A Thermocatalytic Ceramic Membrane by Perovskite Incorporation in the Alumina Framework

Access to clean water is limited by the increasing amount of persistent organic pollutants (POPs), since current methods fail to remove POPs completely. Therefore, new treatment technologies of surface water and wastewater are needed. In this study, two treatment methods are combined in one step, i.e., membrane filtration and thermocatalytic chemical oxidation of POPs. A perovskite‐type catalyst with formula Sr0.85Ce0.15FeO3‐δ (SCF) is incorporated into an alumina membrane using a simple two‐step heat treatment to minimize any chemical reaction of the catalytic active perovskite with alumina. First, a sintering process under inert atmosphere, then, a heat‐treatment under oxidative conditions to oxidize the iron species in the perovskite structure. The well‐known thermocatalytic properties of SCF make the membrane thermocatalytic and thus able to degrade pollutants under dark conditions without addition of oxidants. The SCF content in the membrane is varied between 0 and 15 wt% to explore the change in membrane properties. Results demonstrate that the thermocatalytic membranes have great potential for continuous membrane filtration and simultaneous degradation of POPs. When considering methyl orange, up to 100% removal is obtained at room temperature, whereas up to 93% of bisphenol A is removed at temperatures approaching 60 °C.


Introduction
The industrialization of the modern society has favored an increased amount of persistent organic pollutants (POPs) in DOI: 10.1002/admi.202300435wastewater, originating from pharmaceuticals or industries such as chemicals, textiles, and food. [1,2]POPs are a major concern if not properly treated at wastewater treatment plants, as they will possibly affect aquatic life and human health.Moreover, they usually exist in very small concentrations making them difficult to analyze and remove. [3]reatment processes include multiple steps with different techniques such as membrane filtration, [4] biological degradation, [5] or chemical degradation using ozone [6,7] or catalysts. [8]embrane filtration is an emerging technology suitable in various applications such as water purification and wastewater treatment.However, for some small and non-charged compounds, the rejection is limited, resulting in insufficient removal of POPs, while producing a volume of waste concentrate (retentate). [9,10]Therefore, membrane filtration is often combined with biological or chemical processes to remove and degrade POPs. [11]The membrane material is usually divided into three groups, namely organic, inorganic, and organic-inorganic hybrid. [12]Inorganic (or ceramic) membranes are usually fabricated from alumina, [13,14] silicon carbide, [15,16] silica, [17] or TiO 2 , [18,19] thus having some great benefits, including long lifetime, chemical resistance, and high mechanical strength.In general, -alumina is a commonly used membrane support material, produced in wet [20] or dry conditions. [21,22]The properties of the membrane obtained using dry conditions, i.e.: the sintering of green bodies that are pressed into pellets with or without additives or organic binders, can be greatly altered depending on the heating program. [23,24]Inorganic membranes can also act as support layer for an active catalytic phase, such as TiO 2 on an alumina support. [25]This is interesting as it combines membrane filtration with catalysis in a one-step treatment.Research in combining advanced oxidation processes and membrane filtrations is on-going. [26]atalytic membranes can be prepared by several routes depending on membrane material.The methods include blending, surface coating, and bottom-up synthesis. [27]Catalytic membranes are often prepared by surface coating of the active catalyst onto the membrane supports, [28] allowing minimization of the membrane pore size, [29] as seen for perovskite coatings. [30,31]onsidering catalytic membranes, Li et al. [32] produced a porous membrane electrode based on RuO 2 /TiO 2 nanorods on carbon nanofiber membranes that showed degradation efficiency of >98% of both bisphenol A and sulfadiazine.Lofti et al. [33] prepared a photocatalytic TiO 2 -coated polyethersulfone (PES) membrane for removal of steroids.They immobilized TiO 2 nanoparticles both on the surface and in the pores, which ensured a removal of >80% of non-radiolabeled hormones E1 and E2.Similarly, Wang et al. [34] fabricated a catalytic polymeric ultrafiltration membrane by mixing the spinel-structured CoFe 2 O 4 with PES for activation of peroxymonosulfate (PMS) in the degradation of naproxen.They obtained removal of naproxen up to 87.7% by prolonging the residence time and thus reported for an ultrafiltration membrane a similar removal rate as for a nanofiltration membrane.Concerning ceramic catalytic membranes, Wu et