Saved in:
Bibliographic Details
Main Authors: Zhang, Lei, Banko, Lars, Schuhmann, Wolfgang, Ludwig, Alfred, Stricker, Markus
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.05466
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913574386925568
author Zhang, Lei
Banko, Lars
Schuhmann, Wolfgang
Ludwig, Alfred
Stricker, Markus
author_facet Zhang, Lei
Banko, Lars
Schuhmann, Wolfgang
Ludwig, Alfred
Stricker, Markus
contents Mastering the challenge of predicting properties of unknown materials with multiple principal elements (high entropy alloys/compositionally complex solid solutions) is crucial for the speedup in materials discovery. We show and discuss three models, using property data from two ternary systems (Ag-Pd-Ru; Ag-Pd-Pt), to predict material performance in the shared quaternary system (Ag-Pd-Pt-Ru). First, we apply Gaussian Process Regression (GPR) based on composition, which includes both Ag and Pd, achieving an initial correlation coefficient for the prediction ($r$) of 0.63 and a determination coefficient ($r^2$) of 0.08. Second, we present a version of the GPR model using word embedding-derived materials vectors as representations. Using materials-specific embedding vectors significantly improves the predictive capability, evident from an improved $r^2$ of 0.65. The third model is based on a `standard vector method' which synthesizes weighted vector representations of material properties, then creating a reference vector that results in a very good correlation with the quaternary system's material performance (resulting $r$ of 0.89). Our approach demonstrates that existing experimental data combined with latent knowledge of word embedding-based representations of materials can be used effectively for materials discovery where data is typically sparse.
format Preprint
id arxiv_https___arxiv_org_abs_2411_05466
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Composition-property extrapolation for compositionally complex solid solutions based on word embeddings
Zhang, Lei
Banko, Lars
Schuhmann, Wolfgang
Ludwig, Alfred
Stricker, Markus
Materials Science
Mastering the challenge of predicting properties of unknown materials with multiple principal elements (high entropy alloys/compositionally complex solid solutions) is crucial for the speedup in materials discovery. We show and discuss three models, using property data from two ternary systems (Ag-Pd-Ru; Ag-Pd-Pt), to predict material performance in the shared quaternary system (Ag-Pd-Pt-Ru). First, we apply Gaussian Process Regression (GPR) based on composition, which includes both Ag and Pd, achieving an initial correlation coefficient for the prediction ($r$) of 0.63 and a determination coefficient ($r^2$) of 0.08. Second, we present a version of the GPR model using word embedding-derived materials vectors as representations. Using materials-specific embedding vectors significantly improves the predictive capability, evident from an improved $r^2$ of 0.65. The third model is based on a `standard vector method' which synthesizes weighted vector representations of material properties, then creating a reference vector that results in a very good correlation with the quaternary system's material performance (resulting $r$ of 0.89). Our approach demonstrates that existing experimental data combined with latent knowledge of word embedding-based representations of materials can be used effectively for materials discovery where data is typically sparse.
title Composition-property extrapolation for compositionally complex solid solutions based on word embeddings
topic Materials Science
url https://arxiv.org/abs/2411.05466