**7. Conclusion**

96 Molecular Dynamics – Theoretical Developments and Applications in Nanotechnology and Energy

height of 2 monolayers. The coverage factor of 1% exhibits the minimum in thermal conductivity as the smooth interfaces. For coverage factor of 10% and 50% the minimum dissapears and the results confirm previous theoretical observations (Daly et al 2002, Imamura et al 2003), while the two defect concentrations do not influence the thermal conductivity. For height of roughness of 2MLs, the thermal conductivity is higher than for smooth interfaces and exhibits a maximum for a 80. In the same figure at the right the influence of interdiffusion of the two species is shown, with similar behavior as the interfaces of height of two MLs. These unexpected results are related with the fact that inelastic scattering could enhance the thermal conductivity through interfaces (Lepi et al 2003). The study shows that the thermal conductivity depends strongly on the detailed description of the interfaces, including height, shape of roughness, interdiffusion of species etc. Changing the structure of the interface can favor or deteriorate the thermal transport through interfaces, and we showed that the interface structure is a relevant parameter which

Fig. 10. Cross-plane thermal conductivity as a function of the superlattice period with NEMD for GaAs/AlAs systems. Left: smooth interfaces, rough interfaces of height of 1ML with three different concentrations and rough interfaces of height of 2ML are presented. Right: rough interfaces with interdiffusion. The smooth and rough interfaces with a

concentration of 50% are also shown for comparison.

controls the thermal properties of superlattices.

We hope we have helped in showing the possibilities of the molecular dynamics technique to probe heat transport in solids, and in particular nanomaterials. Molecular dynamics is a relatively simple and flexible method to be used especially today when stable optimized open source codes have become avalaible: LAMMPS, DLPOLY, GROMACS to name a few. The ever increasing number of publications has helped in resolving controversies regarding heat transfer at the nanoscale, and also getting physical insights in classical problems. A physical understanding of energy transport across two solids-a very simple question-still poses a challenge ! MD simulations may help in observing the scattering of phonons at interfaces, which certainly complements experimental investigations. In the context of nanomaterials, MD is well adapted to characterize ballistic heat transport in nano-objects, although care should be taken not to introduce spurious sources of scattering ! Let us hope that further experimental measurements may improve the modeling of the nanomaterials that we have considered.
