|dc.description.abstract||Fresh water scarcity is quickly becoming a serious global challenge as populations grow and resources are depleted. Seawater, which makes up 97% of the water on Earth, can be a viable and sustainable source of usable water if energy-efficient, scalable, and cost-effective methods for desalination can be found. Based on molecular dynamic simulations, graphene-based materials used as desalination membranes can achieve nearly perfect salt rejection while maintaining orders of magnitude higher permeability than current commercial membranes. A new membrane-based separation technique called pervaporation (PV) has fundamental advantages over reverse osmosis (RO) and has potential to become a more energy-efficient technique compared to RO. In our work, we integrate graphene-based membranes with the PV technique for a scalable and high-performance desalination system.
Membranes based on graphene oxide (GO) are promising as a starting material for desalination membranes due to their scalable production, relative ease of processing and their multi-functional surface chemistry which can be beneficial for cross-linking and functionalizing the atomically thin sheets. Due to their unique structure of hydrophilic and hydrophobic domains, they are capable of selective water transport when stacked on top of each other to form thin or thick films.
In order to further enable the large-scale processing of GO into thin membranes, it is imperative to find solvents that meet the manufacturing requirements of high volatility and low toxicity which can also chemically and colloidally stabilize single layer dispersions. To this end, we study GO dispersions in a system of 1-alcohols in comparison to that of water which is conventionally used as a solvent. In this thesis, several unique phenomena are demonstrated that make a subset of these solvents ideal for processing and explore the benefits of casting membranes from these solvent systems using a variety of techniques for application in PV desalination.
In the first experimental chapter, the chemical and colloidal stability of GO was compared in both aqueous solvents and a series of 1-alcohols. The colloidal stability of these alcohol dispersions varies greatly compared to aqueous dispersion. As demonstrated by other groups, GO in water undergoes gelation around 1 wt% due to the strong electrostatic interaction between GO and water molecules, making it difficult to process. This does not occur in the alcohols, enabling us to achieve high dispersion concentrations of up to 8 wt%. This property was studied by simple centrifugation testing, where GO was seen to settle more easily in the alcoholic media. Despite settling, we were able to show with AFM imaging that GO can maintain an exfoliated state of mostly single layers in all the 1-alcohols. We also confirmed liquid crystalline (LC) properties in all alcohol dispersions with the help of polarized light microscopy, which has shown to be useful in making better-performing GO architectures compared to the conventional non-LC GO.
In addition to the processing advantages, alcohol dispersions also have chemical advantages. In aqueous media, GO is known to continuously undergo chemical transformation due to water’s nucleophilic nature. This chemical transformation causes a color change from yellow to dark brown, which suggests a more reduced nature of the material. This color change was observed to be inevitable in the aqueous dispersion but was accelerated in the alcohol dispersions when exposed to light. By quantifying these changes with UV-vis and FTIR spectroscopy, we study the evolution of the absorption coefficient of these dispersions at 230 nm, which is associated with the electronic transition of C=C bonds, and FTIR peaks at 1220 cm⁻¹ (C-O-C), 1300 cm⁻¹ and 1500 cm⁻¹, the latter two associated with carbon-carbon bonds. The absorption coefficient of aqueous GO dispersion at 230 nm was much higher than the alcohols, and it continues to increase with time regardless of conditions. However, all alcohol dispersions demonstrate a resistance to chemical transformation when stocks were sealed, but hexanol demonstrates resistance to chemical transformation even when the stock was continuously used. FTIR spectroscopy was used to support the chemical transformations indicated by UV-vis, suggesting loss of oxygen functionality and more carbon-carbon bonds as the dispersions were exposed to light. The trend with FTIR also suggested that hexanol provides some resistance to chemical transformation induced by light exposure while the other alcohols do not as much. Based on our AFM analysis, the flake sizes were much smaller in the water system compared to the alcohols. All these results indicate that GO in water undergoes chemical and physical changes, which may not be ideal for various applications that require a high oxygen functionality and larger flake sizes. This also suggests that GO is sensitive to light and should be stored away from light to lengthen lifetime.
Finally, we used the various dispersions to prepare vacuum-filtered GO membranes and compared the performance in PV desalination. We developed a custom-made PV module that required optimization of all components, including membrane module, fluid flow dynamics, condenser design, vacuum level, and permeate collection. We compared the performance of the system with a commercial membrane and a commercial PV module, and observed a 5% difference, which we deemed acceptable.
Our data indicated that properties of the vacuum-filtered membranes differed based on the solvent. First off, the interlayer distance depended on the type of solvent, which ranged from 9 to 11 Å. Using the native GO membranes resulted in swelling and leakage, which was mitigated with a zinc crosslinking to enhance the mechanical properties of the membrane. Even with this zinc-enhancement, the water-based membranes were failing with more than a 50% failure rate, compared to the alcohols that were around 20%. To sum up, there were no conclusive differences between the flux and rejection of the different solvent systems, however the 1-propanol and 1-butanol consistently performed better. We also observed a big improvement in flux with more hydrophilic and more porous support membranes. When testing the membranes at a 3.5 wt% NaCl feed solution at 30 °C, we observed a flux of 18.6 L/m²h and >99.8% with the GO membrane from 1-propanol at a loading of 40 μg/cm², where the previous work done by another group achieved a flux of 14.3 L/m²h at the same loading. This improved performance may be due to a combination of effects of zinc enhancement, an optimal interlayer distance and using a better support membrane.
In addition to the work in the application of PV, we tested the capabilities of GO dispersions in 1-alcohols in other membrane preparation techniques, namely solution casting and Langmuir-Blodgett (LB) deposition. Here, we demonstrated successfully that we were able to achieve high-crystallinity GO membranes from 1-alcohol dispersions. We also demonstrate the use of these 1-alcohols dispersions for the high yield transfer of GO monolayers as LB films. Together, these novel methods can yield an inexpensive, reproducible, and most importantly scalable process for creating high performance graphene-based desalination membranes.||en