CAREER:Hybrid Data-driven Synthesis by Design of Atomically Thin Quantum Materials

职业：通过原子薄量子材料设计进行混合数据驱动合成

• 批准号: 1943857
• 负责人: Kasra Momeni
• 金额: $ 50.95万
• 依托单位: Louisiana Tech University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-03-01 至 2020-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1943857&HistoricalAwards=false
• 关键词: CAREER; Hybrid; Data; driven; Synthesis

Two-dimensional (2D) crystalline materials consisting of a single atomic layer have unique quantum mechanical properties that are critical for several advanced technological applications such as photovoltaic and electronic devices. However, the synthesis of 2D materials is generally accomplished through exhaustive trial-and-error experimentation that hinders their commercial exploitation. The main impeding factors are the lack of a comprehensive understanding of the underlying growth mechanisms and the lack of real-time measurement of growth states for implementing feedback process control. The research goals of this CAREER project are to (i) develop a computational model to understand the mechanisms governing the growth of 2D materials, (ii) build a database relating the synthesis process to the properties of these materials, and (iii) use artificial intelligence to find the optimum synthesis conditions. The proposed integration of research and education includes course development and laboratory modules for undergraduate and graduate students and research internships for undergraduate students. The outreach program will engage K-12 students and teachers as well as faculty and minority students from a local minority-serving institution.The proposed research focuses on developing a unified design multiscale framework addressing the growth of 2D materials using the more complex chemical vapor deposition-variant techniques that involve reactive flows of precursors. The objective is to understand the growth mechanisms, such as growth chemistry, and effect of different growth parameters, such as carrier gas flow rates, on the morphology and characteristics of the synthesized 2D materials. This multiscale framework will also be used to build a database of synthesis-morphology conditions to guide the design of new 2D materials. The developed synthesis-morphology database will be used in combination with the ML models, specifically Generative adversarial networks, to predict the morphology and properties of 2D materials significantly faster than the multiscale model. The objective is to develop a model that can be used as an observer with small enough response time that can be useful for real-time control of the synthesis process. The project will also focus on addressing the inverse problem of finding the optimal conditions for growing 2D quantum materials with desired properties. This problem will be first transformed into a classification problem using the synthesis-morphology database, which will be solved utilizing the ML models, and specifically Convolutional Neural Networks. Collaboration with industrial partners is planned through the I/UCRC Center for Atomically Thin Multifunctional Coatings (ATOMIC). The research results will be integrated into a new technical elective course and an existing undergraduate course on engineering materials. A light web-based version of the simulation software will be used for outreach activities and will be made available publicly through the website of the NSF-funded 2D Crystal Consortium - Materials Innovation Platform (2DCC). The outreach program will focus on engaging (1) K-12 students and teachers through STEM training camps and (2) faculty and students belonging to underrepresented minority groups in STEM from a local HBCU, Grambling State University, through computational teaching modules related to the Materials Genome initiative.This project is jointly funded by the Process Systems, Reaction Engineering, and Molecular Thermodynamics Program and the Established Program to Stimulate Competitive Research (EPSCoR).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.

由单个原子层组成的二维（2D）晶体材料具有独特的量子力学性质，这些性质对于诸如光伏和电子器件等若干先进技术应用至关重要。然而，二维材料的合成通常是通过详尽的试错实验来完成的，这阻碍了它们的商业开发。主要的阻碍因素是缺乏对潜在的生长机制的全面理解，以及缺乏对生长状态的实时测量以实施反馈过程控制。该CAREER项目的研究目标是（i）开发一个计算模型来理解2D材料生长的机制，（ii）建立一个将合成过程与这些材料的特性相关联的数据库，以及（iii）使用人工智能来找到最佳合成条件。拟议的研究与教育一体化包括为本科生和研究生开设课程和实验室单元，以及为本科生提供研究实习。外展计划将从事K-12学生和教师以及教师和少数民族学生从当地少数民族服务institution.The拟议的研究重点是开发一个统一的设计多尺度框架，解决二维材料的生长使用更复杂的化学气相沉积变体技术，涉及反应流的前体。其目的是了解生长机制，如生长化学，以及不同的生长参数，如载气流速，对合成的2D材料的形态和特性的影响。该多尺度框架还将用于构建合成形态条件数据库，以指导新的2D材料的设计。开发的合成形态数据库将与ML模型（特别是生成对抗网络）结合使用，以比多尺度模型更快地预测2D材料的形态和属性。我们的目标是开发一个模型，可以用作一个观察员，具有足够小的响应时间，可以是有用的合成过程的实时控制。该项目还将专注于解决逆问题，即找到生长具有所需特性的2D量子材料的最佳条件。这个问题将首先转化为使用合成形态学数据库的分类问题，这将利用ML模型，特别是卷积神经网络来解决。与工业合作伙伴的合作计划通过I/UCRC原子薄多功能涂层中心（原子）。研究成果将被整合到一个新的技术选修课和现有的工程材料本科课程。 该模拟软件的一个基于网络的轻型版本将用于外联活动，并将通过NSF资助的2D晶体联盟-材料创新平台（2DCC）的网站公开提供。 外展计划将侧重于参与（1）K-12学生和教师通过STEM训练营和（2）教师和学生属于代表性不足的少数群体在STEM从当地HBCU，Grambling州立大学，通过计算教学模块相关的材料基因组计划。该奖项反映了NSF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。