PerspectiveMembranes

Scaling up nanoporous graphene membranes

See allHide authors and affiliations

Science  14 Jun 2019:
Vol. 364, Issue 6445, pp. 1033-1034
DOI: 10.1126/science.aax3103

Suitably engineered graphene-based materials could potentially be used to make water-separation membranes for applications such as desalination. Graphene-based materials with water-permeable pores can be made by creating either nanopores in graphene monolayers (see the figure, top) (1, 2) or two-dimensional (2D) channels that form between nanosheets of graphene oxide (GO) (see the figure, middle) (3). Both approaches face challenges for scaling to practical membrane sizes on the meter scale. The former requires creating a high density of subnanometer pores (4) on a defect-free monolayer graphene sheet that has high out-of-plane mechanical strength (5), and materials meeting all these requirements have largely been limited to micrometer-scale lateral dimensions. On page 1057 of this issue, Yang et al. (6) report production of nanoporous graphene on the centimeter scale that can reject between 85 and 97% of the salt from saltwater.

The difference in difficulty of scaling up the two approaches can be best appreciated by noting that unlike the challenging procedures for nanoporous graphene synthesis, centimeter-scale stacked GO membranes were readily made for initial bench-scale studies. Synthesis methods for GO have existed for decades, and it can be mass-produced through chemical oxidization and ultrasonic exfoliation of graphite (7). Thus, GO has proven less complicated and more economical to scale up.

The hydroxyl, carboxyl, and epoxide functional groups on GO can be chemically modified to help control the size of the 2D channel (the interlayer spacing) and produce membranes with different separation capabilities (3, 8). However, these functional groups also create the major problem with the stacked GO membrane—swelling in aqueous solutions (9)—which often results in low salt rejection (fraction of salt removed), typically in the range from 30 to 80%. Physical confinement—for example, with a polymer—can increase salt rejection to 97% (10). However, this strategy is feasible for centimeter-scale GO membranes, but may be more complicated to achieve in large-scale applications.

Unlike GO membranes, nanoporous monolayer graphene membranes are not inherently tolerant to defects. As the graphene membrane area increases, it becomes more prone to defects that can create openings larger than the required nanopore dimension. These defects cause a deterioration in separation capability (essentially leakage), and their presence is why most previous studies could only demonstrate reasonable desalination performance with nanoporous graphene membranes on a micrometer scale. Yang et al. overcome this problem by using chemical vapor deposition (CVD) to produce centimeter-scale graphene with high mechanical strength. Subsequent oxygen plasma etching created relatively uniform nanopores (diameters in the range of 0.3 to 1.2 nm), which are likely decorated by oxygenated functional groups. This study brings the membrane area to the centimeter scale, which is large enough to be tested in a bench-scale membrane system.

Toward practical graphene membranes

Combining two competing approaches may allow fabrication of meter-scale desalination membranes with two-dimensional materials.

GRAPHIC: N. DESAI/SCIENCE

Nonetheless, there are many challenges to extending this approach to create the meter-scale membranes needed for the membrane modules used in the desalination industry. Although the pore-size distribution is impressively narrow, it needs to be further narrowed to improve the salt rejection to >99% for seawater desalination. Atomistic simulations indicate that absolute salt rejection can be achieved when the pore diameter is <0.55 nm (2). Also, the synthesis technique faces substantial challenges to increase the membrane area to the meter scale because larger graphene sheets almost inevitably have larger-than-desired pores and more defects.

Scaling up the membrane area while finely controlling the functional groups on nanopore edges poses another challenge. The presence of certain functional groups (such as hydroxyl groups) decreases salt rejection, likely because of the formation of hydrogen bonds with ions that lower the free-energy barrier for ions to pass through the pores. Other functional groups (such as charged carboxylate groups) can increase salt rejection through charge effects (2).

A promising strategy for addressing the above scaling-up problems would be to make stacked nanoporous graphene (SNG) membranes from nanosheets in a manner similar to the fabrication of GO membranes. As a few layers of nanoporous graphene sheets are stacked, the functional groups on the edge of nanopores will work as spacers to keep interlayer channels open. Water will not only pass through the in-plane nanopores of each layer but also permeate through the interlayer channels (see the figure, bottom). The SNG membrane should have better scalability and performance because it combines several advantages of the nanoporous monolayer graphene membrane and the stacked GO membrane. It should have higher selectivity because it integrates multiple rejection mechanisms, including the gating effect (size exclusion and charge repulsion) of in-plane nanopores, the gating effect at the entrance of the 2D channel between neighboring porous graphene sheets, and the hindered diffusion of ions and molecules caused by the narrowness of the 2D channel. Stacking multiple layers of nanoporous graphene sheets would allow the defects and large pores in each individual sheet to be shielded by the neighboring sheets and minimize the ability of ions and molecules to bypass the membrane.

The SNG membrane would have a slightly lower water flux than the nanoporous monolayer graphene membrane (2) because of multilayer stacking, but would maintain a higher flux than the stacked GO membrane (11) because the stacking of porous graphene sheets would result in a less tortuous transport path than that in the stacked GO. The SNG membrane would also be less subjected to swelling than stacked GO membranes and have a less compromised nanostructure and performance. It should be much easier and cheaper to scale up SNG membranes to a meter-scale area by stacking centimeter-scale or even smaller nanoporous graphene sheets by using widely available coating techniques such as layer-by-layer assembly, vacuum filtration, and electrospray, rather than by synthesizing individual meter-scale nanoporous graphene sheets.

The challenges of scaling up 2D materials for high-efficiency membrane-based desalination call for in-depth exploration of creative ideas. The large family of related 2D materials could be used to overcome some of the technical problems facing the graphene-based membrane technology. For example, a 2D zeolite nanosheet with uniform pores of ∼0.5 nm oriented normal to the plane could be an ideal material for water-salt separation. Molybdenum disulfide (MoS2) has stronger van der Waals interaction between nanosheets and could be used to control the swelling of stacked graphene-based membranes (12) by forming a stacked porous graphene-MoS2 composite structure.

References and Notes

Acknowledgments: The author is supported by NSF award nos. CBET-1565452 and CBET-1706059.
View Abstract

Stay Connected to Science

Navigate This Article