The role of molten-salt distribution in dual-phase ceramic molten-salt membranes

Abstract

Dual-phase ceramic molten-salt membranes offer high permselectivity for carbon dioxide at high temperature (> 400 °C) under continuous operation and offer a reduced environmental footprint compared to current carbon dioxide separation technologies. However, these membranes have not been developed with consideration to controlling the distribution of the molten carbonate phase within their ceramic supports, which can decrease molten carbonate effective thickness and increase surface area for carbon dioxide desorption and feasibly improve flux. Furthermore, the highly wetting nature of molten salts on ceramic supports could be exploited to spread and self-heal cracks developed during long-term operation attributed to the brittle nature of ceramic supports. Here, carbon dioxide permeation was initially investigated in a model dual-phase membrane, where the support did not contribute to the permeation mechanism so that the permeation of carbon dioxide was restricted to the molten phase alone. Surfaceexchange reactions were found to be rate-limiting on carbon dioxide flux between 450 – 750 °C, whereas bulk-diffusion limitations occurred above 750 °C. Thus, carbon dioxide flux can be increased in the surface-exchange-limited region by increasing the available surface area for carbon dioxide desorption (chapter 4). Subsequently, an asymmetric hollow fibre with high and tailorable surface area for carbon dioxide desorption was used as the membrane support. Asymmetric hollowfibre supports (widely used in polymeric and ceramic membrane systems) comprise two distinct porosity domains: micro-channels, conically shaped with an open entrance on the lumen/permeate-side surface (pore size: 2 – 30 μm), and a porous microstructured packed-pore network (pore size: 0.05 – 0.5 μm), located between microchannels and at the shell/feed-side surface of the hollow-fibre supports. Hence, the permeate-side surface area of the hollow fibre supports consists of the projected surface areas of the micro-channels and the areas between them, whereas the feedside surface is the projected surface area of a cylinder. So far, the molten phase has been infiltrated into both porosity domains of the hollow fibre supports in an uncontrolled way, sacrificing gaseous mass transfer advantages of the micro-channels and decreasing the available interfacial area between gaseous carbon dioxide and molten carbonate from the projected surface areas of the micro-channels and the areas between them to that of a cylinder. A new carbonate infiltration method was developed in this work, aiming to control the incorporation and distribution of the molten phase inside the packed-pore network of the hollow-fibre supports alone. As the infiltration targeted the incorporation of the carbonates in the packed-pore network alone, leaving the micro-channels unblocked, the interfacial area between molten carbonate and carbon dioxide increased (chapter 3). In the controlled-infiltrated hollow-fibre membranes the porous micro-channels remained unblocked, increasing the surface area for carbon dioxide desorption to that of the sum of the micro-channels surface areas. In the uncontrolled-infiltrated membranes the surface area for carbon dioxide desorption was ~5 times smaller as the surface area of the micro-channels was blocked. The increase in surface area in the membranes developed by the controlled-infiltration method showed an 8-fold flux improvement (0.036 ml min-1 cm-2) at 600 °C compared to membranes where the molten-salt distribution, and available surface area, was uncontrolled (0.004 ml min-1 cm-2) (chapter 4). Finally, molten carbonate wetting on the ceramic surface, was exploited to self-heal cracks in membrane supports. A membrane system with an incorporated sacrificial crack-inducing material demonstrated that the molten phase can spread and self-heal the catastrophic crack created upon removal of the crack-inducing material. The permselectivity of the membrane was restored after self-healing occurred, demonstrating the first autonomous and intrinsic self-healing membrane (chapter 5).