Digital discovery最新文献

Evaluating the performance and robustness of LLMs in materials science Q&A and property predictions† 评估llm在材料科学问答和性能预测中的性能和稳健性

IF 6.2

Digital discovery Pub Date : 2025-05-28 DOI: 10.1039/D5DD00090D

Hongchen Wang, Kangming Li, Scott Ramsay, Yao Fehlis, Edward Kim and Jason Hattrick-Simpers

{"title":"Evaluating the performance and robustness of LLMs in materials science Q&A and property predictions†","authors":"Hongchen Wang, Kangming Li, Scott Ramsay, Yao Fehlis, Edward Kim and Jason Hattrick-Simpers","doi":"10.1039/D5DD00090D","DOIUrl":"https://doi.org/10.1039/D5DD00090D","url":null,"abstract":"Large Language Models (LLMs) have the potential to revolutionize scientific research, yet their robustness and reliability in domain-specific applications remain insufficiently explored. In this study, we evaluate the performance and robustness of LLMs for materials science, focusing on domain-specific question answering and materials property prediction across diverse real-world and adversarial conditions. Three distinct datasets are used in this study: (1) a set of multiple-choice questions from undergraduate-level materials science courses, (2) a dataset including various steel compositions and yield strengths, and (3) a band gap dataset, containing textual descriptions of material crystal structures and band gap values. The performance of LLMs is assessed using various prompting strategies, including zero-shot chain-of-thought, expert prompting, and few-shot in-context learning. The robustness of these models is tested against various forms of “noise”, ranging from realistic disturbances to intentionally adversarial manipulations, to evaluate their resilience and reliability under real-world conditions. Additionally, the study showcases unique phenomena of LLMs during predictive tasks, such as mode collapse behavior when the proximity of prompt examples is altered and performance recovery from train/test mismatch. The findings aim to provide informed skepticism for the broad use of LLMs in materials science and to inspire advancements that enhance their robustness and reliability for practical applications.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1612-1624"},"PeriodicalIF":6.2,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00090d?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264335","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Precursor reaction pathway leading to BiFeO3 formation: insights from text-mining and chemical reaction network analyses† 导致BiFeO3形成的前体反应途径：来自文本挖掘和化学反应网络分析的见解

IF 6.2

Digital discovery Pub Date : 2025-05-27 DOI: 10.1039/D5DD00160A

Viktoriia Baibakova, Kevin Cruse, Michael G. Taylor, Carolin M. Sutter-Fella, Gerbrand Ceder, Anubhav Jain and Samuel M. Blau

{"title":"Precursor reaction pathway leading to BiFeO3 formation: insights from text-mining and chemical reaction network analyses†","authors":"Viktoriia Baibakova, Kevin Cruse, Michael G. Taylor, Carolin M. Sutter-Fella, Gerbrand Ceder, Anubhav Jain and Samuel M. Blau","doi":"10.1039/D5DD00160A","DOIUrl":"https://doi.org/10.1039/D5DD00160A","url":null,"abstract":"BiFeO3 (BFO) is a next-generation non-toxic multiferroic material with applications in sensors, memory devices, and spintronics, where its crystallinity and crystal structure directly influence its functional properties. Designing sol–gel syntheses that result in phase-pure BFO remains a challenge due to the complex interactions between metal complexes in the precursor solution. Here, we combine text-mined data and chemical reaction network (CRN) analysis to obtain novel insight into BFO sol–gel precursor chemistry. We perform text-mining analysis of 340 synthesis recipes with the emphasis on phase-pure BFO and identify trends in the use of precursor materials, including that nitrates are the preferred metal salts, 2-methoxyethanol (2 ME) is the dominant solvent, and adding citric acid as a chelating agent frequently leads to phase-pure BFO. Our CRN analysis reveals that the thermodynamically favored reaction mechanism between bismuth nitrate and 2ME interaction involves partial solvation followed by dimerization, contradicting assumptions in previous literature. We suggest that further oligomerization, facilitated by nitrite ion bridging, is critical for achieving the pure BFO phase.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1602-1611"},"PeriodicalIF":6.2,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00160a?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Self-driving laboratories in Japan 日本的自动驾驶实验室

