Search for: All records

The melting point is a fundamental property that is time-consuming to measure or compute, thus hindering high-throughput analyses of melting relations and phase diagrams over large sets of candidate compounds. To address this, we build a machine learning model, trained on a database of ∼10,000 compounds, that can predict the melting temperature in a fraction of a second. The model, made publicly available online, features graph neural network and residual neural network architectures. We demonstrate the model’s usefulness in diverse applications. For the purpose of materials design and discovery, we show that it can quickly discover novel multicomponent materials with high melting points. These predictions are confirmed by density functional theory calculations and experimentally validated. In an application to planetary science and geology, we employ the model to analyze the melting temperatures of ∼4,800 minerals to uncover correlations relevant to the study of mineral evolution. 

Ab initio high-throughput efforts are continuously identifying new intermetallic compounds in a wide range of alloy systems that were previously thought to be well-characterized. While such predictions are likely valid near absolute zero, they carry the risk that such phases become unstable at the higher temperature relevant to typical synthesis conditions. We illustrate how this possibility can be rapidly tested by integrating Calphad modeling into the high-throughput loop. As an example, we investigate the Ni-Re system, in which D019 and D1a phases were predicted as possible intermetallic compounds. We confirm that these phases are indeed stable at practical synthesis temperatures and explain how they could have been overlooked in prior assessments. 

The cluster expansion formalism for alloys is used to construct surrogate models for three refractory high-entropy alloys (NbTiVZr, HfNbTaTiZr, and AlHfNbTaTiZr). These cluster expansion models are then used along with Monte Carlo methods and thermodynamic integration to calculate the configurational entropy of these refractory high-entropy alloys as a function of temperature. Many solid solution alloy design guidelines are based on the ideal entropy of mixing, which increases monotonically with N, the number of elements in the alloy. However, our results show that at low temperatures, the configurational entropy of these materials is largely independent of N, and the assumption described above only holds in the high-temperature limit. This suggests that alloy design guidelines based on the ideal entropy of mixing require further examination. 

While rhenium is an ideal material for rapid thermal cycling applications under high temperatures, such as rocket engine nozzles, its high cost limits its widespread use and prompts an exploration of viable cost-effective substitutes. In prior work, we identified a promising pool of candidate substitute alloys consisting of Mo, Ru, Ta, and W. In this work we demonstrate, based on density functional theory melting temperature calculations, that one of the candidates, Mo_0.292Ru_0.555Ta_0.031W_0.122, exhibits a high melting temperature (around 2626 K), thus supporting its use in high-temperature applications.