Simulations on growth mechanism of graphene

The project aims at the investigation, at the atomic level and by multiscale simulations, of the mechanism of growth of pristine and chemically doped graphene.

From atomic carbon to graphene on Cu(111) : stable configurations by ab-initio methods

  Graphene growth by chemical vapor deposition on copper is one of the most popular method to obtain large scale sample. If the commensurability of graphene with Cu(111) plays a determinant role, the most stable geometries for the 2D crystal do not correspond to the most stable adsorption sites of individual carbon atoms on the same surface. In this paper [1], we analyzed this contradiction based on density functional theory calculations. From the three stable sites for isolated carbon atoms on Cu(111), only two of them are involved when small clusters of carbon are adsorbed. However, because of the shift from strong C-Cu interaction for isolated (or unsaturated C atoms) to weak van der Waals C-Cu bonding, other stable geometries are found for adsorbed infinite graphene. We propose here two new stable graphene adsorption geometries and we present a detailed analysis of the various adsorption geometries.

configuration

Figure 1. Supercell used to model the Cu(111) surface. (a) Side view of the supercell. The atoms of the first three Cu layers are displayed with blue, grey and yellow colors (b) top view of the supercell. The top, hcp and fcc adsorption positions correspond to carbon atoms on the top of the first, second and third Cu layer, respectively. Other positions (btop, bhcp and bfcc) correspond to the mid-point between the first top, hcp and fcc sites.

 

Multiscale simulations of the early stages of the graphene growth on copper

We have performed multiscale simulations of the growth of graphene on defect-free copper (111) in order to model the nucleation and growth of graphene flakes during chemical vapour deposition and potentially guide future experimental work. Basic activation energies for atomic surface diffusion were determined by ab initio calculations. Larger scale growth was obtained within a kinetic Monte Carlo approach (KMC) with parameters based on the ab initio results. The KMC approach counts the first and second neighbours to determine the probability of surface diffusion. We report qualitative results on the size and shape of the graphene islands as a function of deposition flux. The dominance of graphene zigzag edges for low deposition flux, also observed experimentally, is explained by its larger dynamical stability that the present model fully reproduced. [2]

 events

Table 1: List of diffusion events. Brown balls and sticks correspond to carbon atoms and bonds. Blue, yellow and grey balls represent top, hcp and fcc sites. Red circles highlight carbon atoms that move between the initial configuration (left) and the final configuration (right). The diffusion events are shown in the schematic and then the diffusion barrier calculated by ab initio is reported, followed by the values used for that diffusion event in the KMC models. The KMC 1 energy barrier is the barrier that is used in theKMC model that includes first neighbours only, while the KMC 2 energy barrier corresponds to the model that also includes second nearest neighbours. The arrows denote the direction of the diffusion event (so← means the barrier corresponds to diffusion from the final configuration to the initial configuration). All barriers are expressed in eV.


Multiscale simulations of the growth of nitrogen-doped graphene on copper

 We used multiscale simulations to model the growth of nitrogen-doped graphene on a copper substrate by chemical vapour deposition (CVD). Our simulations are based on ab-initio calculations of energy barriers for surface diffusion, which are complemented by larger scale kinetic Monte Carlo (KMC) simulations. Our results indicate that the shape of grown doped graphene flakes depends on the temperature and deposition flux they are submitted during the process, but we find no significant effect of nitrogen doping on this shape. However, we show that nitrogen atoms have a preference for pyridine-like sites compared to graphite-like sites, as observed experimentally. [3]


Ndoped

Figure 2. Kinetic Monte-Carlo (kMC) simulation of the first stage of the growth of N-doped graphene on Cu(111) surface. Inset : Morphlogy (ratio of atoms at the edge and within the domain) for different growth conditions.  


[1] T. Chanier, L. Henrard, European Physical Journal B (2015) 88 , 2 , p. 1.

[2] P. Gaillard, T. Chanier, L. Henrard, P. Moskovkin, S. Lucas, Surface Science (2015) 637, 11-18.

[3] P. Gaillard, A. L. Schoenhalz, P. Moskovkin, S. Lucas, L. Henrard, Surface Science (2016) 644, 102-108.