聚合物
班级(哲学)
计算机科学
财产(哲学)
材料科学
共聚物
材料信息学
化学空间
集合(抽象数据类型)
可扩展性
纳米技术
人工智能
化学
医学
哲学
生物化学
护理部
认识论
数据库
公共卫生
复合材料
健康信息学
药物发现
程序设计语言
工程信息学
作者
Shivank S. Shukla,Christopher Kuenneth,Rampi Ramprasad
出处
期刊:Mrs Bulletin
[Springer Nature]
日期:2023-07-31
标识
DOI:10.1557/s43577-023-00561-0
摘要
Polymers are diverse and versatile materials that have met a wide range of material application demands. They come in several flavors and architectures (e.g., homopolymers, copolymers, polymer blends, and polymers with additives). Searching this enormous space for suitable materials with a specific set of property/performance targets is thus nontrivial, painstaking, and expensive. Such a search process can be made effective by the creation of rapid and accurate property predictors. In this article, we present a machine learning framework to predict the thermal properties of homopolymers, copolymers, and polymer blends. A universal fingerprinting scheme capable of handling this entire polymer chemical class has been developed and a multitask deep learning algorithm is trained simultaneously on a large data set of glass-transition, melting, and degradation temperatures. The trained models demonstrate precision and scalability to other properties when relevant data becomes accessible. The chemical and structural variations that can be achieved with the polymers is staggering. Such extraordinary and diverse possibilities translate to attractive combinations of physical properties impacting several application spaces, making the polymeric class of materials ubiquitous in our modern society. This chemistry–structure–property diversity is accompanied by a major challenge. Searching the chemo-structural space to identify suitable application-relevant candidates with the right set of target properties is nontrivial, requiring advanced rapid property prediction and search schemes. In this article, we present a data-driven machine learning framework to instantaneously predict the thermal properties (an important property class) of a dizzyingly large class of polymer archetypes, namely, homopolymers, copolymers, and polymer blends. A state-of-the-art machine learning algorithm has been developed and trained on a large data set of glass-transition, melting, and degradation temperatures, to make instantaneous predictions of these properties for any new-to-the-world polymer that falls in this large important polymer chemical class. This prediction scheme paves the way for discovering polymers with unprecedented thermal stability by allowing searches of enormous chemical spaces at scale.
科研通智能强力驱动
Strongly Powered by AbleSci AI