative computer modelling of the oxygen transport in BSCF and LSCF.
Calculated vacancy formation energy varies nearly linearly as a function of the average oxidation number, whereas in Co-free materials, a step between the 4+/3+ and 3+/2+ regimes is visible. This is in a good agreement with the experimental observation of a plateau for the Fe3+ oxidation state while for Co the change from 4+ to 2+ occurs smoothly. The comparison of the results with the avail-able experimental data shows that the present calculations correctly reproduce the general trend of changing Ev with the average oxidation state and transition metal, although the quantitative magnitude is different.
The fundamental question arises, why (Ba,Sr)(Co,Fe)O3 and (La,Sr)(Co,Fe)O3 perovskites behave so differently - BSCF shows a strong dependence of the migration barriers on the vacancy formation energy leading to low barriers in particular for Ba- and Co-rich compositions, while LSCF hardly shows any dependence and comparably high barriers. Geometrical parameters such as the elongation of the A*A'* and B*B** distances in the transition state, representing the medium-range lattice relaxation and similar in both groups of perovskites. Hence, purely geometrical criteria such as the critical radius (rc) or the metal-oxygen distances in the transition state cannot explain the difference between (Ba,Sr)(Co,Fe)O3 and (La,Sr)(Co,Fe)O3 as the rc values are similar. The substitution of La or Sr by Ba ions expands the lattice constant, but also requires more space itself, thus the geometrical constraints in the transition state do not become more relaxed.
We also performed calculations on how Sr content in LSCF could affect the oxygen vacancy concentration. Oxygen vacancy formation in the Sr-poor perovskites involves a reduction of the transition metals below the 3+ oxidation state. The reduction in the 3+/2+ regime (Sr-poor sample) is energetically more costly than in the 4+/3+ regime. The difference in vacancy formation enthalpy is quite large (around 0.7 eV) and practically temperature independent. This is a direct result of the difference in the transi-tion metal oxidation states, in agreement with experiments. This demonstrates how carefully experimental conditions should be analysed when comparing to theory.
An increase of the temperature from 300 K to 1 200 K leads to a drop in the free enthalpy of vacancy formation. The calculated oxygen vacancy formation energy at zero Kelvin in La0.875Sr0.125Co0.25Fe0.75O3 is close to the experimental reaction enthalpy of 3.3 eV for La0.9Sr0.1FeO3- averaging over the values for 4+/3+ and 3+/2+ valence change. For La0.5Sr0.5Co0.25Fe0.75O3 the agreement is less good (the experimental enthalpies are only 1.0-1.1 eV for La0.5Sr0.5CoO3-d and La0.5Sr0.5FeO3); some improvement might be achieved by including phonons (entropic contributions in the solid).
Summing up, the electronic structure calculations were performed to explore the oxygen vacancy formation and diffusion processes in (La,Sr)(Co,Fe)O3 and (Ba,Sr)(Co,Fe)O3 perovskite solid solutions. The analysis revealed important differences of these two families of materials, the under-standing of which is imperative for applications of these materials in energy conversion devices, such as solid oxide fuel cells and oxygen permeation membranes.
WP5: Application-oriented evaluation
Oxyfuel power plants
This work topic focussed on conducting the necessary studies to examine whether the use of OTMs to separate oxygen from air is - compared with the available cryogenic air separation by distillation - a reliable and economic competitive technology. The studies involve an intensive review on oxyfuel technologies as well as OTM technology to define the operating conditions performances and likely process configurations such as the possibility of recycling the flue gases in order to decrease the oxyfuel combustion temperature or the mode of operation to provide the necessary driving force in the OTM process.