IF 6.2

Digital discovery Pub Date : 2025-05-27 DOI: 10.1039/D4DD00387J

Naruki Yoshikawa, Yuki Asano, Don N. Futaba, Kanako Harada, Taro Hitosugi, Genki N. Kanda, Shoichi Matsuda, Yuuya Nagata, Keisuke Nagato, Masanobu Naito, Tohru Natsume, Kazunori Nishio, Kanta Ono, Haruka Ozaki, Woosuck Shin, Junichiro Shiomi, Kunihiko Shizume, Koichi Takahashi, Seiji Takeda, Ichiro Takeuchi, Ryo Tamura, Koji Tsuda and Yoshitaka Ushiku

引用次数: 0

Exploring noncollinear magnetic energy landscapes with Bayesian optimization 用贝叶斯优化方法探索非共线磁能景观

IF 6.2

Digital discovery Pub Date : 2025-05-24 DOI: 10.1039/D4DD00402G

Jakob Baumsteiger, Lorenzo Celiberti, Patrick Rinke, Milica Todorović and Cesare Franchini

{"title":"Exploring noncollinear magnetic energy landscapes with Bayesian optimization","authors":"Jakob Baumsteiger, Lorenzo Celiberti, Patrick Rinke, Milica Todorović and Cesare Franchini","doi":"10.1039/D4DD00402G","DOIUrl":"https://doi.org/10.1039/D4DD00402G","url":null,"abstract":"The investigation of magnetic energy landscapes and the search for ground states of magnetic materials using ab initio methods like density functional theory (DFT) is a challenging task. Complex interactions, such as superexchange and spin–orbit coupling, make these calculations computationally expensive and often lead to non-trivial energy landscapes. Consequently, a comprehensive and systematic investigation of large magnetic configuration spaces is often impractical. We approach this problem by utilizing Bayesian optimization, an active machine learning scheme that has proven to be efficient in modeling unknown functions and finding global minima. Using this approach we can obtain the magnetic contribution to the energy as a function of one or more spin canting angles with relatively small numbers of DFT calculations. To assess the capabilities and the efficiency of the approach we investigate the noncollinear magnetic energy landscapes of selected materials containing 3d, 5d and 5f magnetic ions: Ba3MnNb2O9, LaMn2Si2, β-MnO2, Sr2IrO4, UO2, Ba2NaOsO6 and kagome RhMn3. By comparing our results to previous ab initio studies that followed more conventional approaches, we observe significant improvements in efficiency.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1639-1650"},"PeriodicalIF":6.2,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d4dd00402g?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reprogramming pretrained language models for protein sequence representation learning† 蛋白质序列表示学习的预训练语言模型重编程[j]

IF 6.2

Digital discovery Pub Date : 2025-05-23 DOI: 10.1039/D4DD00195H

Ria Vinod, Pin-Yu Chen and Payel Das

{"title":"Reprogramming pretrained language models for protein sequence representation learning†","authors":"Ria Vinod, Pin-Yu Chen and Payel Das","doi":"10.1039/D4DD00195H","DOIUrl":"https://doi.org/10.1039/D4DD00195H","url":null,"abstract":"Machine learning-guided solutions for protein learning tasks have made significant headway in recent years. However, success in scientific discovery tasks is limited by the accessibility of well-defined and labeled in-domain data. To tackle the low-data constraint, recent adaptions of deep learning models pretrained on millions of protein sequences have shown promise; however, the construction of such domain-specific large-scale models is computationally expensive. Herein, we propose representation reprogramming via dictionary learning (R2DL), an end-to-end representation learning framework in which we reprogram deep models for alternate-domain tasks that can perform well on protein property prediction with significantly fewer training samples. R2DL reprograms a pretrained English language model to learn the embeddings of protein sequences, by learning a sparse linear mapping between English and protein sequence vocabulary embeddings. Our model can attain better accuracy and significantly improve the data efficiency by up to 104 times over the baselines set by pretrained and standard supervised methods. To this end, we reprogram several recent state-of-the-art pretrained English language classification models (BERT, TinyBERT, T5, and roBERTa) and benchmark on a set of protein physicochemical prediction tasks (secondary structure, stability, homology, and solubility) as well as on a biomedically relevant set of protein function prediction tasks (antimicrobial, toxicity, antibody affinity, and protein–protein interaction).","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1591-1601"},"PeriodicalIF":6.2,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d4dd00195h?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine learning-assisted profiling of a kinked ladder polymer structure using scattering† 利用散射法对一种扭结梯状聚合物结构进行机器学习辅助分析

