Discovery of Low-Modulus Ti-Nb-Zr Alloys Based on Machine Learning and First-Principles Calculations

Abstract

The discovery of low-modulus Ti alloys for biomedical applications is challenging due to a vast number of compositions and available solute contents. In this work, machine learning (ML) methods are employed for the prediction of the bulk modulus (K) and the shear modulus (G) of optimized ternary alloys. As a starting point, the elasticity data of more than 1800 compounds from the Materials Project fed linear models, random forest regressors, and artificial neural networks (NN), with the aims of training predictive models for K and G based on compositional features. The models were then used to predict the resultant Young modulus (E) for all possible compositions in the Ti-Nb-Zr system, with variations in the composition of 2 at. %. Random forest (RF) predictions of E deviate from the NN predictions by less than 4 GPa, which is within the expected variance from the ML training phase. RF regressors seem to generate the most reliable models, given the selected target variables and descriptors. Optimal compositions identified by the ML models were later investigated with the aid of special quasi-random structures (SQSs) and density functional theory (DFT). According to a combined analysis, alloys with 22 Zr (at. %) are promising structural materials to the biomedical field, given their low elastic modulus and elevated beta-phase stability. In alloys with Nb content higher than 14.8 (at. %), the beta phase has lower energy than omega, which may be enough to avoid the formation of omega, a high-modulus phase, during manufacturing.

A Deep Dive into Machine Learning Density Functional Theory for Materials Science and Chemistry

“…The design and discovery of materials with specified structural properties constitutes another important research thrust where DFT data and property mappings in terms of ML play a central role. These investigations [272], [273], [274], [275], [276], [277], [278], [279], [280], [281], [282], [283] are similar in nature to the ones discussed in Sec. III A 2, however, with a stronger emphasis on property selection.…”

“…NNs and RFRs were used in Ref. [272] to predict bulk, shear, and Young's modulus for ternary Ti-Nb-Zr alloys in order to accelerate an otherwise costly high-throughput DFT search for low-modulus alloys, similar to Ref. [231].…”

Multiscale numerical simulation of in-plane mechanical properties of two-dimensional monolayers

“…Construction is done through both traditional learning algorithms (such as support vector machine, decision tree, kernel ridge regression, Gaussian mixture regression) and deep learning algorithms to identify the relationship between the atomic ngerprints and their inplane mechanical properties. 165,166…”

A compilation of experimental data on the mechanical properties and microstructural features of Ti-alloys

“…As for materials discovery, a comprehensive compilation of mechanical properties and microstructure data may allow researchers to spot unexplored regions of the vast available compositional space. Computational Materials Science works have achieved this in the past 21 ; however, due to the lack of organized experimental data, modeling and predictions were primarily performed using theoretical data obtained from first-principles calculations 22 . The public availability of a routinely updated database will allow experimental scientists to act more proactively concerning alloy exploration 23 .…”