al. [35] prepared a MnO 2 -integrated alumina membranes by powder mixing, sieving, and pressing prior to heattreatment.The membranes could activate peroxymonosulfate (PMS) in radical formation to degrade 4-hydroxylbenzoic acid up to 98.9% using only 1.67% MnO 2 .Liang et al. [29] also fabricated catalytic perovskite membranes for degradation of organic pollutants by activation by PMS.This showed to efficiently prevent fouling formation, by long-term flux stability.Ceramic membranes have also been functionalized by incorporation of maghemite for catalytic ozonation to degrade organic pollutants in water. [36]Choi et al. [37] sol-gel coated an alumina membrane with anatase-phased TiO 2 to produce a photocatalytic membrane that showed removal of both methylene blue (full decolorization in 5 h) and creatinine (≈60% after 14 h).Recently, a thermocatalytic membrane based on a Sr 0.85 Ce 0.15 FeO 3- perovskite and graphene oxide coating on a PES membrane support showed abatement of bisphenol A during filtration and no effect on toxicity. [38]n this study, we incorporate a thermocatalyst into the framework of a ceramic alumina membrane using a simple procedure of mixing alumina with a thermocatalytic perovskite (Sr 0.85 Ce 0.15 FeO 3- , SCF) to form a homogeneous membrane with the perovskite incorporated in the alumina framework.Hence, the thermocatalyst is distributed within the ceramic membrane instead of surface coating, which will increase the contact time between catalyst and pollutant and ensure efficient pollutant degradation.SCF has a stabilized cubic structure due to the Ce-doping of the SrFeO 3 perovskite and a high degree of Fe 4+ at B-site. [39]Recently, it has shown ability to degrade various organic compounds in literature including oil, acid orange 7 and 8, rhodamine B, and bisphenol A [40][41][42] without the need of adding oxidants such as PMS.The activation by heat is advantageous compared to light, as there is no requirement for an optical window, and membranes can be activated during filtration of turbid suspensions or in dark modules, e.g., tubular membrane modules.The SCF-alumina mixture is simply uniaxially compressed and then sintered through a two-step thermal treatment, which limits the reactions between perovskite and alumina.The crystalline phases produced by these reactions are investigated at different temperatures in both air and inert atmosphere.Finally, a perovskite-alumina membrane is compared against a pure alumina membrane regarding morphology, porosity, permeability (N 2 gas and water), and ability to remove organic pollutants.

Powder Characterization
X-ray diffraction (XRD) pattern and Rietveld refinement of SCF (Figure S2a, Supporting Information) proves the presence of Sr 0.86 Ce 0.14 FeO 3- and a minor percentage of segregated CeO 2 due to the solubility limit ≈14 mol.% of Ce in SrFeO 3 , as described elsewhere. [39]Phase composition and structural information are given in Table S1 (Supporting Information).Likewise, the XRD patterns and refinement data of the as-received commercial alumina powder (AKP30) and of the SCF-alumina powder mixture with 10 wt% SCF before thermal treatment are found in Figure S2b,c (Supporting Information), respectively, while composition and structural information are given in Table S1 (Supporting Information).

Thermal Stability of the SCF-Alumina Powder Mixtures
The diffraction patterns of SCF-alumina powder mixtures with 15 wt% SCF after thermal treatment at different temperatures and/or atmospheres are shown in Figures 1a-c.Rietveld refinements of all diffractograms are found in Figures S4 and S5 (Supporting Information) and their phase composition is found in Figure 1d−f as well as in Tables S2 and S3 (Supporting Information).An increasing temperature resulted in a loss of SCF due to a reaction between SCF and alumina.The main reaction taking place was the diffusion of strontium and iron from SCF into the alumina, forming strontium iron aluminum oxides identified as SrFe 12-x Al x O 19 or strontium aluminum oxides identified as SrAl 2 O 4 (PDF n°00-034-0379).Strontium monoaluminate and strontium hexaaluminate structures have been previously observed in the literature after the heat treatment of SrO and -Al 2 O 3 mixtures at temperatures >900 °C in air. [43]The changes in phase composition with increasing heat treatment temperature in air are shown in Figure 1a,d, whereas the effect of prolonged heat treatment under argon flow is shown in Figure 1b,e.Initially, strontium monoaluminate (SrAl 2 O 4 ) was formed, but as the temperature increases to 1220 °C, the strontium monoaluminate combined with the iron forming strontium hexaaluminate.As CeO 2 has a limited solubility in SrFeO 3 (maximum 14 mol.%Ce), the diffusion of strontium and iron from SCF resulted in the segregation of a larger CeO 2 amount in the membranes heattreated in argon atmosphere.
