Skip to main content

Graphene-Assembled Inorganic Hollow Fibre Membranes and Membrane Modules for Water Treatment

Final Report Summary - GRAPHENEHF (Graphene-Assembled Inorganic Hollow Fibre Membranes and Membrane Modules for Water Treatment)


Porous membrane-based filtration technologies nowadays are standard separation processes and widely used in water treatment, food processing, pharmaceutical production etc. whenever materials can be separated based on their size. Permeability of a membrane is inversely proportional to the thickness of the membrane. This rule means that the permeation flux can be improved by reducing membrane thickness. Due to this reason, most membranes for water treatment use an asymmetric architecture, where a very thin separation layer with a very small pore size is supported by a macroporous layer, so that the effective membrane thickness can be minimized while the strength of the membrane is maintained. Typically the effective thickness of the separation layer is few microns. Graphene is the ultimate thin two-dimensional material, which is composed of one layer of hexagonally-arranged carbon atoms. The two dimensional feature makes graphene an ideal material to form extremely thin membranes. Therefore it is expected that membranes made from graphene or graphene-like materials can achieve nano-scale thickness, open up a way to achieve higher permeation flux than conventional membranes.

In this research project, we aimed to realize high-performance graphene membranes supported by inorganic hollow fibres, which provide high surface/volume ratio and ease of assembly into industrial modules.


Two approaches were designed to achieve the research targets: the chemical vapour deposition (CVD) method and the dispersion-filtration method.

The CVD can obtain large-area continuous graphene with single- or few-atom layer thickness. The CVD method can only be used on a dense transition-metal surface, so that the obtained graphene will be continuous. However, to be used as membranes for separation, the substrate must be porous. It is possible to transfer the graphene film from the original substrate to a porous substrate, but it only fits for plane substrates and not possible for the hollow fibre geometry. Our design is to use a dual-phase metallic (copper and iron, for example) hollow fibre as the support for CVD growth of graphene films. As illustrated in Fig. 1, once continuous mono-layer or few-layer graphene film is obtained on the dual-phase hollow fibre, the copper phase can then be selectively etched by a proper etching agent, thus leave pores supported by the remaining iron phase. Finally, the graphene film will be selectively oxidized to create pores suitable for separation.

However, the designed CVD method is highly challenging because of the stark novelty. In case the novel method would not work satisfactory, we have an alternative plan which is more conventional but also can achieve the main target of the project. The alternative approach is to use an established dispersion-filtration method. The route is to prepare dispersion of exfoliated graphene oxide flakes from graphite, and then assembled by filtration to get multi-layer graphene oxide films on the prepared hollow fibre substrate. The films obtained are relative thick compared with the CVD method, usually some tens of nanometers, still much thinner than conventional membranes.


Dense Cu-SS (stainless steel) and Cu-Fe single-layer hollow fibres and dual-layer Cu/Cu-Fe and Cu/Cu-SS hollow fibres were successfully prepared for the purpose of growing single- and few-layer graphene by the CVD method. In Fig. 2, a Cu/Cu-Fe dual-layer hollow fibre is clear revealed. The outer copper layer is as thin as 20 µm after sintering, and dense copper surface is suitable for the CVD process. It is surprising that the morphology of the hollow fibre surface was changed to show fascinating patterns after the CVD process, which was perhaps due to the evaporation of copper at high temperature vacuum conditions. Figure 2d shows an SEM image of the surface of a sample after graphene growth, and Figure 2e and 2f give the related Raman spectrum that verifies full coverage of single-layer and few-layer graphene. Difficulties have however been met when trying to leach the copper phase out. Common leaching agents FeCl3, HNO3 and sodium persulfate were found to be destructive to the Cu/Cu-Fe hollow fibre. Cu/Cu-SS hollow fibres can be dissolved in FeCl3, but can survive in HNO3 and sodium persulfate, which can selectively remove copper but remains the porous network of SS, and therefore should be a better choice for the continued research. Few-layer graphene has also been grown on dense nickel layer plated on yttrium stabilised zirconia (YSZ) ceramic hollow fibre support. But again, leaching the dense substrate after graphene growth has however been difficult because the inert few-layer graphene covers both outer and inner surfaces of the nickel layer. The make the leaching possible, UV etching and catalytic oxidation were tried to make pores on the graphene/nickel/YSZ hollow fibres, but these efforts also failed within the time frame of the project. More efforts are needed in the continuous research to improve the pore creation and leaching process.

Due to the practical difficulties met with the CVD method, attentions were paid to the dispersion-filtration method to prepare graphene oxide membranes on hollow fibre supports. Graphene oxide dispersions were prepared first, via a modified Hummer’s method. GO membranes supported on ceramic hollow fibres were then successfully prepared by the filtration method. The microstructure of the GO membranes has been found unstable when being kept in air, and evidence from shrinkage tests points the reason to the stress generated during drying when the GO membrane start to shrink. To tackle this problem, a sacrificial layer is used to eliminate the stress. The sacrificial layer creates an extra space between the ceramic hollow fibre substrate and the GO membrane, which allows the GO membrane to shrink freely without introducing stress. The sacrificial layer stabilised GO membrane is very stable, a stringent long-term gas-tightness test showed that the membrane was impermeable within the 50-day testing duration. The membrane has also shown very low water permeation flux, but the permeability can be dramatically improved by UV treatment, which generates disorder in the GO laminar structure that serve as shortcuts for liquid transport. Such UV treated membranes allow water and acetone to pass through with a flux of 2.8 and 7.54 L m-2 h-1 bar-1, respectively, with a rejection rate of >90% to a dye methyl red (Mw = 269.3 Da).


The research shows that growing single-layer and few-layer graphene on hollow fibre by the CVD method is feasible, but difficulties of leaching the sacrificial phase in the support and pore creation on the graphene need to be solved. GO membranes are easier to be fabricated on hollow fibre supports, and the use of a sacrificial layer can stabilised the GO membrane, and proper permeation fluxes can be achieved with a low molecular weight cutoff. This research is the first attempt to grow graphene membrane on hollow fibre by the CVD method, and the experience accumulated here would be a foundation for similar studies. And the interesting pattern observed on the post-CVD copper surface will attract interests from materials scientists. The GO membrane study in this project help to clarify some facts that puzzle people in the field: it confirms that GO membranes are not permeable to gases, and a perfect GO membrane only allows liquids to pass with an extremely low permeability. It contradicts with some claims made by previous researches that GO membranes are gas permeable and water transport in GO laminar structures are fast. Furthermore, the ceramic and metal hollow fibres developed in this project can serve as porous microfiltration and ultrafiltration membranes for water treatment, and these hollow fibre themselves could attract more attention from the water treatment industry.