IF 6.2

Digital discovery Pub Date : 2025-05-21 DOI: 10.1039/D5DD00051C

Lijie Ding, Chi-Huan Tung, Zhiqiang Cao, Zekun Ye, Xiaodan Gu, Yan Xia, Wei-Ren Chen and Changwoo Do

{"title":"Machine learning-assisted profiling of a kinked ladder polymer structure using scattering†","authors":"Lijie Ding, Chi-Huan Tung, Zhiqiang Cao, Zekun Ye, Xiaodan Gu, Yan Xia, Wei-Ren Chen and Changwoo Do","doi":"10.1039/D5DD00051C","DOIUrl":"https://doi.org/10.1039/D5DD00051C","url":null,"abstract":"Ladder polymers consisting of fused rings in the backbone have very limited conformational freedom, which results in very different properties from traditional linear polymers. However, accurately determining their size and chain conformations from solution scattering remains a challenge. Their chain conformations of kinked ladder polymers are largely governed by the structures and relative orientations or configurations of the repeat units, unlike conventional polymer chains whose bending angles between repeat units follow a unimodal Gaussian distribution. Meanwhile, traditional scattering models for polymer chains do not account for these unique structural features. This work introduces a novel approach that integrates machine learning with Monte Carlo simulations to construct a model that can describe the geometry of a type of kinked CANAL ladder polymers. We first develop a Monte Carlo simulation model for sampling the configuration space of CANAL ladder polymers, where each repeat unit is modeled as a biaxial segment. Then, we establish a machine learning-assisted scattering analysis framework based on Gaussian Process Regression. Finally, we conduct small-angle neutron scattering experiments on a CANAL ladder polymer solution to apply our approach. Our method uncovers structural features of such ladder polymers that conventional methods fail to capture.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1570-1577"},"PeriodicalIF":6.2,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00051c?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264328","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Large language models for material property predictions: elastic constant tensor prediction and materials design† 材料性能预测的大型语言模型：弹性常数张量预测和材料设计

IF 6.2

Digital discovery Pub Date : 2025-05-20 DOI: 10.1039/D5DD00061K

Siyu Liu, Tongqi Wen, Beilin Ye, Zhuoyuan Li, Han Liu, Yang Ren and David J. Srolovitz

{"title":"Large language models for material property predictions: elastic constant tensor prediction and materials design†","authors":"Siyu Liu, Tongqi Wen, Beilin Ye, Zhuoyuan Li, Han Liu, Yang Ren and David J. Srolovitz","doi":"10.1039/D5DD00061K","DOIUrl":"https://doi.org/10.1039/D5DD00061K","url":null,"abstract":"Efficient and accurate prediction of material properties is critical for advancing materials design and applications. Leveraging the rapid progress of large language models (LLMs), we introduce ElaTBot, a domain-specific LLM for predicting elastic constant tensors and enabling materials discovery as a case study. The proposed ElaTBot LLM enables simultaneous prediction of elastic constant tensors, bulk modulus at finite temperatures, and the generation of new materials with targeted properties. Integrating general LLMs (GPT-4o) and Retrieval-Augmented Generation (RAG) further enhances its predictive capabilities. A specialized variant, ElaTBot-DFT, designed for 0 K elastic constant tensor prediction, reduces the prediction errors by 33.1% compared with a domain-specific, materials science LLM (Darwin) trained on the same dataset. This natural language-based approach highlights the broader potential of LLMs for material property predictions and inverse design. Their multitask capabilities lay the foundation for multimodal materials design, enabling more integrated and versatile exploration of material systems.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1625-1638"},"PeriodicalIF":6.2,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00061k?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Composition-property extrapolation for compositionally complex solid solutions based on word embeddings† 基于词嵌入的复合型复杂固溶体的组合属性外推

