Search

The lattice thermal conductivity ($\kappa_{\ell}$) is a key materials property in power electronics, thermal barriers, and thermoelectric devices. Identifying a wide pool of compounds with low $\kappa_{\ell}$ is particularly important in the development of materials with high thermoelectric efficiency. The present study contributed to this with a reliable machine learning (ML) model based on a training set consisting of 268 cubic compounds. For those, $\kappa_{\ell}$ was calculated from first principles using the temperature-dependent effective potential (TDEP) method based on forces and phonons calculated by density functional theory (DFT). 238 of these were preselected and used to train an initial ML model employing Gaussian process regression (GPR). The model was then improved with active learning (AL) by selecting the 30 compounds with the highest GPR uncertainty as new members of an expanded training set. This was used to predict $\kappa_{\ell}$ of the 1574 cubic compounds in the \textsc{Materials Project} (MP) database with a validation R2-score of 0.81 and Spearman correlation of 0.93. Out of these, 27 compounds were predicted to have very low values of $\kappa_{\ell}$ ($\leq 1.3$ at 300~K), which was verified by DFT calculations. Some of these have not previously been reported in the literature, suggesting further investigations of their electronic thermoelectric properties.

A finite electronic band gap is a standard filter in high-throughput screening of materials using density functional theory (DFT). However, because of the systematic underestimation of band gaps in standard DFT approximations, a number of compounds may incorrectly be predicted metallic. In a more accurate treatment, such materials may instead appear as low band gap materials and could e.g. have good thermoelectric properties if suitable doping is feasible. To explore this possibility, we performed hybrid functional calculations on 1093 cubic materials listed in the MaterialsProjects database with four atoms in the primitive unit cell, spin-neutral ground state, and a formation energy within 0.3 eV of the convex hull. Out of these materials, we identified eight compounds for which a finite band gap emerges. Evaluating electronic and thermal transport properties of these compounds, we found the compositions MgSc2 Hg and Li2 CaSi to exhibit promising thermoelectric properties. These findings underline the potential of reassessing band gaps and band structures of compounds to indentify additional potential thermoelectric materials., Comment: 6 papes, 4 figures

Low lattice thermal conductivity is essential for high thermoelectric performance of a material. Lattice thermal conductivity is often computed using density functional theory (DFT), typically at a high computational cost. Training machine learning models to predict lattice thermal conductivity could offer an effective procedure to identify low lattice thermal conductivity compounds. However, in doing so, we must face the fact that such compounds can be quite rare and distinct from those in a typical training set. This distinctness can be problematic as standard machine learning methods are inaccurate when predicting properties of compounds with features differing significantly from those in the training set. By computing the lattice thermal conductivity of 122 half-Heusler compounds, using the temperature-dependent effective potential method, we generate a data set to explore this issue. We first show how random forest regression can fail to identify low lattice thermal conductivity compounds with random selection of training data. Next, we show how active selection of training data using feature and principal component analysis can be used to improve model performance and the ability to identify low lattice thermal conductivity compounds. Lastly, we find that active learning without the use of DFT-based features can be viable as a quicker way of selecting samples.

The thermoelectric properties of 54 different group 4 half-Heusler (HH) alloys have been studied from first principles. Electronic transport was studied with density functional theory using hybrid functionals facilitated by the $\mathbf{k} \cdot \mathbf{p}$ method, while the temperature dependent effective potential method was used for the phonon contributions to the figure of merit $ZT$. The phonon thermal conductivity was calculated including anharmonic phonon-phonon, isotope, alloy and grain-boundary scattering. HH alloys have an ${\it XYZ}$ composition and those studied here are in the group 4-9-15 (Ti,Zr,Hf)(Co,Rh,Ir)(As,Sb,Bi) and group 4-10-14 (Ti,Zr,Hf)(Ni,Pd,Pt)(Ge,Sn,Pb). The electronic part of the thermal conductivity was found to significantly impact $ZT$ and thus the optimal doping level. Furthermore, the choice of functional was found to significantly affect thermoelectric properties, particularly for structures exhibiting band alignment features. The intrinsic thermal conductivity was significantly reduced when alloy and grain boundary scattering were accounted for, which also reduced the spread in thermal conductivity. It was found that sub-lattice disorder on the ${\it Z}$-site, i.e. the site occupied by group 14 or 15 elements, was more effective than ${\it X}$-site substitution, occupied by group 4 elements. The calculations confirmed that ZrNiSn, ZrCoSb and ZrCoBi based alloys display promising thermoelectric properties. A few other n-type and p-type compounds were also predicted to be potentially excellent thermoelectric materials, given that sufficiently high charge carrier concentrations can be achieved. This study provides insight into the thermoelectric potential of HH alloys and casts light on strategies to optimize thermoelectric performance of multicomponent alloys.

