Potable water scarcity is a global problem which could be ameliorated by membrane desalination technologies as there is a significant proportion of saline and/or sea water on our planet. Led by a maturing reverse osmosis (RO) technology, pervaporation and/or membrane distillation processes have also attracted considerable attention for desalination, though inorganic membranes are still at early research stages. For example, silica membranes in pervaporation processes have delivered low water fluxes (2 kg m–2 h–1). However, there are many advantages in using sol-gel derived silica membranes for desalination because of the micropore size control of silica structures, which allow the diffusion of smaller water molecules and hinder the passage of larger hydrated salt ions. The silica membranes reported in the literature are typically synthesized from acid-catalyzed sol-gel method (pH ~2), which forms a large fraction of hydrophilic silanol groups but renders the silica membranes unstable for desalination. Moreover, silica membranes are generally coated on α-alumina substrates that are primed with γ-alumina layers as interlayers. These interlayers add extra cost in the membrane production and are required to smooth the coating surface to reduce pin holes and micro-crack defects of silica matrices.
It is initially postulated that silica membranes can be prepared via a base-catalyzed sol-gel method to produce more siloxane bridges, creating both mesoporous and microporous structures, whereby the mesopores enhance water fluxes and the micropores control salt rejection, thus leading to high performance with the added benefit of being more hydrostable. A second postulation was raised about the robustness of base-catalyzed sol-gel method in preparing defect-free silica thin films, a novel process called “interlayer-free” membranes. Therefore, a systematic investigation of sol-gel synthesis accompanied by silica xerogel characterization, thin-film calcination and membrane performance was carried out. All membranes were prepared by the base-catalyzed sol-gel and interlayer-free methods. Results for pure silica membranes show that the average water fluxes (and salt rejection) decrease with increasing salt concentration from 9.5 (99.6%) to 1.55 kg m–2 h–1 (89.2%) for the 0.3 and 15 wt% saline solutions, respectively. It is noteworthy that these membranes could deliver higher water fluxes than reported silica membranes and could also process brine concentrations (15 wt%), which are well beyond the capability of RO membranes. The results strongly suggest that an increase in the silica porosity caused by the siloxane species did not compromise the ability of the membrane to reject salt. It was further postulated that percolation pathways were still controlled by the bottle necks in the silica matrix attributed to the amorphous nature of silica. Long term (250 hours) testing showed a ~20% decrease in water flux, though 99% salt rejection was maintained. The decrease in water flux is associated with closure of micropores (possibly containing silanol groups), and slight densification of silica matrix.
To further improve hydrostability to the silica membranes, it was postulated that the silica matrices could be functionalized with other materials whilst maintaining high water fluxes and salt rejections. To test these hypotheses, three different strategies were selected namely: (i) using a new silica precursor based on organo-silica (tetraethyl vinylsilane - TEVS), (ii) carbon-templating the silica matrix with tri-block copolymer (Pluronic® P123), and (iii) with metal oxide using cobalt nitrate hexahydrates. Initial results were remarkable, as salt rejection was very high above 99.6% whilst reaching high fluxes of 23.9 (Si-TEVS), 6.81 (Si-P123) and 7.66 (Si-CoOx) kg m-2 h-1. Similarly to pure silica membranes, the water fluxes were a function of temperature, thus demonstrating that the transport of water molecules was dependent on the water vapour pressure difference.
The further final test was the long-term reliability and hydrostability of the functionalized silica membranes operating at room temperature and 3.5 wt% salt solution over 500 hours. The high flux (Si-TEVS) proved to be unstable within 250 hours testing as fluxes increased from 10.6 to 15.8 kg m-2 h-1 and salt rejections greatly reduced from 99.9 to 50.1%. The Si-CoOx membrane was more stable with a slight reduction in water fluxes from 4.66 to 3.71 kg m-2 h-1 and salt rejections from 99.6 to 99.4%. It was observed that cobalt oxide leached from the membrane matrix leading to performance deterioration. Lastly, the lower flux (Si-P123) proved to be very hydrostable with minor variations in water fluxes from 2.53 to 2.28 kg m-2 h-1 and salt rejections from 99.9 to 99.6%. P123 carbon-templating proved to be a better strategy in maintaining the long term integrity of the silica matrix.
In summary, this work is the first time that high performance silica membranes were prepared via base-catalyzed sol-gel and interlayer-free methods. Molecular sieving structures were achieved as the membranes separated water from ionic salts. Long-term testing showed that carbon templated silica membranes were robust to oppose the detrimental effect of silica deterioration.