IF 6.2

Digital discovery Pub Date : 2025-05-19 DOI: 10.1039/D5DD00169B

Lei Zhang, Lars Banko, Wolfgang Schuhmann, Alfred Ludwig and Markus Stricker

{"title":"Composition-property extrapolation for compositionally complex solid solutions based on word embeddings†","authors":"Lei Zhang, Lars Banko, Wolfgang Schuhmann, Alfred Ludwig and Markus Stricker","doi":"10.1039/D5DD00169B","DOIUrl":"https://doi.org/10.1039/D5DD00169B","url":null,"abstract":"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 experimentally measured electrocatalytic performance data from two ternary systems (Ag–Pd–Ru; Ag–Pd–Pt), to predict electrocatalytic performance in the shared quaternary system (Ag–Pd–Pt–Ru). As a starting point, we apply Gaussian Process Regression (GPR) based on composition as the feature, which includes both Ag and Pd, achieving an initial correlation coefficient for the prediction (r) of 0.63 and a determination coefficient (r2) of 0.08. Second, we present a version of the GPR model using word embedding-derived materials vectors as features. Using materials-specific embedding vectors significantly improves the predictions, evident from an improved r2 of 0.65. The third model is based on a ‘standard vector method’ which synthesizes weighted vector representations of material properties as features, then creating a reference vector that results in a very good correlation with the quaternary system's material performance (resulting r of 0.94). Our approach demonstrates that existing experimental data combined with the latent knowledge of word embedding-derived representations of materials can be used effectively for materials discovery where data is typically scarce.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1578-1590"},"PeriodicalIF":6.2,"publicationDate":"2025-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00169b?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Artificial intelligence approaches for anti-addiction drug discovery 抗成瘾药物发现的人工智能方法。

IF 6.2

Digital discovery Pub Date : 2025-05-13 DOI: 10.1039/D5DD00032G

Dong Chen, Jian Jiang, Nicole Hayes, Zhe Su and Guo-Wei Wei

引用次数: 0

MOFChecker: a package for validating and correcting metal–organic framework (MOF) structures MOFChecker：一个用于验证和纠正金属有机框架（MOF）结构的软件包。

IF 6.2

Digital discovery Pub Date : 2025-05-08 DOI: 10.1039/D5DD00109A

Xin Jin, Kevin Maik Jablonka, Elias Moubarak, Yutao Li and Berend Smit

{"title":"MOFChecker: a package for validating and correcting metal–organic framework (MOF) structures","authors":"Xin Jin, Kevin Maik Jablonka, Elias Moubarak, Yutao Li and Berend Smit","doi":"10.1039/D5DD00109A","DOIUrl":"10.1039/D5DD00109A","url":null,"abstract":"Metal–organic frameworks are promising porous materials for applications like gas adsorption, separation, transportation, and photocatalysis, but their large-scale computational screening requires high-quality, computation-ready structural data. Existing databases often contain errors due to experimental limitations, including inaccurately determined hydrogen positions, atomic overlaps, and missing components. We introduce MOFChecker to address these issues, providing tools for duplicate detection, geometric and charge error checking, and structure correction. Some errors can be systematically corrected through atomic adjustments on structures in the database, including deleting duplicated structures and adding missing hydrogen atoms, counterions, and linkers. Evaluation of established MOF databases, like the CoRE2014 database, indicates that 38% of structures contain significant errors, highlighting the importance of MOFChecker in ensuring accurate structural data for subsequent density functional theory (DFT) optimizations and computational studies. This work aims to enhance the reliability of MOF databases for high-throughput screening and practical applications.","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 1560-1569"},"PeriodicalIF":6.2,"publicationDate":"2025-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12091091/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144129560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0