To avoid the unwanted SCF-alumina reaction, samples were sintered at 870 °C under argon atmosphere and different durations were explored (Figure 1b).This temperature was high enough to favor the sintering of the powder, although low enough to minimize the Sr-Al reaction (see Rietveld refinements in Figure S5, Supporting Information).In fact, when a mixture of SCF and alumina was sintered at 1070 °C for 1 h under argon atmosphere, the perovskite phase totally disappeared (Figure S5c, Supporting Information).The SCF phase was present up to 5 h of argon sintering at 870 °C, although it disappeared after 10 h of argon sintering at the same temperature (Figure S5d, Supporting Information).In parallel, the strontium aluminate phase percentage increased (up to 13.5 wt%) with the duration of the heat treatment under argon atmosphere (Figure 1e; Table S3, Supporting Information).However, the total percentage of Sr-Al phases formed under argon treatment was much smaller than under air treatment at the same temperature and time (Tables S2 and S3, Supporting Information).The lack of oxygen during argon sintering might be responsible for the reduced formation of Sr-Al compounds, as a significant amount of oxygen is necessary to form some of the strontium aluminum oxides.CeO 2 , present in very low percentage already in the starting SCF powder (Figure S2a, Supporting Information), reacts with alumina under argon, forming trace amounts of CeAlO 3 after prolonged heat treatment (10 h at 870 °C) or higher temperatures (1 h at 1070 °C), as shown in Figure S5c,d and Table S3 (Supporting Information).Formation of Fe 3 O 4 was also observed along with CeAlO 3 .The formation of CeAlO 3 has been already reported in the literature after reducing treatments at high temperatures. [44]t is worth to notice that the cubic perovskite structure present in SCF expanded after high-temperature sintering under inert atmosphere (Figure S6, Supporting Information), due to the reduction of the smaller Fe 4+ to the larger Fe 3+ , as previously observed in the literature. [39]The expansion of the cubic phase is better evidenced in the inset of Figure S6 (Supporting Information), showing a decrease in the detector angle (2), corresponding to a larger d value (d = distance between the crystal planes), according to the Braggs Law.Unfortunately, the reaction kinetic constant of the membrane with the expanded perovskite decreased from 0.0409 to 0.0067 min −1 (see Figure S7, Supporting Information and Section 2.4), thus making a re-oxidation of the perovskite structure very important to enhance its catalytic activity.Therefore, an additional heat treatment in air at 600 °C was performed, which resulted in a shift of the perovskite diffraction peak to higher 2, due to the re-oxidation of the larger Fe 3+ to the smaller Fe 4+ , bringing the cell volume back to the initial value (inset of Figure S6, Supporting Information).The re-oxidation resulted in a significant improvement of the reaction kinetic constant, from 0.0067 to 0.0211 min −1 (see Figure S7, Supporting Information).
The cubic perovskite was present in the re-oxidized membranes at any SCF/alumina ratio, as observed in the XRD diffractograms in Figure 1c (compare also Rietveld refinements in Figure S8, Supporting Information and phase composition and other structural parameters in Table S4, Supporting Information), although the percentage of unwanted crystalline phases increased as well at increasing SCF content in the membranes.In fact, a larger amount of strontium available for diffusion into alumina and reaction caused formation of SrAl 2 O 4 (from 1.1 to 4.0 wt%) and SrFe 12-x Al x O 19 (from 1.3 to 3.0 wt%) as the SCF content increases from 5 to 15 wt%.Moreover, CeO 2 was also present, since it segregated from SCF (from 0.3 to 0.8 wt%) to compensate the destabilization of the perovskite structure caused by the Sr-Al reaction.

