Incorporation of a high density of molecular-sieving nanopores in the graphene lattice by the bottom-up synthesis is highly attractive for high-performance membranes. Herein, we achieve this by a controlled synthesis of nanocrystalline graphene where incomplete growth of a few nanometer-sized, misoriented grains generates molecular-sized pores in the lattice. The density of pores is comparable to that obtained by the state-of-the-art postsynthetic etching (10
12 cm
−2) and is up to two orders of magnitude higher than that of molecular-sieving intrinsic vacancy defects in single-layer graphene (SLG) prepared by chemical vapor deposition. The porous nanocrystalline graphene (PNG) films are synthesized by precipitation of C dissolved in the Ni matrix where the C concentration is regulated by controlled pyrolysis of precursors (polymers and/or sugar). The PNG film is made of few-layered graphene except near the grain edge where the grains taper down to a single layer and eventually terminate into vacancy defects at a node where three or more grains meet. This unique nanostructure is highly attractive for the membranes because the layered domains improve the mechanical robustness of the film while the atom-thick molecular-sized apertures allow the realization of large gas transport. The combination of gas permeance and gas pair selectivity is comparable to that from the nanoporous SLG membranes prepared by state-of-the-art postsynthetic lattice etching. Overall, the method reported here improves the scale-up potential of graphene membranes by cutting down the processing steps.The direct bottom-up synthesis of two-dimensional (2D) films hosting molecular-sieving apertures is a longstanding goal in material chemistry. Materials with a high density of molecular-sized nanopores such as zeolites (
1,
2), metal-organic frameworks (MOFs) (
3,
4), covalent organic frameworks (
5), graphitic carbon nitride (
6), and protein channels (
7) can be synthesized in a 2D topology, albeit in the form of nanosheets with lateral dimension limited to ∼O (1 µm), which prohibits the realization of porous, unit-cell–thick films by the bottom-up synthesis. On the other hand, the synthesis of high-quality atom-thick polycrystalline films of graphene (
8) and hexagonal boron nitride (
h-BN) (
9) can be carried out in a scalable manner, with the production capacity of graphene films exceeding three million m
2 in 2017 (
10). However, their lattice is impermeable in the absence of a sufficient electron-density gap for molecular transport with a reasonable energy barrier (
11).The lattice of graphene can be engineered by postsynthetic etching to obtain attractive separation performance, for example, by using an electron (
12) and ion (
13) beam or by chemical etching techniques involving O
2 (
14), O
2 plasma (
15,
16), O
3 (
17–
19), or UV/O
3 (
20). However, the postsynthetic lattice etching adds an additional and often complex processing step in the overall fabrication process. It is attractive to avoid this step from the point of view of membrane scale-up. This makes the bottom-up approach, capable of synthesizing a nanoporous graphene film in a single synthesis step, extremely desirable. For example, Ullmann coupling is a highly promising route that yields porous lattice with a precise pore structure (
21); however, significant advances are needed to enlarge the ordered domains, which are currently only a few nanometers in size. Another promising strategy is to engineer the catalyst to selectively avoid graphene growth where pores are desired. For example, Park and coworkers deposited 2 to 10 nm tungsten islands in a Cu foil to locally shield the Cu surface during growth giving rise to a porous graphene with 20 to 50 nm pores and a pore density of ∼10
10 cm
−2 (
22).An attractive approach for the bottom-up synthesis of nanoporous graphene is to promote the incorporation of intrinsic vacancy defects in the graphene lattice during its crystallization. A proof of principle of this idea for molecular separation has been demonstrated by altering the chemical vapor deposition conditions of polycrystalline single-layer graphene (SLG) on a catalytic Cu foil (
17,
18). Tuning the synthesis temperature (
23), the precursor (methane, benzene) (
24), and the roughness/orientation of the catalytic Cu foil (
25) has been shown to yield intrinsic vacancy defects that could separate small molecules based on their size, for example H
2 from CH
