Molecular dynamics (MD) is a computational method which simulates the real behavior of materials and calculates physical properties of these materials by simultaneously solving the equations of motion for a system of atoms interacting with a given potential. The molecular basis and computationally intensive nature of the MD technique render it particularly effective for modeling systems of limited size and for examining short time scale phenomena.

It is well known from measurements on thin films that such materials display markedly lower thermal conductivities than their bulk counterparts. The chief advantage of MD over other approaches used to predict thin-film thermal conductivity is that it can easily handle the nonuniformly distributed impurities, pores, and cracks which exist in real thin film materials. The current study uses MD to predict the thermal conductivity of a solid argon thin film in the direction perpendicular to the film plane. An argon-model system is used because the widely-accepted Lennard-Jones 12-6 (LJ) potential matches liquid-phase experimental data fairly well, gives good qualitative agreement with the measured specific heat behavior of nanoparticles, and has a simple, two-body form that is computationally economical. A schematic of the computational configuration is shown in Fig. 1.

Figure 2 displays calculation results for normalized perpendicular thermal conductivity versus dimensionless film thickness at several mean lattice temperatures. A rough thin film thermal conductivity estimate based on Majumdar's equation of phonon radiative transfer (EPRT) is also shown to illustrate the expected trend. The results at each temperature are normalized by the experimental conductivity value at that temperature, and film thickness is normalized by the argon lattice parameter. Three observations can be made from Fig. 2. The first is that, as expected, thermal conductivity at all temperatures increases with film thickness. The unexpected undulation of the T = 0.5 curve is fully contained within the envelope of its error bars. The second is that the conductivities of lower-temperature films constitute a smaller fraction of their corresponding bulk values than identically-sized higher-temperature films. This shows that thin-film size effects are more pronounced at lower temperatures. The third observation is that while the expected behavior is an asymptotic increase of the thin film results toward the bulk value, the thicker films for the T = 0.5 and 0.6 cases appear to have thermal conductivities that exceed their corresponding experimental bulk values. This discrepancy is about 30% for the T = 0.6 case. This modest overprediction is likely caused by the perfect purity of the idealized argon model used in the simulations. Additional simulation results indicate that changing the boundary conditions and the imposed fluxes in the thin film simulations produces no significant change in thermal conductivity.

Future work will be done to apply the MD technique to microscale problems where other experimental and analytical approaches are difficult. Molecular dynamics is especially suited to study the thermophysical properties of doped and nanoporous thin films, buckyballs and buckytubes, quantum wires and dots, and other novel nanoscale materials with complex geometries.

References

Lukes, J. R., D. Li, X. G. Liang, and C. L. Tien, "Molecular Dynamics Studyof Solid Thin-Film Thermal Conductivity," Journal of Heat Transfer 122, 536-543 (2000).

Chou, F. C., J. R. Lukes, X. G. Liang, K. Takahashi, and C. L. Tien,"Molecular Dynamics in Microscale Thermophysical Engineering," Annual Review of Heat Transfer, 10, 141-176, 1999.

Tien, C. L., J. R. Lukes, and F. C. Chou, "Molecular Dynamics Simulationof Thermal Transport in Solids," Microscale Thermophysical Engineering,2,133-137 (1998).

Last updated November 4, 2000.
Please send your comments and suggestions to Jennifer Lukes