Cut-Off Pore Size
The maximum membrane pore size (i.e.: cut-off pore size) increased with the SCF content (Figure 2) and, based on image analyses, the cut-off pore size was 280, 320, 340, and 400 nm for membranes with SCF content of 0, 5, 10, and 15 wt%, respectively.A comparison between SEM images and image analysis using ImageJ is found in Supporting Information for the pure alumina membrane (Figure S9a,b, Supporting Information).The pore size distributions are based on equivalent pore diameter (i.e., the diameter determined from the area of the pore assuming the pore to be a perfect circle) and these show a majority of small pores <200 nm (Figure S9c, Supporting Information).The pore size did not show a clear relation to perovskite content (Table 1), but porosity increased with SCF content (inset Figure 3a).The morphology of the membranes observed from the SEM images (Figure 2) indicates a more open structure when SCF is incorporated.The alumina membrane seems wellsintered and relatively homogenous.When SCF is added the images indicate more loosely sintered membranes with small particles less connected.The incorporation of SCF in the alumina membrane possibly hinders the sintering process due to the SCF limiting the alumina-alumina particle interaction, causing a rougher surface,a more open structure, a larger porosity and an increased pore size.In addition, the energy required for a reaction between SCF and alumina (formation of strontium aluminates) is possibly higher than that required for the thermodynamically advantageous sintering process for surface minimization.Therefore, the SCF-alumina membranes will exhibit larger pores than the alumina membrane due to the limited sintering, in agreement with the pore size distributions (Figure S9, Supporting Information) and porometry (Table 1).This morphological difference between alumina and SCF-alumina membranes, combined with the X-ray diffrac- tion results, indicates that the majority of SCF is possibly located near pores, as the sintering is hindered.The minor amount of SCF that reacts with alumina possibly forms a Sr-O-Al interface or an interlayer with the strontium aluminate phases.
Analysis by gas-liquid displacement porometry confirmed the presence of a majority of small pores <200 nm (Table 1) with the pure alumina membrane having the smallest pore size, whereas the functionalized membranes both have larger pores, suggesting a larger porosity.It should be noted that the complete pore size distribution could not be measured, as pores with diameters <74 nm could not be measured, since the membranes would break at the required pressures (>6 bar) to measure smaller pores.

Strength, Permeability, and Resistance
The porosity of the fabricated membranes increased from 51.0 to 53.5% with increasing SCF content, as observed in the inset of Figure 3a.The porosity of these membranes is like that of alumina membranes reported in literature and prepared by a similar method, [45] but higher than alumina membranes prepared by colloidal filtration. [46]This can be either caused by the less efficient packing during uniaxial compression, caused by the difference in shape and dimension of alumina and SCF, or it can be caused by the reaction between alumina and SCF forming the SrAl 2 O 4 phase.Therefore, the alumina-SCF contact causes formation of new particles due to reaction rather than sintering.The porosity greatly affects the flexural strength of the membranes, which decreased with increasing porosity (Figure 3a).The close-to-linear inverse relationship between strength and porosity has been already reported in the literature. [47]The strength of the membranes presented in this work is comparable to those obtained for alumina membranes fabricated in a similar manner, [45] including alumina-based supports used for catalytic membranes. [48]or example, Co 3 O 4 coated attapulgite catalytic ceramic membranes show bending strength in the range ≈ 30-60 MPa, although these were sintered at higher temperatures (up to 1050 °C), promoting the strength. [49]In general, the porositystrength trade-off for alumina-based membranes prepared from dry pressing is lower than for the membranes obtained by colloidal filtration. [46]However, our fabrication method is necessary to incorporate SCF in the alumina membrane and, although it reduces the temperature of the sintering process and thereby weakens the membrane, also gives the advantage of a higher permeability, thus balancing the strength with permeability.
