Uniform hexagonal graphene film growth on liquid copper surface: Challenges still remain

Uniform hexagonal graphene film growth on liquid copper surface: Challenges still remain

Journal Article



In a recent publication Geng et al. (1) reported the growth of graphene thin films on liquid Cu surface, using chemical vapor deposition. The authors state that the graphene nucleation typically occurs on grain boundaries in solid polycrystalline Cu, which results in an inhomogeneous density and size distribution of hexagonal graphene flakes. Therefore, liquefaction of the Cu leads to the formation of a uniform single-layer graphene. Although I agree with the general description and the fact that liquefaction of polycrystalline Cu will eliminate grain boundaries, I think the conclusion that it leads to the synthesis of continuous, uniform graphene film in general is misleading.

In fact, a liquid layer (Cu) with an adsorbate on the surface (C) inevitably produces domains/cells based on solutal or thermal instabilities driven by variations in surface tension, which can be described for instance by Benard-Marangoni convection (2). In a recent article (3) we showed that the solutal Marangoni number for the surface-melted Cu–C system can be higher than the critical number M > MC = 80, which initiates the convection phenomenon driven by a gradient in carbon concentration. As a result, the topographic patterns in the form of convectional isolated/connected hexagonal or vermicular cells have been observed in grown graphene. Depending on specific experimental conditions such as the thickness of the liquid layer, concentration gradient of the carbon, and temperature gradient between the surface and the core, these cells assemble into very regular hexagonal (honeycomb) patterns like the authors have observed in (1). Hence, on the one hand, liquefaction of Cu does indeed eliminate the grain boundaries, but on the other hand, it causes the formation of domains driven by surface tension variation on the Cu–C interface. Moreover, the source of new domain centers can be a small amount (10−6–10−2 wt%) of common impurities, because the authors used Cu with 99.8% purity.

In addition, cooling down of the Cu–C melt to room temperature accompanied by the solidification process inevitably initiates planar or convective instabilities driven by solutal or thermal capillary forces, which will break down the planar interface of a dilute alloy into cells, leading to inhomogeneous distribution of carbon on the surface and thereby formation of ripples and variation of the numbers of graphene layers (3). These processes can be described on the basis of Mullins–Sekerka instabilities in dilute binary alloys (4). According to this theory, an arbitrary perturbation due to the imposed temperature/solutal gradient initiates a cellular structure and develops a pattern with an initial wavelength λ typically on the order of microns, which is in the same size range as that of the images observed by Geng et al. (1). In the absence of specific experimental information, more accurate domain size estimation is hard to perform.

In my opinion the above discussion highlights the interplay between the onset of instability and graphene formation, which can be seen as an obstacle for large single-domain graphene growth. The wavelength of the instability will put a new upper limit on the domain size of graphene grown from the melt.


Proceedings of the National Academy of Sciences of the United States of America 109, E2099-E2099
31 Jul 2012
A. Harutyunyan