Structural chemistry of amorphous metal-organic frameworks from machine-learning-driven modelling

该项目将专注于开发一种集成的计算和实验方法来表征和理解非晶金属有机框架（mof）的微观结构。计算工作将建立在机器学习（ML）潜在模型的基础上，以较低的成本获得量子精确的模拟。这项工作将通过实验来验证计算预测，然后合成和表征目标非晶mof （aMOFs）。mof是一种被广泛研究的材料类别，在学术研究和工业领域都很受关注。由于其多孔性和相对易于功能化，mof在催化、气体吸附和储存以及气体分离等许多应用中都有很好的应用。aMOFs缺乏晶体对应物的长程顺序，但由于其有前景的应用潜力而引起了人们的兴趣。在阐明非晶化合物的结构特征方面的困难仍然是该领域的一个主要障碍，并阻碍了材料在实践中的应用。通用的基于ml的力场开始被开发用于简单的系统，以相对较低的成本实现良好的材料性能和结构预测。然而，对于复杂材料，ML力场很难开发。因此，需要新的、经过仔细设计和验证的方法。该项目将建立在小组先前的研究基础上，其中使用基于结构简化的方法对复杂材料表达的几何多样性进行定量研究。我们的目标是在计算预测的指导下合成新的、量身定制的aMOFs。项目期间开发的原子间势模型将提供给更广泛的研究界使用。求解非晶固体的结构是具有挑战性的，对于复杂的有机/无机混合固体更是如此。这个项目将开发一种新的方法来应对这一挑战，进一步在原子尺度上理解这些材料。该项目属于EPSRC“计算和理论化学”和“功能陶瓷和无机物”研究领域，主要属于“物理科学”主题的范围。该项目旨在通过提高我们对与实际应用相关的新兴功能材料的理解，与EPSRC关于“依靠基础科学解决社会和经济挑战”这一主题的既定战略保持一致。关于第一个研究领域，该项目显然具有根本性的性质：它将导致新的计算方法的发展和传播，用这些方法来表示复杂的原子结构及其随时间的演变。关于第二个，也许更面向应用的研究领域，该项目与EPSRC的战略一致，具有“基础和高风险”的性质：它将加深对一类功能材料的理解，这些材料现在才开始合成和研究，但具有实质性的应用前景，因为它们与被广泛研究的晶体类似。

This project will focus on developing an integrated computational and experimental approach to characterise and understand the microscopic structure of amorphous metal-organic frameworks (MOFs). The computational work will be built on machine learning (ML) potential models to obtain quantum-accurate simulations at orders of magnitude lower cost. The work will be complemented by experiments to validate the computational predictions, and later, to synthesise and characterise target amorphous MOFs (aMOFs). MOFs are a widely studied materials class, of interest in both academic research and industry. Owing to their porosity and relative ease of functionalisation, MOFs are excellent for many applications, including catalysis, gas adsorption and storage, and gas separation. aMOFs lack the long-range order of their crystalline counterparts, but are of interest owing to their promising application potential. The difficulty in elucidating the structural characteristics of amorphous compounds remains a major barrier in the field, and hinders the use of the materials in practice. General-purpose ML-based force fields are beginning to be developed for simple systems, achieving good material property and structure predictions at relatively low costs. However, ML force fields are difficult to develop for complex materials. New, carefully designed and validated methodologies are therefore required. This project will build upon previous research carried out in the group, in which geometric diversity expressed by complex materials was investigated quantitatively, using methods based on structural simplification. We are aiming to synthesise new, tailor-made aMOFs, as guided by computational predictions. Interatomic potential models developed during the project will be made available for use in the wider research community. Solving the structures of amorphous solids is challenging, and even more so for complex hybrid organic/inorganic solids. This project will develop a new approach to tackling this challenge, to further the atomic-scale understanding of these materials. This project falls within the EPSRC "Computational and theoretical chemistry" and "Functional ceramics and inorganics" research areas, and is primarily within the remit of the "Physical sciences" theme. The project aims align with EPSRC's stated strategy for this theme of meeting "societal and economic challenges that rely on fundamental science for solutions" by improving our understanding of an emerging class of functional materials relevant to practical applications. Regarding the first research area, the project is clearly of a fundamental nature: it will lead to the development and dissemination of new computational methodology with which to represent complex atomistic structures and their evolution over time. Regarding the second, perhaps more application-oriented research area, the project is of a "fundamental and high-risk" nature in line with EPSRC strategy: it will develop a deepened understanding of a class of functional materials that are only now beginning to be synthesised and studied, but hold substantial promise for applications, because of the analogy to their widely-studied crystalline counterparts.