Scientists Engineer Thinner Ceramic-Carbonate Membranes for High-Temperature CO2 Capture

…

…

The choice of 873 K as the operating temperature for the permeation tests is not arbitrary. The ternary eutectic mixture of lithium, sodium, and potassium carbonates used in this study melts at approximately 661 K, which means that at 873 K the carbonate phase is fully molten and mobile within the pores of the ceramic scaffold. This is a critical requirement for the membrane to function, because the separation mechanism depends entirely on the presence of a continuous liquid carbonate pathway. If the carbonate were solid, carbonate ions could not diffuse across the membrane, and no CO2 transport would occur. The eutectic composition, 43.5 mol% lithium carbonate, 31.5 mol% sodium carbonate, and 25 mol% potassium carbonate, was selected because it offers the lowest melting point of any combination of these three salts, together with relatively high ionic conductivity in the molten state. Lowering the melting point matters for practical deployment: the further the operating temperature sits above the melting point, the lower the viscosity of the molten phase and the faster the diffusion of carbonate ions through it, both of which favor higher permeance. At the same time, operating at 873 K rather than at much higher temperatures reduces thermal stress on the stainless-steel support and on the seals and module components that would surround the membrane in an industrial setting.

The ionic transport mechanism underlying these membranes deserves closer examination, because it explains why the microstructure of the ceramic layer is so central to performance. When CO2 from the feed gas contacts the membrane surface, it reacts at the gas-carbonate interface with oxide ions that are supplied by the ceramic phase, forming carbonate ions. These ions then hop through the molten carbonate, driven by the difference in CO2 partial pressure between the feed side and the permeate side. At the downstream interface, the reverse reaction releases CO2 back into the gas phase and regenerates the oxide ion, which diffuses back through the ceramic scaffold to complete the cycle. The ceramic phase therefore acts as a solid-state shuttle for oxide ions, while the molten carbonate acts as the medium in which carbon is carried as CO3 2-. Neither phase alone can transport CO2; the membrane only works where the two phases are in intimate, continuous contact. This is precisely why the pore-forming agents evaluated in the study are so important. By increasing the porosity of the YSZ layer and controlling the size and connectivity of the pores, PMMA and starch increase the interfacial area between the ceramic and the carbonate, shortening the diffusion path for oxide ions in the solid and for carbonate ions in the liquid. A well-developed, interconnected pore network also ensures that the molten carbonate, introduced by vacuum-assisted immersion, spreads uniformly rather than leaving isolated pockets that would act as dead volume.

The use of vacuum-assisted dip coating for the ceramic deposition and vacuum-assisted immersion for the carbonate infiltration reflects a broader principle in membrane fabrication: removing trapped air is often the difference between a functional composite and a defective one. In conventional dip coating onto a porous substrate, capillary forces draw the suspension into the surface pores, but air trapped within the pore network can prevent complete wetting, leading to pinholes, incomplete coverage, and delamination during calcination. Applying vacuum before or during immersion evacuates this air, so that when the suspension contacts the substrate it penetrates uniformly into every accessible pore. For the carbonate infiltration step this is even more consequential, because the molten carbonate must fill the fine porosity of the ceramic layer completely while being prevented from flooding the full thickness of the porous stainless-steel support. If carbonate were to penetrate beyond the ceramic layer and reach the metal support, it could short-circuit the selective layer or react with the steel at high temperature. The in situ melting step, in which the infiltrated carbonate powders are heated to 873 K within the membrane itself, allows the molten salt to flow and redistribute under capillary forces, healing small voids and establishing the continuous liquid pathway that the permeation mechanism requires.

The selection of 316L stainless steel as the substrate material introduces both advantages and constraints worth considering. Stainless steel offers mechanical robustness, thermal shock resistance, and ease of sealing compared with ceramic supports, which are brittle and difficult to join to metal housings in module assemblies. The ability to weld or gasket a metallic support simplifies the engineering of high-temperature membrane modules, which is one of the persistent obstacles to deploying ceramic membranes at industrial scale. However, 316L steel contains chromium, nickel, and molybdenum, and at 873 K in contact with molten alkali carbonates these elements can corrode, particularly in the presence of dissolved oxygen or water vapor. The ceramic YSZ layer serves a dual role here: it is not only the oxide-ion conductor but also a protective barrier that limits direct contact between the molten carbonate and the metal. This makes the completeness of the coating, verified by the scanning electron microscopy observations showing uniform coverage with no uncovered zones after three deposition-calcination cycles, important for long-term chemical stability as well as for separation performance. Any pinhole or crack would expose the steel to the corrosive melt and could become a non-selective leak path for CO2.