4. However, the SLG’s true potential of ultrahigh molecular permeance could not be realized in these reports because the density of vacancy defects was low. To put this into perspective, the graphene lattice hosts 3.8 × 10
15 atoms cm
−2, and the highest reported density of molecular-sieving vacancy defects is ∼5 × 10
10 cm
−2 (
24), whereas postsynthetic etching has been successfully used to incorporate defect density of 10
12 cm
−2 (
16,
19,
26). Therefore, the development of a direct bottom-up method incorporating a high density of molecular-sized intrinsic vacancy defects in graphene would be highly attractive.A major source of intrinsic vacancy defects in polycrystalline graphene is the incomplete intergrowth of misaligned grains (
27–
34). Therefore, to increase the density of such defects, one needs to reduce the grain size. Also, to make the vacancies large enough for selective molecular translocation, one needs to increase the misalignment between the grains. To the best of our knowledge, growth conditions that promote small, misoriented grains incorporating a high density of molecular-sized vacancies in graphene have not been reported. We note that while there have been several reports on nanocrystalline graphene, none of them reported a porous lattice. For example, a helium leak test carried out on sealed nanocrystalline graphene film, synthesized via the quenching approach, verified the absence of He-permeable vacancy defects in such films (
35). Nanocrystalline graphene films have also been synthesized by electron-radiation–induced cross-linking of aromatic self-assembled monolayers on Au (
36), graphitization of polymeric precursor on Ge (
37), diffusion of carbon through Ni grain boundaries (
38), quenching of Pt foil in a carbon solution (
35,
39), etc. However, molecular-scale vacancy defects were not observed.Herein, we report a graphene synthesis route based on the controlled precipitation and crystallization of carbon on Ni surface at a low temperature that limits the grain size in graphene to less than 5 nm and allows the realization of a high density (∼10
12 cm
−2) of molecular-sieving intrinsic vacancy defects or nanopores by the bottom-up synthetic route. The nanocrystalline graphene is few layered (two to four layers) except near the nanopores where the grains taper into a single layer and eventually terminate leading to vacancies (). This nanostructure, which we refer to as porous nanocrystalline graphene (PNG), is highly attractive for the membrane-based molecular separation because the layered domains increase the mechanical robustness of the film, and the atom-thick apertures allow the realization of high-permeance attributing to the single rate-limiting transition state at the center of the pore (
40).
Open in a separate windowThe nanostructure of PNG film. (
A) Schematic of PNG showing the nanopores that form when three or more nanocrystalline graphene grains with different orientations meet. (
B) AC-HRTEM image of the PNG-1 synthesized by exposing a 25 μm thick Ni foil to the pyrolysis products of PS-
b-P4VP and turanose (1:2 weight ratio) at 500 °C. The orange arrows point toward the nanopores, and the yellow squares indicate the areas used for computing the FFT images presented at the bottom of the panel. The scale bars in the FFT indicate 2 nm
−1, and the angles indicate the relative rotation of the graphene grains from an arbitrary reference point. (
C) AC-HRTEM images of several nanopores. (Scale bar, 1 nm.) The PNG-1 samples were imaged directly after transferring to the TEM grid. No posttreatment or cleaning protocol was used before imaging.The presence of high-density intrinsic vacancy defects led to extremely high H
2 permeance (38,000 gas permeation units or GPU; 1 GPU = 3.35 × 10
−10 mol m
−2 s
−1 Pa
−1). The pores were commensurate with the size of H
2 indicated by activated transport of H
2 and attractive H
2/CH
4, H
2/N
2 and CO
2/N
2 selectivities. The combination of permeance and selectivity is comparable to the state-of-the-art graphene membranes prepared by postsynthetic lattice etching (
19,
41). Additionally, functionalization and masking methods previously demonstrated to improve the performance of postsynthetically etched graphene could be applied to PNG to fine-tune the application-specific performance. For example, oxygen functionalization of the pores increased H
2/CH
4 and H
2/N
2 selectivities by 150 and 130%, respectively. Masking PNG with CO
2-phillic polymers led to attractive carbon capture performance with CO
2 permeance of 5,700 GPU and a CO
2/N
2 separation factor of 31.
相似文献