Water and gas (nitrogen) permeabilities were measured for both pure alumina membrane and membranes with embedded SCF.Both permeabilities increased with increasing SCF content in the membranes (water permeability is shown in Figure S10, Supporting Information and gas permeability is shown in Figure S11, Supporting Information) because of the increasing pore size and/or number of pores (Table 1) easing the gas and water to pass through the membrane.Accordingly, the membrane hydraulic resistance showed decreasing resistance with higher SCF content and larger pore sizes (Figure 3b).The  13 6.0×10 13 8.0×10 13 1.0×10 14 1.2×10 14 1.4×10 14 1.6×10 14 1.8×10 14 low water permeability and hence the high hydraulic resistance (Figure 3b) is an effect of the larger thickness of 3.5 mm of the homogeneous membrane compared to other MF membranes.52], E.g., in Qin et al., [52] membranes with coated films of thicknesses 7.5-15.6μm ranged from 0.75 × 10 12 to 1.08 × 10 12 m −1 in hydraulic resistance.Moreover, the higher resistance of the homogeneous membranes fabricated in this study can be explained by the thicker active layer if compared with the conventional heterogeneous membranes.A Ce/TiO x coated ceramic membrane was found to have a membrane resistance of 1.00 to 1.23 × 10 11 m −1 , with an increase after coating. [53]An alternative approach is the incorporation of the catalytic material in ceramic hollow fiber membranes, whose low wall thickness is typically in the range 300-500 μm, i.e., an order of magnitude lower than that of commercial alumina membranes.By following this method, Wang et al. [54] prepared catalytic -Al 2 O 3 /CoFe 2 O 4 membranes with hydraulic resistance lower than 1.5 × 10 11 m −1 .

Catalytic Degradation of Organic Substances by Thermocatalytic Membrane
The retention or degradation of methyl orange at pH of 2 using an alumina membrane was negligible compared to the feed, and the minor drop in methyl orange concentration was most likely caused by adsorption to the membrane, as the pore size in the membrane is too large to retain methyl orange (Figure 4a).In contrast, methyl orange was removed (>99% after 9 mL) using the thermocatalytic SCF-alumina membranes.The removal was caused by degradation due to the thermocatalytic properties of the membrane, as no additional oxidants such as PMS (peroxymonosulfate) were added.If only adsorption had occurred, SCF catalyst should have been fully occupied by methyl orange, and therefore, the methyl orange concentration would have increased gradually.This was previously observed with SCF in a packed-bed reactor for oil removal when the flow was not sufficiently low. [40]The catalytic membranes showed great stability during methyl orange degradation, as >85% was removed at all times.The small active membrane area resulted in a low flux, and the duration of collecting 18 mL of permeate was ≈7 h.In addition, a previous study using SCF powder,in a distillation membrane apparatus, showed good stability of the catalyst through four consecutive cycles of degradation of bisphenol A. [42] As seen in Figure 4a, not all methyl orange was removed initially (85-92% removal).This is possibly due to an observed change in pH (Figure S12, Supporting Information), as the pH has great influence on the thermocatalytic degradation of methyl orange (Figure S3, Supporting Information).This is also supported by a study by Wang et al. who investigated ferrocene-based polymeric membranes for catalyzing Fenton reactions.They found pH (and temperature) to be of great importance in the degradation of methylene blue. [55]The pH increased rapidly from 2 to 8 during initial filtration followed by a slow decrease toward acidic environment with further filtration (Figure S12, Supporting Information).It is worth noting that the rate at which the pH returns to 2 depends on the amount of SCF in the membrane formulation.The higher the SCF concentration, the slower was the rate, supporting the argument of a significant effect of the SCF content.As the pH change was only observed for SCFalumina membranes and not for the pure alumina membrane nor in the batch tests using SCF, the reason for pH change must be due to newly occurring crystal phases that are being formed during the sintering (Figure 1).An additional experiment was performed to prove this hypothesis.SCF, AKP30, and a crushed membrane were all added to methyl orange at pH 2. No change was observed for SCF and AKP30, but the pH changed for the crushed membrane supporting the hypothesis that the newly formed phases are responsible for the pH change due to reactions with the acid.The degradation of bisphenol A (BPA) at neutral pH using thermocatalytic membranes is shown in Figure 4b.In contrast to of methyl orange, the removal of BPA required additional heat, as the kinetics was slow around room temperature even in batch degradation tests with SCF suspended in water. [42]ere, the bisphenol A solution was led through the membrane, and a sample was taken at room temperature to determine the initial concentration (c 0 ) just before heating the membrane module.The concentration of bisphenol A was found not to change with increasing temperature through the pure alumina membrane, whereas the concentration continuously decreased for the SCF-functionalized membranes (Figure S13, Supporting Information).A fast response to the heat was found for the thermocatalytic membranes, that instantly started to degrade bisphenol A. The continuous decrease of the BPA concentration is explained by the continuous increase in temperature from room temperature toward 60 °C, as the membrane module was heated by pouring 60 °C water over the module (see illustration of the setup in Figure S1, Supporting Information).This clearly indicates the importance of the temperature on the extent of degradation.Comparing the thermocatalytic membranes, the response shown for 5 and 15% was similar (both 48% at 12 mL), while the 10% SCF membrane showed a better performance (92-93% removal from 6 to 12 mL permeate).The higher degradation of the 10% SCF membrane compared to that of 15% SCF is interesting, as a higher perovskite concentration would have been expected to increase the catalytic activity.This might be related to the increasing reaction between alumina and SCF, which limits the SCF content in the 15% SCF-functionalized membrane.Another reason could be the increased permeability with increasing SCF content (see resistance in Figure 3b), resulting in too short contact times for the reaction between the SCF catalyst in the membrane and the BPA in solution when the SCF content exceeds 10 wt%.