The calcination protocol used between deposition cycles also merits comment. Each cycle comprised five five-minute immersions in the YSZ-polyvinyl alcohol suspension followed by drying, and calcination at 723 K for twelve hours with a slow heating ramp of one degree Celsius per minute. The slow ramp allows the polyvinyl alcohol binder and the pore-forming agents to decompose gradually; heating too quickly would cause rapid burnout of the organic phases, generating gas pressures that crack or blister the coating. The twelve-hour hold at 723 K ensures that the organic material is fully removed and that the zirconia particles sinter into a mechanically coherent layer without densifying to the point of closing the porosity that the carbonate infiltration later requires. The fact that three such cycles were needed to achieve near-complete coverage of the substrate illustrates a common trade-off in slip coating: each cycle adds thickness and closes defects, but also adds processing time and cost, and excessive thickness would increase the diffusion resistance that the study is trying to minimize.

Comparing the two pore-forming agents highlights how template chemistry shapes the resulting microstructure. PMMA spheres are synthetic polymer particles available in controlled, narrow size distributions, and they burn out cleanly during calcination, leaving behind well-defined spherical pores whose size can be tuned by selecting the particle grade. Starch, by contrast, is a natural biopolymer whose granules vary in size and shape and which decomposes through a more complex sequence of gelatinization, dehydration, and combustion. Starch tends to produce a broader and more irregular pore size distribution, which can be advantageous where highly interconnected porosity is desired, while PMMA offers more reproducible, monodisperse porosity when tight control is needed. The study evaluated both agents precisely because the optimal pore architecture for carbonate infiltration, oxide-ion conduction, and mechanical integrity is not known a priori and depends on the interplay between pore volume, pore connectivity, and the wetting behavior of the molten carbonate on zirconia.

The preliminary permeation results, showing measurable CO2 transport at 873 K with flux values comparable to related ceramic-carbonate dual-phase membranes reported in the literature, place this work within a growing body of evidence that supported thin-film configurations can match the performance of thicker self-supported membranes while using far less material. Because the transport of CO2 through these membranes is controlled primarily by oxygen-ion conduction through the ceramic phase, and because that resistance scales with thickness, confining the dual-phase layer to the first few micrometers of the substrate surface is among the most direct routes to higher flux. The porous stainless-steel support carries the mechanical load, allowing the selective layer to be thin without sacrificing strength. This asymmetric architecture, with a graded or thin functional layer on a robust porous backing, mirrors design principles long established in gas separation and reverse osmosis membranes and is now being adapted to high-temperature electrochemical separation systems.

Several questions remain open as this approach matures. The long-term chemical stability of the YSZ-carbonate-steel assembly under sustained thermal cycling and realistic flue-gas compositions, which contain water vapor, sulfur oxides, and particulate contaminants, will need to be demonstrated over thousands of hours. Sulfur dioxide in particular can react with alkali carbonates to form sulfate salts, potentially depleting the molten phase or altering its melting behavior. The selectivity of the membrane toward CO2 relative to nitrogen and oxygen, which determines its usefulness for post-combustion capture, depends on the membrane being effectively dense to gaseous species while permitting only ionic transport, and any residual open porosity in the carbonate-filled layer would degrade that selectivity. Finally, scaling the vacuum-assisted deposition and infiltration procedures from 1.27 cm discs to large-area modules will require process controls that maintain coating uniformity across much greater dimensions. The morphological optimization reported here, demonstrating complete, adherent, porosity-tuned YSZ coatings on metallic supports with successful molten carbonate infiltration, provides the materials foundation on which those subsequent engineering steps can build.

Subject of Research: …

Article Title: Upgrading morphological properties of ceramic-carbonate coating for high-temperature CO2 capture

Article References: Tarditi, A. M. (2026). Upgrading morphological properties of ceramic-carbonate coating for high-temperature CO2 capture. Frontiers in Membrane Science and Technology, 5, Article 1871627. https://doi.org/10.3389/frmst.2026.1871627

Image Credits: AI Generated

DOI: 10.3389/frmst.2026.1871627

Keywords: carbon capture, membranes, ceramics, CO2 separation, materials science, Upgrading, morphological, properties, ceramic-carbonate, coating, high-temperature, capture