Graphene Oxide Membrane Technology for Refrigerant-Free Air Dehumidification

Abstract

Graphene oxide membranes have emerged as promising candidates for refrigerant-free air dehumidification because they combine high water-vapour permeance with high selectivity. Practical deployment has remained limited for several reasons. Most studies have used vacuum filtration to fabricate graphene oxide composite membranes. Although vacuum filtration aligns flakes well and yields high selectivity and permeance, scale-up remains costly. The brittle nature of graphene oxide has also necessitated porous membrane supports, which add mass-transfer resistance that can exceed the resistance of the selective layer itself. As a result, reported composite membranes often show permeances close to, or below, those of graphite oxide films ≈ 6 μm thick (1.01×10-5 \frac{mol}{m^2sPa}). This outcome persists despite the thinner graphene oxide layer that should produce a higher permeance from the composite membrane architecture. As such, measured permeance and selectivity are available for graphene oxide composite membranes with varied supports and oxidation levels, but few studies have examined their integration into larger-scale applications such as HVAC dehumidification systems. These studies rarely quantify how module-scale mass-transfer resistances affect measured membrane properties across operating conditions, even at the laboratory scale. Graphene oxide membranes make this question especially important because their high water-vapour flux and selectivity can make the bulk gas-phase transport resistance non-negligible. Bulk-gas concentration polarization can then skew the measured permeance as water-vapour flux approaches the diffusion rate from the bulk gas to the membrane surface. Most laboratory studies have also explored only a narrow range of vacuum conditions and therefore do not resolve the operating envelope that governs flux, crossover, and pump demand. Reported thermodynamic metrics often rely on idealized assumptions, including isothermal vacuum compression and vacuum pumps with 90% polytropic efficiency, rather than actual vacuum-pump power consumption under realistic operation. Together, these limitations leave no clear framework for separating intrinsic membrane performance from module-scale transport losses or for identifying the operating conditions that govern practical dehumidification. This thesis addressed these gaps through scalable graphene oxide membrane fabrication, simulations with bulk mass-transfer resistances, and system-level models that include transport and realistic vacuum-pump behaviour. This work produced composite membranes with coated areas of 225 cm2 with water-vapour permeances up to 1.5×10-5 \frac{mol}{m^2sPa}, and selectivity above 1000. Electrospun polystyrene supports with 93% porosity increased permeance by up to 27%, confirming that support resistance set the observed performance ceiling. Additional experiments and simulations showed that test-cell hydrodynamics changed the apparent permeance by ≈ 20% which required correction for gas-phase and support-layer resistances to recover the intrinsic membrane properties. System-level modelling showed that real vacuum pumps, pressure losses, residence time, and driving-force collapse reduced performance and limited the coefficient of performance to below 0.5 at the tested scale. These findings establish a scalable route to high-performance graphene oxide membranes and identify membrane materials, module design, and vacuum hardware as coupled constraints on HVAC implementation. Although the work in this thesis has moved the needle forward slightly, future work should target full module optimization under realistic operating conditions. Emphasis should be placed on membrane support architecture, gas-phase transport, thermal integration, and vacuum-pump losses. Reduced-order and dimensionless models could then provide a clearer basis for membrane sizing, operating-envelope mapping, and system scale-up.