The low lattice thermal conductivity in tin selenide is key to its excellent thermoelectric properties. The dominant mechanism for this phononic property, however, is debated. Past lattice-dynamic calculations for SnSe consistently predicted the existence of a phonon bandgap, which is puzzling as it has not been observed in spectroscopic experiments and is typically associated with low phonon-scattering rates in contrast to the observed physical properties. In this work, we accurately investigate the size of the phonon bandgap in SnSe using a direct comparison between experimental neutron spectroscopy, lattice-dynamic calculations, and molecular dynamics. We theoretically investigate the effects of van der Waals interactions in density functional theory calculations using the popular Grimme D2/D3 and semilocal methods and we also assess the effect of finite-temperature driven anharmonicity using Born-Oppenheimer molecular dynamics. The D2/D3 lattice dynamics calculations, which assume 0 K, predict a phonon bandgap, although the size and structure depends on the details of the van der Waals correction. In contrast, both the experiment and dynamics identify there is a valley in the density of states at finite temperature, ascribed to strong anharmonicity that shifts specific phonon modes. This finding highlights the need for great care when applying 0 K lattice dynamics and dispersion corrections to SnSe, with general implications for designing thermoelectric materials via computational methods.
The “phonon-glass electron-crystal” is an infamously challenging thermoelectric material design principle due to the interconnectedness of thermal and electronic transport in materials. Here, the incorporation of ∼5-nm particles of diamond—a phonon crystal—into the thermoelectric matrix of SnTe is explored as a route toward low lattice thermal conductivity. This counterintuitive strategy works because the large acoustic property mismatch at the SnTe–diamond interface blocks thermal transport. Between 300 and 773 K, SnTe with 1.0 vol. % nanodiamond inclusion exhibits the lowest average and absolute lattice thermal conductivities of any reported SnTe material in this temperature range. The ultra-low lattice thermal conductivity of the nanocomposites is investigated in the two-channel framework—recently advanced in the context of glassy and disordered materials—whereby heat is transported by propagating and non-propagating phonons termed propagons and diffusons, respectively. Above ∼650 K, calculations demonstrate the breakdown of the phonon gas (propagon-only) model for describing the nanocomposite conductivity. At ∼773 K, conductivity reaches the glassy limit where thermal transport is mediated by diffusons. Neutron spectroscopy reveals that with the increase in temperature, phonon modes in SnTe broaden and overlap in energy. We propose that linewidth broadening from nanodiamond-induced scattering and Umklapp processes promotes coupling and wave-like tunneling between overlapping modes, thereby enhancing diffuson-mediated transport at the expense of propagon transport. This progression toward diffuson-dominated conduction represents a novel transport paradigm in primarily crystalline nanocomposites.
Powder-averaged inelastic neutron spectroscopy was performed in order to compare the phonons of microdiamond, few-layer graphene, and amorphous carbon fibers at room temperature. Both acoustic and optical phonons were observed in the crystalline allotropes. We present average group velocities, phonon densities of states, and heat capacities relevant to intrinsic thermal transport. High-temperature measurements were performed to evaluate whether quasiparticle broadening from phonon–phonon scattering could be detected. The glassy carbon form in the fibers exhibits pronounced spectral broadening and an enhanced low-energy density of states beyond the Debye prediction. Methods for estimating thermal conductivity using inelastic neutron data as input to the phonon-gas model are discussed. We highlight opportunities and challenges for employing higher-resolution methods to directly determine phonon lifetimes.