We apply inline electron holography to investigate the electrostatic potential across an individual BaZr0.9Y0.1O3 grain boundary. With holography, we measure a grain boundary potential of -1.3 V. Electron energy loss spectroscopy analyses indicate that barium vacancies at the grain boundary are the main contributors to the potential well in this sample. Furthermore, geometric phase analysis and density functional theory calculations suggest that reduced atomic density at the grain boundary also contributes to the experimentally measured potential well.

Boron (B) has the potential for generating an intermediate band in cubic silicon carbide (3C-SiC), turning this material into a highly efficient absorber for single-junction solar cells. The formation of a delocalized band demands high concentration of the foreign element, but the precipitation behavior of B in the 3C polymorph of SiC is not well known. Here, probe-corrected scanning transmission electron microscopy and secondary-ion mass spectrometry are used to investigate precipitation mechanisms in B-implanted 3C-SiC as a function of temperature. Point-defect clustering was detected after annealing at 1273 K, while stacking faults, B-rich precipitates and dislocation networks developed in the 1573 - 1773 K range. The precipitates adopted the rhombohedral B13C2 structure and trapped B up to 1773 K. Above this temperature, higher solubility reduced precipitation and free B diffused out of the implantation layer. Dopant concentrations E19 at.cm-3 were achieved at 1873 K., Comment: 18 pages, 10 figures

Thermoelectric application of half-Heusler compounds suffers from their fairly high thermal conductivities. Insight into how effective various scattering mechanisms are in reducing the thermal conductivity of fabricated XNiSn compounds (X = Hf, Zr, Ti, and mixtures thereof) is therefore crucial. Here, we show that such insight can be obtained through a concerted theory-experiment comparison of how the lattice thermal conductivity kLat(T) depends on temperature and crystallite size. Comparing theory and experiment for a range of Hf0.5Zr0.5NiSn and ZrNiSn samples reported in the literature and in the present paper revealed that grain boundary scattering plays the most important role in bringing down kLat, in particular so for unmixed compounds. Our concerted analysis approach was corroborated by a good qualitative agreement between the measured and calculated kLat of polycrystalline samples, where the experimental average crystallite size was used as an input parameter for the calculations. The calculations were based on the Boltzmann transport equation and ab initio density functional theory. Our analysis explains the significant variation of reported kLat of nominally identical XNiSn samples and is expected to provide valuable insights into the dominant scattering mechanisms even for other materials.

In spite of their relatively high lattice thermal conductivity $\kappa_{\ell}$, the XNiSn (X=Ti, Zr or Hf) half-Heusler compounds are good thermoelectric materials. Previous studies have shown that $\kappa_{\ell}$ can be reduced by sublattice-alloying on the X-site. To cast light on how the alloy composition affects $\kappa_\ell$, we study this system using the phonon Boltzmann-transport equation within the relaxation time approximation in conjunction with density functional theory.The effect of alloying through mass-disorder scattering is explored using the virtual crystal approximation to screen the entire ternary Ti$_x$Zr$_{y}$Hf$_{1-x-y}$NiSn phase diagram. The lowest lattice thermal conductivity is found for the Ti$_x$Hf$_{1-x}$NiSn compositions; in particular, there is a shallow minimum centered at Ti$_{0.5}$Hf$_{0.5}$NiSn with $\kappa_l$ taking values between 3.2 and 4.1 W/mK when the Ti content varies between 20 and 80\%. Interestingly, the overall behavior of mass-disorder scattering in this system can only be understood from a combination of the nature of the phonon modes and the magnitude of the mass variance. Mass-disorder scattering is not effective at scattering acoustic phonons of low energy. By using a simple model of grain boundary scattering, we find that nanostructuring these compounds can scatter such phonons effectively and thus further reduce the lattice thermal conductivity; for instance, Ti$_{0.5}$Hf$_{0.5}$NiSn with a grain size of $L= 100$ nm experiences a 42\% reduction of $\kappa_{\ell}$ compared to that of the single crystal.

