In response to the renewed interest in solute drag and solute trapping models fueled by their applications to additive manufacturing, a novel treatment is proposed to describe the diffusional behaviors of solute at a migrating solid-liquid interface during rapid solidification of multicomponent alloys. While the treatment is still built on irreversible thermodynamics and linear kinetic law, its novelty lies in breaking up the classical trans-interface diffusional flux into two separate fluxes one is the transferred-back flux with its ending point at the interface and the other is the bumping-back flux with its starting point at the interface. This novel treatment entails three significant improvements in reference to the existing models. Firstly, it reveals that the degree of solute drag is dependent on the ratio of liquid diffusive speed over interface diffusive speed. Secondly, a novel relation between the distribution coefficient and interface velocity is derived. It amends the confusing behavior seen in Aziz’s without-drag continuous growth model. Thirdly, the proposed treatment eliminates the need of prescribing the degree of solute drag parameter for the kinetic phase diagram calculation. The numerical solution to the proposed model is presented, and it is ready to be used for the kinetic phase diagram calculation.
Chalcogenide phase-change materials (PCM) have been explored in novel nonvolatile memory and neuromorphic computing technologies. Upon fast crystallization process, the conventional PCM undergo a semiconductor-to-semiconductor transition. However, some PCM change from a semiconducting amorphous phase to a metallic crystalline phase with low conductivity (“bad metal”). In this work, we focus on new “bad metal” PCM, namely, AgSnSe2, and carry out multiscale simulations to evaluate its potential for reconfigurable nanophotonic devices. We study the structural features and optical properties of both crystalline and amorphous AgSnSe2 via density functional theory (DFT) calculations and DFT-based ab initio molecular dynamic (AIMD) simulations. Then we use the calculated optical profiles as input parameters for finite difference time domain (FDTD) modeling of waveguide and metasurface devices. Our multiscale simulations predict AgSnSe2 to be a promising candidate for phase-change photonic applications.
Solidification is a critical process in the manufacturing of metals and alloys, with nucleation being the initial stage that determines the resulting microstructure and mechanical properties. Among various nucleation methods, heterogeneous nucleation is particularly effective in controlling the solidified structure and properties. However, the underlying mechanisms and atomic characteristics of heterogeneous nucleation remain a topic of debate. This paper reviews recent advancements and the current state of research on heterogeneous nucleation during the solidification of aluminum alloys. It focuses on three key areas: the methods and mechanisms for influencing heterogeneous nucleation, existing theories on the subject, and recent experimental and modeling studies on the effect of atomic-scale interactions at the solid/liquid interface on nucleation. The paper also addresses the ongoing challenges and future directions, highlighting the importance of atomic-scale experimental characterization, the validity and reliability of atomic-scale simulations, the role of the pre-nucleation layer at the solid/liquid interface, and the impact of solute elements on the formation of the pre-nucleation layer.
Traditional alloy design typically relies on a trial-and-error approach, which is both time-consuming and expensive. Whilst physical metallurgical (PM) models offer some predictive capabilities, their reliability is limited by errors accumulating across space scales. To address this, this study proposes a novel framework that combines PM knowledge graphs (PMKGs) with graph neural networks (GNNs) to predict the tensile properties of quenching and partitioning steels, using genetic algorithms for dual-objective optimization. Compared to traditional artificial intelligence (AI) models, this framework shows significant advantages in predicting ultimate tensile strength (UTS) and total elongation (TEL) with higher accuracy and stability. Notably, the R2 for TEL prediction improved by approximately 15%. Furthermore, this framework successfully balances UTS and TEL, resulting in the design of alloys with superior overall properties. The designed alloys, with a composition of approximately 0.3 wt.% C, 3 wt.% Mn, 1.2 wt.% Si, and minor amounts of Cr and Al, achieve a UTS exceeding 1500 MPa and TEL near 20%, aligning with PM principles and validating the rationality and feasibility of this method. This study offers new insights into applying AI in complex multi-objective alloy design, highlighting the potential of integrating expert knowledge with GNNs.