Considering the optimum thermocatalytic membrane, multiple parameters must be taken into account.First, the thermocatalytic behavior is of course important to have the maximum activity that can be exploited by the thermocatalyst.Here, we showed that the membrane can operate at different pH depending on the targeting pollutant (pH 2 for methyl orange, and neutral pH for bisphenol A).Second, the pore size is important to both reject unwanted materials and providing sufficient contact time for catalytic degradation.Lastly, the strength of the membrane is of utmost importance, as this is one of the areas where ceramic membranes should be outstanding compared to polymeric membranes.According to these three parameters, the optimum SCF content in the membrane was found to be 10 wt%.This is a compromise between the wanted alumina and SCF phases and unwanted Sr-Al phases, such as SrAl 2 O 4 and SrFe 12-x Al x O 19 .These phases are critical for the catalytic performance and the higher is the SCF amount imbedded in the membrane, the more SCF reacts with the alumina phase forming the strontium aluminate phase as well as segregated ceria.The segregation of ceria itself does not assist in the degradation, however, it might result in a distortion of the perovskite structure enhancing the thermocatalytic activity as recently shown in literature. [56]The strength of these thermocatalytic membranes decreased with higher SCF content, and, thus, the SCF content was minimized.Alternatively, the membrane should be strengthened by other processes, and this is a key point for further research on fabrication of thermocatalytic ceramic membranes, thinking of full-scale filtration setups.The flux should be also controlled by varying the applied pressure and should not be too high, although in the current setup, it was not possible to study it.Therefore, future research will require upscaling of the membranes to increase the active membrane area, allowing a deeper study of filtration parameters such as pollutant concentration, temperature, and flux (contact time).In summary, a SCF content ≈10 wt% seems to be the optimum to have the highest catalytic rate of pollutant degradation and maintain a relatively good strength.

Conclusion
The thermocatalytic Ce-doped strontium ferrate perovskite with nominal composition Sr 0.85 Ce 0.15 FeO 3 (SCF) was successfully incorporated into an -alumina membrane while maintaining the thermocatalytic activity.For successful membrane fabrication and optimization of the perovskite structure in the membrane, a two-step heat treatment was applied to limit the reaction between SCF and alumina, ie.: first sintering in inert atmosphere, then re-conversion of Fe 3+ into Fe 4+ under oxidative atmosphere.All thermocatalytic membranes degraded methyl orange at room temperature without the use of oxidants.In contrast, the degradation of bisphenol A required higher temperatures, with the 10 wt% SCF-doped membrane degrading the most, while no removal was observed for the pure alumina membrane.This proves the thermocatalytic nature of this ceramic thermocatalytic membrane.When the SCF content increases from 0 to 15 wt%, porosity, pore size, and permeability also increased, while the strength of the membrane decreased.Therefore, the alumina membrane with 10 wt% SCF was recognized as the best compromise among the studied membranes.

