Thermal boundary resistance dominates the overall resistance of nanosystems. This effect can be utilized to improve the figure of merit of thermoelectric materials. It is also a concern for thermal failures in microelectronic devices. The interfacial resistance sensitively depends on many interrelated structural details including material properties of the two layers, the system dimensions, the interfacial morphology, and the defect concentrations near the interface. The lack of an analytical understanding of these dependences has been a major hurdle for a science-based design of optimum systems on a nanoscale. Here we have combined an analytical model with extensive, highly-converged direct method molecular dynamics simulations to derive analytical relationships between interfacial thermal boundary resistance and structural features. We discover that thermal boundary resistance linearly decreases with total interfacial area that can be modified by interfacial roughening. This finding is further elucidated using wave packet analysis and local density of state calculations.

In nanosystems, the thermal resistance between materials typically dominates the overall resistance. While size effects on thermal conductivity are well documented, size effects on thermal boundary conductance have only been speculated. In response, we characterize the relationship between interfacial resistance and material dimension using molecular dynamics. We find that the interfacial resistance increases linearly with inverse system length but is insensitive to cross-sectional area. Also, from the temperature-dependence of interfacial resistance, we conclude that contributions of short-wavelength phonons dominate. Lastly, by coupling the molecular dynamics to a two-temperature model, we show that electron-mediated transport has little effect on thermal resistance.

No abstract

The multilayer-relaxation geometry of a tungsten (111)surface has been calculated using both a firstprinciples approach within the local-density approximation and an empirical approach using an embedded-atom-type potential with angular forces. Both calculations predict the same relaxation pattern of a triplet of W layers moving toward each other and an expansion of the layer spacing between each triplet. The first-principles calculations were carried out for three-, five-, and seven-layer thin films using mixed-basis pseudopotential techniques and including scalar-relativistic interactions. Within these approximations, the electronic structure of the W(111) surface is characterized by a surface resonance near the Fermi level and near the I point of the surface Brillouin zone, which is insensitive to surface relaxation. The empirical calculations were carried out for 3to 15-layer thin films. The relaxation geometries calculated for the three-, five-, and seven-layer films are consistent with the first-principles results; geometries calculated for the larger films indicate that the main relaxation effects occur in the first four layers near the surface, although measurable relaxations occur far from the surface.

We investigate the influence of crystallographic orientation and anisotropy on local phonon density of states, phonon transmissivity, and Kapitza conductance at interfaces between Lennard-Jones solids via classical molecular dynamics simulations. In agreement with prior works, we find that the Kapitza conductance at an interface between two face-centered cubic materials is independent of crystallographic orientation. On the other hand, at an interface between a face-centered cubic material and a tetragonal material, the Kapitza conductance is strongly dependent on the relative orientation of the tetragonal material, albeit this dependence is subject to the overlap in vibrational spectra of the cubic and tetragonal materials. Furthermore, we show that interactions between acoustic phonons in the cubic material and optical phonons in the tetragonal material can lead to the interface exhibiting greater “thermal anisotropy” as compared to that of the constituent materials. Finally, it is noted that the relative match or mismatch between the Debye temperatures of two materials comprising an interface does not serve an accurate gauge of the efficiency of interfacial thermal transport when those materials have different crystal structures.

We present a critical investigation of the validity of the harmonic approximation for developing constitutive models for multiscale simulations. We examine models using the Cauchy–Born hypothesis within the quasiharmonic, local harmonic, and modified local harmonic approximations in order to characterize the strain and temperature dependence of the Cauchy stress for uniaxial, equibiaxial, and equitriaxial deformations. We compare these predictions with molecular dynamics simulations to evaluate the suitability of each harmonic model over a wide range of strains and temperatures. The various harmonic approximations are found to be very robust over a large temperature range. All the approximations make very similar predictions at small strains and temperatures. At larger strains and temperatures, the quasiharmonic model is the most accurate but also the most computationally expensive. The modified local harmonic model is seen to provide an accurate alternative to the full quasiharmonic model over a wide range of strains while being much less computationally expensive. The local harmonic model is similar in absolute accuracy to the modified local harmonic model, but the modified harmonic model is seen to more accurately predict the elastic moduli.

This paper describes “digital origami” from geometrically frustrated tiles: arrays of structures that cannot attain an energetically favorable flat state because of internal constraints. Each tile can typically snap between two symmetric energy-minimizing states, and neighboring tiles are coupled so that a collection of binary tile states determines the local curvature of the entire sheet. Modular structures like these tiles give great advantages in manufacturing and in predictive simulation, and their discrete nature is a good match for digital readout and control of self-folding systems. The digital origami concept applies to materials from the nanoscale to the macroscale. An example from previous researchers is a metal sheet with an array of dimples that can flex up or down. In this paper we investigate more general techniques that can develop planar sheets into bistable structures. Such methods include installing compressed pieces into a flat sheet of material, or tying together parts of a sheet (smocking). These methods work with a large variety of technologically important materials including circuit boards and semiconductor substrates. While there are clear benefits to such structures, significant obstacles to design exist in manufacturing, in evaluating their mechanical properties, and in choosing the best arrangement of tile states to match a desired shape. Determining the optimal flipping order of tile states to change the sheet from one shape to another is a sequencing problem analogous to protein folding, and origami from non-planar surfaces is a little-explored area in the fine arts. The paper discusses algorithms for curve-matching with one-dimensional arrays by error diffusion, and shape prediction for two-dimensional sheets with pre-programmed tile states. Low computational cost is required for creating structures that can predict, detect, and even change their own three-dimensional shapes using low-power onboard microprocessors. Motivators for this challenge include shape measurement over a large size range — for example, detecting the changing shapes of biological microstructures or endowing robots with a spatial sense similar to human proprioception — and self-modeling of structural properties for lightweight morphing structures that can strengthen for impact in a given direction using a limited amount of material.

No abstract