This study presents ∆τ, a novel descriptor that captures the compositional dependence of phase transformation temperature (Ap) in NiTi-based shape memory alloys (SMAs). Designed to address the complexity of multicomponent SMAs, ∆τ was integrated into symbolic regression (SR) and kernel ridge regression (KRR) models, yielding substantial improvements in predicting key functional properties: transformation temperature, enthalpy, and thermal hysteresis. Using the KRR model with ∆τ, we explored the NiTiHfZrCu compositional space, identifying six promising alloys with high Ap (>250℃), large enthalpy (>27 J/g), and low thermal hysteresis. Experimental validation confirmed the model's accuracy with the alloys showing high-temperature transformation behavior and low hysteresis, suitable for high-performance applications in aerospace and nuclear industries. These findings underscore the power of domain-informed descriptors like ∆τ in enhancing machine learning-driven materials design.
Graphene-metal (G-M) composites have attracted tremendous interests due to their promising applications in electronics, optics, energy-storage devices and nano-electromechanical systems. Especially, phase formations of graphene combined with different metals are considered valuable for discovering and designing advanced G-M composites. However, the phase formations in G-M systems have rarely been systematically described since graphene was first extracted from graphite in 2004. Here, we propose a data-driven approach to predict the phase formations in G-M systems leveraging G-M binary phase diagrams, which were established using the calculation of phase diagrams method. Phase relationships obtained from G-M phase diagrams of 34 systems and formation enthalpies of corresponding carbides were employed as the training dataset in a machine learning model to further predict the phase formations in additional 13 G-M systems. Phase formation predictions achieved an accuracy of 87.5% in the test dataset. Three distinct phase formations were characterised in G-M systems. Finally, we propose a general phase formation rule in the G-M systems: metals with smaller atomic numbers in the same period are more likely to form secondary solutions with graphene.
Recently, density functional theory (DFT) has been a powerful tool to model the corrosion behaviors of materials, provide insights into the corrosion mechanisms, predict the corrosion performance of materials, and design the corrosion-resistant alloys and organic inhibitors. DFT enables corrosion scientist to fundamentally understand the corrosion behaviors and corrosion mechanisms of materials from the perspective of atomic and electronic structures, combining with the traditional and advanced experimental tests. This review briefly summarizes the main features of DFT calculations and present a comprehensive overview of their typical applications to corrosion and corrosion prevention of metals, involving potential-pH diagrams, hydrogen evolution reaction, anodic dissolution, passivity and passivity breakdown, and organic inhibitor for metals. The paper also reviews the correlations between DFT-computed descriptors and the micro/macro physiochemical parameters of corrosion. Despite the great progress achieved by DFT, there are still some challenges in addressing corrosion issues due to the lack of bridges between the DFT-calculated electronic parameters and the macro corrosion performance of materials. The DFT modeling-experiment-engineering-theory model will be a potential method to clarify and build the links.
The sublimation enthalpy, ΔHsub, is a key thermodynamic parameter governing the phase transformation of a substance between its solid and gas phases. This transformation is at the core of many important materials' purification, deposition, and etching processes. While ΔHsub can be measured experimentally and estimated computationally, these approaches have their own different challenges. Here, we develop a machine learning (ML) approach to rapidly predict ΔHsub from data generated using density functional theory (DFT). We further demonstrate how combining ML and DFT methods with active learning can be efficient in exploring the materials space, expanding the coverage of the computed dataset, and systematically improving the ML predictive model of ΔHsub. With an error of ~15 kJ/mol in instantaneous predictions of ΔHsub, the ML model developed in this work will be useful for the community.
As the lightest structural metal materials, Mg alloys are promising for wider applications but are limited by low strength and poor corrosion resistance. Precipitation is an effective way to improve the strength and other performance of Mg alloys. Facing the extremely complex precipitation process, the crystal structures of precipitates, precipitation sequence, and precipitation thermodynamic and kinetics behaviors have stimulated extensive research interests. Precipitation kinetics, which connects composition, aging processes, and precipitate microstructure, is pivotal in determining the performance of age-hardening Mg alloys. Despite numerous studies on this topic, a comprehensive review remains absent. This work aims to bridge that gap by analyzing precipitation from thermodynamic and kinetic perspectives. Thermodynamically, the stability of precipitates, nucleation driving forces, and resistances of precipitation are discussed. Kinetically, the various kinetic theories including semi-empirical models, mean-field models, phase-field model, and atomistic approaches and their applications in Mg alloys are systematically summarized. Among these, mean-field models emerge as particularly promising for accurately predicting precipitation processes. Finally, the framework for property prediction based on precipitation kinetics is introduced to illustrating the role of integrated computational materials engineering (ICME) in designing advanced Mg alloys.