Energy filtering has been suggested by many authors as a means to improve thermoelectric properties. The idea is to filter away low-energy charge carriers in order to increase Seebeck coefficient without compromising electronic conductivity. This concept was investigated in the present paper for a specific material (ZnSb) by a combination of first-principles atomic-scale calculations, Boltzmann transport theory, and experimental studies of the same system. The potential of filtering in this material was first quantified, and it was as an example found that the power factor could be enhanced by an order of magnitude when the filter barrier height was 0.5~eV. Measured values of the Hall carrier concentration in bulk ZnSb were then used to calibrate the transport calculations, and nanostructured ZnSb with average grain size around 70~nm was processed to achieve filtering as suggested previously in the literature. Various scattering mechanisms were employed in the transport calculations and compared with the measured transport properties in nanostructured ZnSb as a function of temperature. Reasonable correspondence between theory and experiment could be achieved when a combination of constant lifetime scattering and energy filtering with a 0.25~eV barrier was employed. However, the difference between bulk and nanostructured samples was not sufficient to justify the introduction of an energy filtering mechanism. The reasons for this and possibilities to achieve filtering were discussed in the paper.

Sodium cobaltate with high Na content is a promising thermoelectric material. It has recently been reported that oxygen vacancies can alter the material properties, reducing its figure of merit. However, experimental data concerning the oxygen stoichiometry are contradictory. We therefore studied the formation of oxygen vacancies in Na_xCoO_2 with first principles calculations, focusing on x = 0.75. We show that a very low oxygen vacancy concentration is expected at the temperatures and partial pressures relevant for applications., Comment: 4 pages

Adsorption of gaseous species, and in particular of hydrogen atoms, on graphene is an important process for the chemistry of this material. At the equilibrium geometry, the H atom is covalently bonded to a carbon that puckers out from the surface plane. Nevertheless the \emph{flat} graphene geometry becomes important when considering the full sticking dynamics. Here we show how GGA-DFT predicts a wrong spin state for this geometry, namely $S_z$=0 for a single H atom on graphene. We show how this is caused by the self-interaction error since the system shows fractional electron occupations in the two bands closest to the Fermi energy. It is demonstrated how the use of hybrid functionals or the GGA+$U$ method an be used to retrieve the correct spin solution although the latter gives an incorrect potential energy curve.

Adsorption of hydrogen atoms on a single graphite sheet (graphene) has been investigated by first-principles electronic structure means, employing plane-wave based, periodic density functional theory. A reasonably large 5x5 surface unit cell has been employed to study single and multiple adsorption of H atoms. Binding and barrier energies for sequential sticking have been computed for a number of configurations involving adsorption on top of carbon atoms. We find that binding energies per atom range from ~0.8 eV to ~1.9 eV, with barriers to sticking in the range 0.0-0.2 eV. In addition, depending on the number and location of adsorbed hydrogen atoms, we find that magnetic structures may form in which spin density localizes on a $\sqrt{3}{x}\sqrt{3}{R}30^{\circ}$ sublattice, and that binding (barrier) energies for sequential adsorption increase (decrease) linearly with the site-integrated magnetization. These results can be rationalized with the help of the valence-bond resonance theory of planar $\pi$ conjugated systems, and suggest that preferential sticking due to barrierless adsorption is limited to formation of hydrogen pairs., Comment: 12 pages, 8 figures and 4 tables

We summarize and discuss some of the available experimental and theoretical data important for understanding the role played by subsurface sites in dissociative chemisorption calculations for the H$_2$/Pd(111) system. Then we use a semi-empirical potential energy surface (PES) to model the interaction of a H$_2$ molecule impinging on a Pd(111) surface. The London-Eyring-Polanyi-Sato (LEPS) construction has been extended to make direct subsurface absorption possible. A 2-dimensional wave packet calculation is used to find qualitative trends in the direct subsurface absorption and to reveal the time scales involved. We suggest that a partial in-plane relaxation occurs for the slowest incoming particles, thus resulting in a higher direct subsurface absorption probability for low energies., Comment: Journal of Chemical Physics (in press), 19 pages, REVTeX, 4 Postscript figures