CAREER: Data-driven design of graphene oxide for environmental applications enabled by natural language processing and machine learning techniques

氧化石墨烯(GO)是一种很有前途的功能纳米材料和构建块，可用于各种环境技术，包括用于饮用水滤膜的抗菌涂层、用于去除空气和水中污染物的吸附剂，以及用于去除和破坏受污染水中有机污染物的光催化剂。然而，目前用于设计、合成和优化用于有针对性的环境应用的GO基纳米材料的方法存在缺乏标准化的问题，导致大量的试验和错误运行，材料开发成本高昂。这个职业项目的总体目标是探索利用数据驱动的方法来表征和揭示GO合成条件和材料性能之间的关键关联，这将使基于GO的净水和环境修复技术的设计和开发成为可能。为了推进这一目标，首席调查员(PI)建议使用自然语言处理和机器学习技术来1)从数十万篇致力于围棋合成、表征和应用的科学论文中提取同行评议的信息，2)将这些知识构建成稳健的数据集，3)利用这些数据集来开发和实验验证围棋合成条件和性质之间的结构-性质关系。这一项目的成功完成将使社会受益，因为它将产生新的基本知识，并创建经过整理的数据集和相关的计算工具，以推动基于GO的环境技术的设计和开发。学生教育和培训将为社会带来更多好处，包括指导佛罗里达大学的一名研究生。用于水净化和环境修复的定制氧化石墨烯(GO)功能纳米材料的设计和合成将需要了解材料合成输入参数和所产生的材料性能之间的关系。GO的结构、物理化学和功能表面/本体性质取决于几个参数，包括前驱体石墨的性质和特性、合成条件和合成后处理方案。在这个职业项目中，首席调查员(PI)建议将自然语言处理(NLP)与机器学习(ML)以及定向材料合成和表征实验相结合，以开发和验证GO合成条件与性能之间的结构-性质关系，以促进基于GO的净水和环境修复技术的基本设计和开发。研究的具体目标是：1)使用自然语言处理工具(例如，潜在语义分析和命名实体识别)来自动提取关于GO的合成(例如，合成条件、前体材料和制造后处理)和所得到的GO的材料性能(例如，片材的大小和各种物理/化学性能)的相关信息；2)将这些信息构建到数据集，并使用经典的频率分析方法和其他数据分析技术(例如，主成分分析、线性判别分析和T分布随机邻近嵌入)来找到合成输入参数与GO的最终材料特性之间的相关性，以及3)通过有针对性的实验来验证这些相关性，包括合成的GO纳米材料作为饮用水滤膜和吸附剂/光催化剂的抗菌涂层的材料合成、表征和性能评估，以去除和破坏污染水中的有机污染物。该项目的成功完成有可能通过基础知识和结构化数据集的产生产生变革性的影响，以推进用于水净化和环境修复的GO基纳米材料的基本设计。为了实现这一职业项目的教育和培训目标，国际和平研究所建议开发和教授一门关于纳米材料数据驱动设计的研究生课程，重点是环境应用。此外，PI计划1)从代表性不足的群体中招募和指导本科生，2)开发一个在线研讨会系列，目的是为佛罗里达大学的学生在材料科学、数据分析和环境工程领域解决重要问题的培训做出贡献。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Graphene oxide (GO) has emerged as a promising functional nanomaterial and building block for various environmental technologies including antimicrobial coatings for drinking water filtration membranes, sorbents for the removal of pollutants from air and water, and photocatalysts for the removal and destruction of organic pollutants from contaminated water. However, the current approaches used to design, synthesize, and optimize GO-based nanomaterials for targeted environmental applications suffer from a lack of standardization leading to numerous trial and error runs with prohibitive material development costs. The overarching goal of this CAREER project is to explore the utilization of data-driven approaches to characterize and unravel critical correlations between the synthesis conditions of GO and the material properties that will enable the rationale design and development of GO-based water purification and environmental remediation technologies. To advance this goal, the Principal Investigator (PI) proposes to use natural language processing and machine learning techniques to 1) extract peer-reviewed information from hundreds of thousands of scientific papers devoted to GO synthesis, characterization, and applications, 2) structure this knowledge into robust datasets, and 3) leverage these datasets to develop and experimentally validate structure-property relationships between the synthesis conditions and properties of GO. The successful completion of this project will benefit society through the generation of new fundamental knowledge and the creation of curated datasets and associated computational tools to advance the design and development of GO-based environmental technologies. Additional benefits to society will be achieved through student education and training including the mentoring of a graduate student at the University of Florida.The design and synthesis of tailored graphene oxide (GO)-based functional nanomaterials for water purification and environmental remediation will require a knowledge of the relationships between the material synthesis input parameters and the resulting material properties. The structural, physicochemical, and functional surface/bulk properties of GO depend on several parameters including the nature and characteristics of the precursor graphite, the synthesis conditions, and the post-synthesis treatment protocols. In this CAREER project, the Principal Investigator (PI) proposes to combine natural language processing (NLP) with machine learning (ML) and targeted materials synthesis and characterization experiments to develop and validate structure-property relationships between the synthesis conditions and properties of GO to advance the rationale design and development of GO-based water purification and environmental remediation technologies. The specific objectives of the research are to: 1) use NLP tools (e.g., latent semantic analysis and named-entity recognition) to automatically extract relevant information about the synthesis of GO (e.g., synthesis conditions, precursor materials, and post-fabrication treatments) and the resulting material properties of GO (e.g., size of sheets, and various physical/chemical properties); 2) structure this information into datasets and use classical frequentist approaches and other data analysis techniques (e.g., principal component analysis, linear discriminant analysis, and T-distributed stochastic neighbor embedding) to find correlations between the synthesis input parameters and the resulting material properties of GO, and 3) validate these correlations using targeted experiments including material synthesis, characterization, and performance evaluation of the synthesized GO nanomaterials as antimicrobial coatings for drinking water filtration membranes and sorbents/photocatalysts for the removal and destruction of organic pollutants from contaminated water. The successful completion of this project has the potential for transformative impact through the generation of fundamental knowledge and structured datasets to advance the rationale design of GO-based nanomaterials for water purification and environmental remediation. To implement the educational and training goals of this CAREER project, the PI proposes to develop and teach a graduate course on data-driven design of nanomaterials with a focus on environmental applications. In addition, the PI plans to 1) recruit and mentor undergraduate students from underrepresented groups and 2) develop an online seminar series with the goal of contributing to the training of students at the University of Florida to work on important problems at the interface of materials sciences, data analysis, and environmental engineering.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.