Leveraging Language Representation for Material Recommendation, Ranking, and Exploration
Authors:
Jiaxing Qu,
Yuxuan Richard Xie,
Kamil M. Ciesielski,
Claire E. Porter,
Eric S. Toberer,
Elif Ertekin
Abstract:
Data-driven approaches for material discovery and design have been accelerated by emerging efforts in machine learning. However, general representations of crystals to explore the vast material search space remain limited. We introduce a material discovery framework that uses natural language embeddings derived from language models as representations of compositional and structural features. The d…
▽ More
Data-driven approaches for material discovery and design have been accelerated by emerging efforts in machine learning. However, general representations of crystals to explore the vast material search space remain limited. We introduce a material discovery framework that uses natural language embeddings derived from language models as representations of compositional and structural features. The discovery framework consists of a joint scheme that first recalls relevant candidates, and next ranks the candidates based on multiple target properties. The contextual knowledge encoded in language representations conveys information about material properties and structures, enabling both representational similarity analysis for recall, and multi-task learning to share information across related properties. By applying the framework to thermoelectrics, we demonstrate diversified recommendations of prototype structures and identify under-studied high-performance material spaces. The recommended materials are corroborated by first-principles calculations and experiments, revealing novel materials with potential high performance. Our framework provides a task-agnostic means for effective material recommendation and can be applied to various material systems.
△ Less
Submitted 19 May, 2023; v1 submitted 1 May, 2023;
originally announced May 2023.
Origin of ultra-low thermal conductivity in unconventional clathrates: Strong scattering from extremely low-frequency rattling modes
Authors:
Kamil M. Ciesielski,
Brenden R. Ortiz,
Lidia C. Gomes,
Vanessa Meschke,
Jesse M. Adamczyk,
Tara L. Braden,
Dariusz Kaczorowski,
Elif Ertekin,
Eric S. Toberer
Abstract:
Recent discoveries of materials with ultra-low thermal conductivity open a pathway to significant developments in the field of thermoelectricity. Here, we conduct a comparative study of three chemically similar antimonides to establish the root causes of their extraordinarily low thermal conductivity ($0.4-0.6$ Wm$^{-1}$K$^{-1}$ at 525 K). The materials of interest are: the unconventional type-XI…
▽ More
Recent discoveries of materials with ultra-low thermal conductivity open a pathway to significant developments in the field of thermoelectricity. Here, we conduct a comparative study of three chemically similar antimonides to establish the root causes of their extraordinarily low thermal conductivity ($0.4-0.6$ Wm$^{-1}$K$^{-1}$ at 525 K). The materials of interest are: the unconventional type-XI clathrate K$_{58}$Zn$_{122}$Sb$_{207}$, the tunnel compound K$_{6.9}$Zn$_{21}$Sb$_{16}$, and the type-I clathrate K$_8$Zn$_{15.5}$Cu$_{2.5}$Sb$_{28}$ discovered herein. Calculations of the phonon dispersions show that the type-XI compound exhibits localized (i.e., rattling) phonon modes with unusually low frequencies that span the entire acoustic regime. In contrast, rattling in the type-I clathrate is observed only at higher frequencies, and no rattling modes are present in the tunnel structure. Modeling reveals that low-frequency rattling modes profoundly limit the acoustic scattering time; the scattering time of the type-XI clathrate is half that of the type-I clathrate and a quarter of the tunnel compound. For all three materials, the thermal conductivities are additionally suppressed by soft framework bonding that lowers the acoustic group velocities, and structural complexity that leads to diffusonic character of the optical modes. Understanding details of thermal transport in structurally complex materials will be crucial for develo** the next generation of thermoelectrics.
△ Less
Submitted 25 January, 2023;
originally announced January 2023.