An industrial power plant of 500 MW was modelled. It includes the combustion process with flue gas pollutant purification as start-up baseline. After that simulations were run with an integrated ASU. On one hand, the OTM-membrane system was modelled considering a second generation membrane permeability equation and with parameters provided by experimental data from the project partners. On the other hand, cryogenic air distillation was simulated to evaluate its energy consumption taking a single column and a double column processes into account. The CO2 compression stage was evaluated too. The result of the simulations was a CO2-emissions-free power plant concept. Lastly, the steam cycle simulation was incorporated to the scheme for the purpose of reviewing the power plant energy production. The result was that the electrical efficiency of an oxyfuel power plant is about 3 % points higher with an integrated OTM-ASU than with a cryogenic ASU. Meaning that with high oxygen flux membranes the membrane alternative can show under certain circumstances an advantage over the conventional cryogenic air separation.
Economic evaluations on basis of the investment costs of the OTM and the cryogenic ASU using a factored-cost method support the previous studies.
An important item was the development of a conceptual membrane module design. Taking into account several considerations as well as mass and energy balances and the use of the second generation membrane permeability equation an optimal design was found.
OCM and HCM synthesis (AR)
Selected chemical applications of technical interest aiming at the direct oxidative conversion of methane to useful products were also in focus. In the case of the OCM Na2(Mn)WO4 catalysts supported on SiO2 were used for the fabrication of MCAs. In case of the AR new catalysts has to be developed resulting in Pt or Pt/Rh suspensions with particle diameters of circa 5 nm and powder-supported (powders = MgO, GCO, perowskite) catalysts with a Pt or Pt/Rh load up to 30 %.
The designed most favourable concepts for MCAs were the following:
1. The incorporation of unsupported catalysts in the porous substructure. For this concept a nano particulate suspension of the catalyst is needed to impregnate the porous substructure. The deposition of a sup-ported catalyst on top of porous substructure requires a catalyst on a support.
2. The coating of a powder supported catalysts on the gas-tight separation layer to get a porous catalyst layer. This concept would require catalysts supported on perovskite, CGO or inactive material powders as manganese oxide (MgO).
Four different reactors have been evaluated in OCM to C2. It seems that the disc membrane reactor is not suitable for high C2 yields. As one of the major problems it turned out that in case of the OCM the residence time of the reactants is too short and not defined in membrane reactors with planar disc-membrane geometry. Another drawback was the lack of a reaction model for the OCM which is based on molecular kinetics. Therefore, based on our present knowledge, the milestone MS8 could not be achieved in its full complexity for scientific reasons. With tubular or hollow fibre membrane reactors however, higher yields and selectivities could be obtained.
As general problem for the AR it was figured out during the first reaction tests, that ammonia decomposes at metallic surfaces at reaction temperature to hydrogen and nitrogen. It could be proven that quartz or dense alumina do not show this effect up to 950 °C. This finding led to the development of special adapted membrane reactors. Because membrane materials, the catalyst supports and the Pt-Rh catalyst particles themselves promote the ammonia (NH3) decomposition too, dif-ferent materials were tested in a quartz tube with a feed of 16.8 Nml/min of a He (90%) - NH3 (10%) mixture at different temperatures. The best catalyst in terms of the lowest decomposition is Pt/Rh supported by GCO (without sinter aid).
Different membrane reactor designs were used at LUH and BASF. The reactor developed at BASF was consequently built from quartz glass. Used were simple quartz glass ground-joints, where the sealing is done by gluing the membrane disc with gold paste to the lower part of the ground-joint. The air side was open to the air atmosphere of the furnace. At LUH beside a reactor for flat membrane disks a reactor for hollow fibre membranes with complete outside cold sealing were used.
For the fabrication of membrane catalyst assemblies the powder-supported catalysts were coated as a suspension with a paint-brush on the membrane. In case of the unsupported catalyst the porous side of the membrane was several times immersed with the developed Pt/Rh particle suspension.
All reaction tests at LUH and BASF with flat BSCF (Jülich) and CGO membranes (Risoe) with different catalysts either on the porous or on the gas tight separation layer do not lead to detectable HCN concentrations.