Experimental Section
Perovskite Oxide Synthesis: Ce-doped SrFeO 3 with perovskite structure and Sr 0.85 Ce 0.15 FeO 3- (SCF) nominal composition was synthesized by dissolving 1.80 g Sr(NO 3 ) 2 (Acros Organics, purity 99+%), 0.65 g Ce(NO 3 ) 3 •6H 2 O (Sigma Aldrich, purity 99%), and 4.04 g Fe(NO 3 ) 3 •9H 2 O (Sigma Aldrich, purity ≥98%) in a 3 L glass beaker containing 200 mL distilled water.The solution was stirred using a magnetic stirrer with heating function.Then, 7.68 g of citric acid (Carl Roth, purity ≥99.5%) was added as a fuel and 9.26 g of NH 4 NO 3 (Sigma Aldrich, purity ≥ 99.5%) was added to regulate the reducers-to-oxidizers ratio at 1.5 (overstoichiometric).A pH of 6 was obtained by adjustment with NH 4 OH solution (25 vol.%).The solution was heated at 80 °C under continuous magnetic stirring and the water was left to evaporate until a green gel was formed.The magnet was then removed before the gel thickened and the temperature of the hotplate was set to 300 °C to induce a selfcombustion of the gel.The resulting powder was calcined at 1000 °C for 5 h with a temperature ramp of 2°C min −1 [57] to obtain the perovskite structure.
Membrane Fabrication: Alumina powder (AKP30, Sumitomo Chemicals) with mean particle size of 0.26 μm (according to manufacturer) was used as main membrane material.Pure alumina powder and mixtures of alumina powder with the synthesized SCF powder (5, 10, and 15 wt%) were prepared by hand mixing using a mortar and pestle for 10 min.One gram of powder was uniaxially compressed at 36 MPa in a stainless-steel mold into flat discs (Ø 13, H 3 mm).The discs were then sintered in a custom-made heating microscope (EM201x, Hesse Instruments) by heating at 5 °C min −1 to the desired temperature followed by cooling at 10 °C min −1 .Different isothermal heat treatments at maximum temperatures and different atmospheres (air and argon) were used for the various experiments.Additional membranes were prepared in green body dimensions of 80 × 50 × 3 mm 3 using 30 g of powder mixtures to investigate the mechanical properties.
Substrates and Membranes Characterization: The crystal structures of SCF, alumina, and the fabricated membranes were characterized by XRD using an Empyrean diffractometer (PANalytical).The SCF and alumina powders were analyzed as prepared and received, respectively, while the membranes were crushed and grinded prior to measurement.The diffractometer used a Cu K radiation ( = 1.5418Å) and operated at 45 kV and 40 mA.The recorded diffraction pattern was analyzed using Highscore+ software (PANalytical) by comparing to diffraction pattern from the International Centre for Diffraction Data (ICDD), for qualitative assessment.Quantitative analysis of the XRD data was performed by Rietveld refinement using the GSAS II software. [58]Chebyschev polynomial functions with eight polynomial terms were chosen for the background and Pseudo-Voigt (sum of Gaussian & Lorentzian) shape function for the peak profile fitting.In the structure refinement, unit cell parameters, sample displacement, and crystalline size were considered as variable parameters.Microstrain and thermal factors were not refined due to the multiple superposition of the peaks and the multiphase nature of the powder.From fitting results, the structural parameters of the investigated compounds and phase composition and the relative cell edge lengths (a-c) were obtained.The agreement factors and  2 were generally acceptable.
The total porosity ( tot ) of the membranes was calculated by Equation (1) by measuring the apparent ( app ) and powder ( pow ) densities.The apparent density was calculated by weight and dimensions of the pellets, while powder density was measured by a He-pycnometer (Ultrapyc 1200e, Quantachrome).
The morphology of selected samples was analyzed using scanning electron microscopy (SEM) operating at an accelerating voltage of 10 kV (EVO60, Zeiss).From the SEM images, the pore sizes were estimated using ImageJ-Fiji software. [59]The images were processed with a Fast Fourier Transformation Bandpass Filter followed by adjusting the threshold 10-15%.The pore size (determined as equivalent sphere diameter) was determined using the Analysis tool for Particle Size which is the voids between the solid phases.In addition, the pore size distribution was measured using gas-liquid porometry (Porolux 1000, Porometer GmbH, Germany), using a PorofilTM (Porometer GmbH, Germany) wetting solution.Membranes were adjusted in size (Ø = 25 mm), and the pore size distribution was determined by measuring liquid rates of displacement through the membranes with N 2 at pressures up to 6 bar in 20 steps.