High-entropy alloys (HEAs) have revolutionized alloy design by integrating multiple principal elements in equimolar or near-equimolar ratios to form solid solutions, vastly expanding the compositional space beyond traditional alloys based on a primary element. However, the immense compositional complexity presents significant challenges in designing alloys with targeted properties, as billions of new alloy systems emerge. High-throughput approaches, which allow the parallel execution of numerous experiments, are essential for accelerated HEA design to navigate this extensive compositional space and fully exploit their potential. Here, we reviewed how advancements in high-throughput synthesis tools have accelerated HEA database development. We also discussed the advantages and limitations of each high-throughput fabrication methodology, as understanding these is vital for achieving precise HEA design.
Large language models (LLMs) excel at extracting information from literatures. However, deploying LLMs necessitates substantial computational resources, and security concerns with online LLMs pose a challenge to their wider applications. Herein, we introduce a method for extracting scientific data from unstructured texts using a local LLM, exemplifying its applications to scientific literatures on the topic of on-surface reactions. By combining prompt engineering and multi-step text preprocessing, we show that the local LLM can effectively extract scientific information, achieving a recall rate of 91% and a precision rate of 70%. Moreover, despite significant differences in model parameter size, the performance of the local LLM is comparable to that of GPT-3.5 turbo (81% recall, 84% precision) and GPT-4o (85% recall, 87% precision). The simplicity, versatility, reduced computational requirements, and enhanced privacy of the local LLM makes it highly promising for data mining, with the potential to accelerate the application and development of LLMs across various fields.
Selecting appropriate material features is essential for effective data-driven materials design. Here, we propose a multi-objective feature optimization strategy that identifies feature subsets to improve both prediction accuracy and active learning efficiency for iterative experimentation. Our approach integrates an evolutionary genetic algorithm to explore an expanded feature space, encompassing both traditional feature pools and a continuous numerical representation of elements rather than relying solely on discrete values. We demonstrate this strategy by identifying high-entropy alloys (HEAs) with optimal strength and ductility. Results show that the optimized feature subsets reduce prediction errors by 20% for strength and 11% for ductility. Additionally, within fewer than three feedback iterations, HEAs with outstanding combinations of yield strength and ductility are identified, highlighting the high efficiency of this approach. This multi-objective feature optimization strategy is adaptable to other material systems, offering a pathway to improve machine learning performance and accelerate materials discovery.
In this study, we propose a novel joint training model for named entity recognition (NER) that combines BERT, BiLSTM, CRF, and a reading comprehension (RC) mechanism. Traditional BERT-BiLSTM-CRF models often struggle with inaccurate boundary detection and excessive fragmentation of named entities due to their lack of specialized vocabulary. Our model addresses these issues by integrating an RC mechanism, which helps refine fragmented results by enabling the model to more precisely identify entity boundaries without relying on an expert-annotated dictionary. Additionally, segmentation issues are further mitigated through a segmented combined voting- and positive-sample-coverage technique. We applied this model to develop a database for mesoporous bioactive glass (MBG). Furthermore, a classifier was developed to automatically detect the presence of pertinent information within paragraphs. For this study, 200 articles were searched using MBG-related keywords, and the data were split into a training set and a test set in a 9:1 ratio. A total of 492 paragraphs were automatically extracted for training, and 50 paragraphs were extracted for testing the model. The results demonstrate that our joint training model achieves an accuracy of 92.8% in named entity recognition, which is 4.3% higher than the 88.5% accuracy of the traditional BERT-BiLSTM-CRF model.
Polymer crystallization, an everlasting subject in polymeric materials, holds great significance not only as a fundamental theoretical issue but also as a pivotal basis for directing polymer processing. Given its multistep, rapid, and thermodynamic nature, tracing and comprehending polymer crystallization pose a formidable challenge, particularly when it encounters practical processing scenarios that involve complex coupled fields (such as temperature, flow, and pressure). The advent of high-time and spatially resolved experiments paves the way for in situ investigations of polymer crystallization. In this review, we delve into the strides in studying polymer crystallization under the effects of coupled external fields via state-of-the-art high-throughput experiments. We highlight the intricate setup of these high-throughput experimental devices, spanning from the laboratory and pilot levels to the industrial level. The individual and combined effects of external fields on polymer crystallization are discussed. By breaking away from the conventional “black box” research approach, special interest is paid to the in situ crystalline behavior of polymers during realistic processing. Finally, we underscore the advancements in polymer crystallization via high-throughput experiments and outline its promising development.