Only during the tests of a BCFZ hollow fibre (made by Fraunhofer IGB Stuttgart) with a porous cata-lyst layer of Pt/Rh on BCFZ powder (16.8 % metal content) on the outside of the fibre a HCN yield of 8.5 % at nearly 100% NH3 conversion could be achieved at 950 °C. We have to stress that this result is far away from the performance of the practiced conventional Andrussow synthesis.
The reason for this is too long residence times of feed and products in the lab-scale membrane reactors which results in a fast decomposition of NH3 before HCN can be formed and an additional decay or further oxidation of HCN. Enforced were these side reactions by the reactor geometry especially for flat membrane discs where an intermixing of educts, products and oxygen was occurring. A partly better flow regime could be achieved in the hollow fibre reactor. This results as shown in the formation of HCN in low concentrations.
In the LCA the project specific membrane information from other WPs were combined with additional life cycle data to describe the environmental effects of the entire process chain of electricity production by oxyfuel power plants. Considered life cycle stages beside the operation of the power plant are coal supply, setup of the power plant, ASU plant and membrane module, as well as CO2-compression, transport and storage.
The results were compared to those of an oxyfuel system using conventional cryogenic ASU and set into relation to a basic power plant without CCS. Five environmental effects were considered for comparison on the basis of 1 kWh of electricity produced, e.g. global warming potential (GWP), acidification potential (AP), eutrophication potential (EP), photochemical ozone creation potential (POCP), human toxicity potential (HTP).
For each impact the effect of a power plant without CCS is compared to those of oxyfuel plants using either cryogenic or membrane-based ASU systems. The GWP, AP and POCP decrease, because direct emissions at the power plant are avoided and rest emissions are discharged with the CO2 stream. At the same time, the effects due to a higher demand on coal and the additional CO2 transport and storage increase. However, the sum of all life cycle stages is smaller than for power plants without CCS. This cannot be found for EP and HTP. The impacts for EP increase by 7 % and 14 % for membrane-based and cryogenic ASU, respectively and for HTP even by 16 % and 25 %, mostly due to the higher coal demand. The high dependency of the coal supply results in an always better performance of the membrane-based ASU, as the efficiency drop is smaller than for cryogenic ASU.
Only the POCP is also sensible to the production of the membrane (emissions of burned binder during sintering). A reduction of the assumed membrane life time from 40 to 1 year and a 10 times smaller permeability would increase the POCP considerably and the total POCP would exceed that of the cryogenic system.
Among the resources necessary to produce BSCF powder, only Cobalt is classified among the moment as a critical material for the EU. However, an estimated capacity build-up of 1 GW would demand only 0.02 % of the yearly mining capacity.
Potential impact:
Membranes developed in this project provide potential for future commercial exploitation. Major attention should be drawn to two approaches, i.e. perovskitic thin film membranes (d = 10 to 50 µm) as well as metal supported ultra-thin film ceria membranes (d around 1 µm).
Thin film perovskite membranes needs to be scaled up and modules have to be designed considering the transport limitation in the support as discovered in this project. Scaling up of membrane components as well as the development of appropriate transport models is already ongoing in different groups. One major scientific issue is the sealing of the membrane components to the metallic parts of the modules, which is also addressed already in different projects. Demonstration of this concept can be expected as of 2014. The potential impact is expected to be high in specific applications.
CGO membranes showed high potential for membrane reactors, e.g. oxidative coupling of methane as investigated in this project using thin film ceria membranes. In this respect the metal supported ultra-thin film ceria membranes are very promising although experimental validation is still missing. Further research required comprises the demonstration of reliable and reproducible membranes and the investigation of the catalysis needed for the targeted applications. Especially the results of NASA-OTM for the oxidative coupling of methane to ethane and ethylene led to new insights for the design and the operation of an OTM reactor. The impact of such a development is quite high because there is a high need for the direct conversion of methane to useful products like HCN or hydrocarbons.
Project website: http://www.nasa-otm.eu/
Dr Wilhelm A. Meulenberg, Forschungzentrum Juelich GmbH, w.a.meulenberg@fz-juelich.de
Phone: +49-246-1616323
FAX: +49-246-1612455