On the effectiveness of the thermoelectric energy filtering mechanism in low-dimensional superlattices and nano-composites

Abstract

Electron energy filtering has been suggested as a promising way to improve the power factor and enhance the ZT figure of merit of thermoelectric materials. In this work we explore the effect that reduced dimensionality has on the success of the energyfiltering mechanism for power factor enhancement. We use the quantum mechanical nonequilibrium Green's function (NEGF) method for electron transport including electronphonon scattering to explore 1D and 2D superlattice / nanocomposite systems. We find that, given identical material parameters, 1D channels utilize energy filtering more effectively than 2D as they: i) allow one to achieve maximal power factor for smaller well sizes / smaller grains (which is needed to maximize phonon scattering), ii) take better advantage of a lower thermal conductivity in the barrier/boundary materials compared to the well/grain materials in both: enhancing the Seebeck coefficient; and in producing a system which is robust against detrimental random deviations from optimal barrier design. In certain cases we find that the relative advantage can be as high as a factor of 3. We determine that energy-filtering is most effective when the average energy of carrier flow varies the most in the wells and the barriers along the channel, an event which appears when the energy of the carrier flow in the host material is low and when the energy relaxation mean-free-path of carriers is short. Although the ultimate reason these aspects, which cause a 1D system to see greater relative improvement than a 2D, is the 1D system's van Hove singularity in the density-of-states, the insights obtained are general and inform energy-filtering design beyond dimensional considerations.

Nanostructured potential well/barrier engineering for realizing unprecedentedly large thermoelectric power factors

“…Energy relaxation in semiconductors is dominated by inelastic scattering processes, primarily electron-optical phonon scattering. In Si, for example, the electron-optical phonon meanfree-path is around λE ~15 nm, which results in wells sizes of LW ~ 50 nm to exhibit semirelaxation of the current energy [27,64]. Well sizes of the order of LW = 30 nm will only exhibit some degree of relaxation as we show below.…”

“…Thus, an optimal compromise is achieved under semi-relaxation conditions. The advantage of energy semi-relaxation to the Seebeck coefficient is described by us and others in several works [27,26,62,63,64]. 4d shows the mobility of p-type Si (dashed line), and the mobility of the carriers that travel above VB alone, which in this case is more than double (solid line).…”

“…Interestingly, from Fig. 7b we can extract that to achieve 5×10 19 [26,27,64]. Energy relaxation in semiconductors is dominated by inelastic scattering processes, primarily electron-optical phonon scattering.…”

“…8a). For this, we have used our in-house 2D simulator that we have developed with details explained in Refs [26,64,66]…”

Enhanced thermoelectric performance actuated by inelastic processes in the channel region

“…It is worth mentioning that, unlike the given Refs. [23][24][25], where much of the enhancement in generated power stems from an enhanced electrical conductivity σ( ) at higher energy, the proposed concept in this paper is independent of such constraints. Rather, in this case, I aim towards finding out the optimal bounded transmission functions for maximizing the efficiency at a generated power.…”