Gas and water permeability of the membranes was tested using a custom-made membrane module in stainless steel at room temperature.The module could fit membranes with sizes dimensions from 11 mm ≤ Ø ≤ 13.5 mm and 2 mm ≤ H ≤ 4 mm, with an effective membrane diameter of 10 mm.The gas permeability was measured volumetrically using N 2 gas at room temperature and a transmembrane pressure (TMP) of 1-3 bar using a bubble-flowmeter and measuring the time for 5, 10, 15, and 20 mL of gas molecules to permeate through the membrane.The permeability was then determined from the molar gas flow at different gas pressures.The water permeation was determined from the mass flow by weighing the permeate passing through the membrane every 2 min using a Mettler Toledo PB3002-S balance and a custom-made script.The membrane hydraulic resistance, R m , was calculated from Equation (2): Where J is the permeate flux and μ is the dynamic viscosity of water.The mechanical strength of the fabricated membranes was determined by measuring the flexural strength using a three-point bending test of beams of the membrane material.Membrane beams were cut in the dimensions WxLxH of 4×50×3 mm 3 from an 80 × 50 × 3 mm 3 sheet using a Secotom-10 (Struers) and polishing using a 500 mesh SiC.The beams were bended using a CB500 microindenter (Nanovea).The bending strength ( f3r ) [MPa] was then calculated from Equation (3): Here, F [N] is the measured force at the point of fracture, L [mm] is the length between the supporting pins, while b [mm] and d [mm] are the width and height of the beam.
Catalytic Testing: For catalytic filtration experiments, a batch dead-end setup was prepared for testing of the membranes, where the driving force is compressed air to generate a TMP of 4 bar (Figure S1, Supporting Information).The membrane module was fabricated in Teflon (membrane dimension Ø = 13 mm and H = 3 mm, effective membrane diameter of 10.5 mm).For filtration experiments at elevated temperature, 60 °C water from a water bath was pumped over the membrane module, where the heat was transported through the module toward the membrane and feed solution by conduction.
Bisphenol A and methyl orange were used as model pollutants.Both compounds were prepared in 10 mg L −1 solutions, and the pH of the methyl orange solution was adjusted to two as this was shown to increase the rate of degradation (Figure S3, Supporting Information), which is in accordance with observations by Verduzco et al. [60] Both pollutants were degraded through the thermocatalytic properties of SCF without adding oxidants.The solutions were filtered through the membranes and the concentration in the permeate was analyzed for every 3 mL permeate.
The concentration of BPA and methyl orange was determined using HPLC-UV-vis.First, the collected samples were filtered through a 0.45 μm RC filter, then the liquid phase was analyzed through a HPLC using UV-detection (Summit-Dionex) using a Kinetex 5 μm EVO C18 column (150 × 4.6 mm).For methyl orange analysis, the mobile phase flow was 0.8 mL min −1 (acetonitrile/0.01m pH 6.8 ammonium acetate

Figure 1 .
Figure 1.XRD patterns of a) SCF-alumina powder mixtures heated at different temperature in air for 1 h, b) SCF-alumina powder mixtures sintered for different durations under argon flow at 870 °C, and c) crushed membranes with different SCF content after heat-treatment at 870 °C under argon flow follow by re-oxidation of the perovskite at 600 °C for 2 h in O 2 atmosphere.The phase compositions based on Rietveld refinements are shown in (d-f), where the errors are generally smaller than the size of the symbols.

Figure 3 .
Figure 3. a) Change in flexural strength with porosity (inset shows porosity change with SCF content) and b) membrane resistance of SCF-alumina membranes with increasing SCF.

Figure 4 .
Figure 4. Organic substance degradation by filtrations using SCF-doped alumina membranes.a) Concentration of methyl orange in permeate relative to initial concentration versus collected permeate volume by filtration at room temperature, and b) bisphenol A concentration after 12 mL permeate collected relative to initial concentration after heating membrane module by 60 °C water externally.Starting conditions are c 0 = 10 mg L −1 and pH of 2 (methyl range) and neutral pH (BPA).

Table 1 .
Mean pore size and total number of pores of alumina membranes with SCF loading between 0 and 15%, as measured by gas-liquid porometry.