CDS&E: Development of Computational Library for Accurate Binding Energies of Emerging Organic Contaminants on Environmental Interfaces

当化学品泄漏事件发生时，应急响应人员采取行动保护人类健康和环境。然而，有关泄漏的化学物质如何相互作用并在环境中移动，最终会在哪里，或者如何过滤掉的信息并不总是可用的。在漏油事件发生后，能够快速预测这些行为的方法将是有用的。在化学学部环境化学科学（ECS）项目的支持下，孟菲斯大学的William Alexander教授正在使用计算机模型来快速估计泄漏后污染物的行为。亚历山大教授与他的学生一起开发了自动化计算化学工具，以预测污染物的物理性质，并预测环境或基础设施表面上“它们可能会粘在哪里”。该小组正在建立一个与环境相关的表面模型数据库，并使用这些模型来计算筛选污染物可能粘附的表面。该小组得出的模型和工具可以使响应者获得对新型污染物化合物行为的快速估计。这将有助于援助和指导可能对公共安全和危机后经济复苏产生重大影响的补救战略。该项目还为来自代表性不足的学生群体的研究生和本科生提供多学科合作科学培训。外展工作旨在扩大家庭教育学生接触实验室科学的机会，重点是环境科学和化学实验室经验。从2004年到2014年，美国国家应急中心报告了大约17.2万起影响美国水体的化学品泄漏事件。其中一些是由于具有已知特性和毒性的化学品造成的，而对于其他化学品，可获得的信息相对较少。定性模型通常用于预测污染物的命运和运输。这些模型通常不考虑污染物行为的重要动力学方面，如溶剂化效应和构象平均。此外，污染物在环境（淤泥，粘土等）和水系统（过滤介质，聚合物管道等）中相互作用的大多数相关表面在本质上是动态的和无定形的。将这些动力学方面纳入预测模型将提高所得到的物理性质和结合估计的准确性。在该项目中，输入文件和作业管理是自动化的，用于快速量子力学（QM）预测污染物的物理性质，包括构象和溶剂化效应。从分子力学（MM）模拟中产生了相关的非晶表面模型库（即非晶二氧化硅，碳，聚合物）。这些方法被结合到QM/MM方案中，以实现污染物结合能的快速计算，从而指导过滤介质的选择并预测污染物可能集中在哪里。这些方法是通过一系列化合物类别的已知有机污染物分子的大型测试集来验证的，包括从公共卫生机构优先列表中选择的数十种化合物。通过新的分配和吸附研究，选定的分子用于实验验证计算的分子性质。在整个调查过程中，开发了新的工作流程来管理大型数据集。除了准确预测污染物行为外，由此产生的方法还可以通过减少现场样品和时间来帮助减少人类在现场接触污染物化合物。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。

When a chemical spill incident happens, emergency responders act to protect human health and the environment. However, information is not always available about how the spilled chemical will interact and move through the environment, where it will end up, or how it might be filtered out. After a spill, methods that can quickly predict these behaviors would be useful. With support from the Environmental Chemical Sciences (ECS) Program of the Chemistry Division, Professor William Alexander of The University of Memphis is using computer models to quickly estimate how contaminants may act after a spill. Working with his students, Professor Alexander develops automated computational chemistry tools to predict the physical properties of contaminants, and to predict "where they might stick?" on environmental or infrastructure surfaces. The group is building a data library of environmentally-relevant surface models and using these models to computationally screen what surfaces the contaminants may stick to. The group's resulting models and tools could allow responders to obtain rapid estimates of the behavior of novel contaminant compounds. This will help aid and direct remediation strategies that could have a significant impact on public safety and post crisis economic recovery. The project is also providing multidisciplinary collaborative scientific training for graduate and undergraduate students from underrepresented student populations. Outreach efforts are designed to expand access to laboratory science for home-schooled students, with a focus on environmental science and chemistry laboratory experiences.About 172,00 chemical spills impacting US bodies of water were reported to the US National Response Center from 2004-2014. Some of these are due to chemicals with known properties and toxicity, whereas for others, relatively little information is available. Qualitative models are often used to predict contaminant fate and transport. These models generally do not treat important dynamical aspects of contaminant behavior such as solvation effects and conformational averaging. Also, most relevant surfaces that contaminants interact with in the environment (silts, clays, etc.) and within the water system (filter media, polymer plumbing, etc.) are dynamical and amorphous in nature. Incorporating these dynamical aspects into prediction models will increase the accuracy of the resulting physical properties and binding estimates. In the project, input files and job management are automated for rapid quantum mechanical (QM) prediction of contaminant physical properties, including conformation and solvation effects. A library of relevant amorphous surface models (i.e. amorphous silica, carbon, polymers) derived from molecular mechanics (MM) simulations is generated. These approaches are combined into QM/MM schemes to enable rapid computation of contaminant binding energies to direct filter media selection and to predict where contaminants may concentrate. The methods are validated using a large test set of known organic contaminant molecules in a range of compound classes, including dozens of compounds selected from public health agency priority lists. Selected molecules are used to experimentally validate computed molecular properties by new partitioning and adsorption studies. Throughout the investigations, new workflows are developed to manage the large datasets. In addition to accurately predicting contaminant behavior, the resulting methods may help to reduce human exposure to contaminant compounds in the field by requiring less field samples and time.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.