DMREF: Computational Discovery of Polymeric Membranes for Dehydration of Polar Solvents

膜分离已经彻底改变了一些行业（如海水淡化），并有可能推动许多应用朝着更节能和可持续的过程发展。化学分离通常是高能耗的，占美国每年一次能源消耗的5%以上，膜分离方法在降低能源利用率和资本成本方面具有巨大的潜力。然而，用于分离混合溶剂的膜虽然在实践中很重要，但在技术上具有挑战性，因为需要区分具有细微性质差异的小分子的运输。考虑到巨大的化学和设计空间，合理设计用于这种分离的下一代膜是一个重大挑战，但可以通过材料基因组计划（MGI）启发的筛选来实现。也就是说，这种mgi风格的努力可能会改变目前长期存在的膜开发模式。在这项工作中，功能和性能驱动的筛选，与模拟和实验之间的紧密耦合，将导致设计和制造为目标分离量身定制的高性能膜。具体地说，通过渗透蒸发的极性溶剂脱水将作为首要的初始目标。这是因为有效的新材料的发现和部署将消除对高成本和高能量分离的需求，并实现有效的溶剂再利用，以实现可持续制造。技术概述：这项工作将开发一个集成的，受mgi启发的计算和实验筛选平台，旨在加速极性溶剂脱水膜的合理设计。这将通过使用分子模拟和设计框架（MoSDeF）的广泛分子模拟相结合来实现；使用DeepForge进行机器学习；开环复分解聚合化学与自旋涂层相结合的合成实验表征;在渗透汽化过程中。这种从分子到过程的方法将使从一个共同的中心支架合成各种聚合物膜组合物成为可能。此外，在分子水平筛选模拟和实验之间将有一个整体的协同作用，以分子设计和识别适合特定脱水分离的新膜。该项目的目标是鉴定、合成、表征和测试用于极性有机溶剂-水膜分离的新候选聚合物，开发一个强大的聚合物膜特性库，开发将化学与膜膜测量特性相关联的机器学习模型，以及发布一套普遍适用的软件工具，以实现软物质系统的快速筛选和机器学习研究。此外，这项工作将计算建模、机器学习、材料合成、表征和目标分离的性能评估集成在一起，将为参与的研究生和博士后研究人员提供一个极好的教育平台，让他们体验MGI愿景中描述的全套互联组件。通过培养实验、计算和数据科学这三个基本支柱的能力，该项目将培养一支与MGI模型相一致的劳动力队伍。此外，在本科和K-12阶段的多种综合教育活动将突出计算材料科学的潜力以及与实验和数据科学密切结合的需求，激励下一代MGI劳动力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

NON-TECHNICAL SUMMARYMembrane-based separations have revolutionized some industries (e.g., seawater desalination) and have the potential to drive numerous applications towards more energy efficient and sustainable processes. In chemical separations, which are typically highly energy intensive and account for greater than 5% of the annual primary energy consumption in the U.S., membrane-based approaches have significant potential to reduce both energy utilization and capital cost. However, membranes for the separation of mixed solvents, while practically important, are technically challenging as differentiating between the transport of small molecules that have subtle differences in properties is required. The rational design of the next generation of membranes for such separations is a significant challenge, given the vast chemical and design space, but could be realized by Materials Genome Initiative (MGI)-inspired screening. That is, the MGI-style of this effort could alter the current, long-standing membrane development paradigm. In this work, functionality- and performance-driven screening, with close coupling between simulations and experiment, will result in the design and fabrication of high-performance membranes tailored for targeted separations. Specifically, the dehydration of polar solvents by pervaporation will be targeted as an overarching initial target. This is because the discovery and deployment of effective new materials will eliminate the need for high-cost and high-energy separations and enable effective solvent reuse for sustainable manufacturing.TECHNICAL SUMMARYThis effort will develop an integrated, MGI-inspired computational and experimental screening platform with the goal of accelerating the rational design of membranes for the dehydration of polar solvents. This will be achieved through the combination of extensive molecular simulations using the Molecular Simulation and Design Framework (MoSDeF); machine learning using DeepForge; syntheses based around the combination of ring-opening metathesis polymerization chemistry combined with spin coating; experimental characterization; and in operando evaluation in a pervaporation process. This molecule-to-process approach will enable the synthesis of a wide array of polymer membrane compositions from a common central scaffold. Moreover, there will be an integral synergy between molecular-level screening simulations and experiment to molecularly design and identify new membranes that are tailored to specific dehydration separations. The goals of this project are the identification, synthesis, characterization, and testing of new candidate polymers for polar organic solvent-water membrane separations, development of a robust library of polymer membrane properties, development of machine learning models that relate chemistry to measured properties of membrane films, and the release of a generally applicable set of software tools that will enable rapid screening and machine learning studies on soft matter systems. Additionally, this effort, with its integration of computational modeling, machine learning, material synthesis, characterization, and performance evaluation for targeted separations, will serve as an excellent educational platform for participating graduate students and postdoctoral researchers to experience the full suite of interconnected components described in the MGI vision. By developing competency in the three foundational pillars of experiment, computation, and data science, the project will develop a workforce aligned with the MGI model. Also, multiple integrated educational activities at the undergraduate and K-12 levels will highlight the potential of computational materials science and the need for close coupling with experiment and data science, inspiring the next generation of the MGI workforce.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.