Abstract
We demonstrate a new mechanism in the early stages of sub-monolayer epitaxial island growth, using Monte Carlo simulations motivated by experimental observations on the growth of graphene on copper foil. In our model, the substrate is "dynamically rough", by which we mean (i) the interaction strength between Cu and C varies randomly from site to site, and (ii) these variable strengths themselves migrate from site to site. The dynamic roughness provides a simple representation of the near-molten state of the Cu substrate in the case of real graphene growth. Counterintuitively, the graphene island size increases when dynamic roughness is included, compared to a static and smooth substrate. We attribute this effect to destabilisation of small graphene islands by fluctuations in the substrate, allowing them to break up and join larger islands which are more stable against roughness. In the case of static roughness, when process (ii) is switched off, island growth is strongly inhibited and the scale-free behaviour of island size distributions, present in the smooth-static and rough-dynamic cases, is destroyed. The effects of the dynamic substrate roughness cannot be mimicked by parameter changes in the static cases.
Highlights
While recent density functional tight binding (DFTB) simulations[14] have probed the early stages of graphene nucleation on semi-molten copper, these cannot access the wider range of time and length scales over which important processes occur.[15,16]
We find that dynamic substrate disorder enhances the growth of large, regular islands, and preserves the scaling of the island size distribution (ISD) over a wide range of coverages
In this paper we have presented an abstract lattice Monte Carlo (MC) model for surface monolayer growth
Summary
While recent density functional tight binding (DFTB) simulations[14] have probed the early stages of graphene nucleation on semi-molten copper, these cannot access the wider range of time and length scales over which important processes occur.[15,16] These range from atomistic events on a timescale around 10À12 s to the scale of hundreds of microns and minutes for graphene grain completion. The key ingredient of a MC model is the list of microscopic events which can occur and their rates or probabilities. Such models are constructed on a static lattice: monomers (atoms) can occupy discrete sites, which are identical in the substrate, so that only occupancy by monomers in the dynamic growing layer differentiates the sites. This type of model does not seem to be appropriate for a growth system where the substrate is highly active during growth, such as copper in graphene CVD
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
