Reactive force field simulation on thermal conductivities of carbon nanotubes and graphene

Graphene and carbon nanotubes, which are grouped as low carbon allotropes, exhibit unique thermal conductivities. This property has made them attractive for offering solutions in thermal management technologies. Unfortunately, chemical functionalization, interfaces, and junctions are the principle causes of the reduced thermal conductivity when manufacturing carbon nanotubes and graphene based thermal management systems. For example, the measured thermal conductivities of carbon nanotube arrays, graphene metal composites, and graphene nanocomposites, are uncompetitive when compared to thermal conductivities of copper, silver, and aluminum.

This sharp contrast in the thermal conductivities is founded on the strong phonon scattering on interfaces as well as limited volume fraction of fillers. For this reason, appropriate assembly of carbon nanotubes and graphene with the excellent filling ratio and minimum interface resistance is desired for implementing them in heat management technologies. This however calls for more insights into the interaction of carbon nanotubes and graphene with their substrates as well as host materials.

Molecular dynamics simulation has been adopted to analyze the heat transfer processes. However, researchers have reported varying values of thermal conductivities. These discrepancies can be attributed to different force field potentials and algorithms. Varying system sizes can also lead to varying results.

Chenghao Diao, Yuan Dong and Jian Lin at University of Missouri implemented non-equilibrium simulations to investigate the thermal conductivities of graphene and carbon nanotubes by adopting ReaxFF potentials, all of which enable a wider choice of element category. Their research work is published in peer-reviewed journal, International Journal of Heat and Mass Transfer.

The authors used in the simulations, carbon nanotubes and graphene with maximum lengths of 400nm. They investigated the length-dependent thermal conductivities in the systems. The ReaxFF potentials predicted higher thermal conductivities as well as longer phonon mean free paths as compared to the conventional AIREBO potential. When the system size was 50nm, the ReaxFF potentials were smaller than the AIREBO potential.

Thermal conductivities obtained from the Reax-03 and Reax-12 indicated a linear relationship as the system sized up to approximately 400nm. This was indicative that the phonon mean free paths of carbon nanotubes and graphene were longer than 400nm. This was well in agreement with the experiments as opposed to the values from AIREBO potential. The impact of potentials on phonon attributes of graphene were in addition investigated by computing the phonon densities of states.

Reax-03 as well as Reax-12 potentials offered reasonable locations as well as shapes of G peaks, while the Reax-15 indicated a broader band. The low frequency bands of out-of-plane mode were wider for the Reax-03 and Reax-12 potentials. This was an indication of longer phonon mean free paths for the two potentials. The authors found that the Reax-03 and the Reax-12 potentials predicted precise out-of-plane phonon modes compared with experimental results.

The whole idea of this study is not to replace the conventional potentials in solving problems relating to pure carbon nanotubes and graphene. The idea is to implement successfully ReaxFF potentials, which involve a larger variety of elements.