FINALES (06/2022) – Electrolyte Optimization for Minimum Density and Maximum Viscosity

This study presents the initial implementation of the Fast INtention-Agnostic LEarning Server (FINALES) in a demonstration of a distributed Materials Acceleration Platform (MAP) including experimental and computational methods and a machine learning (ML)-based optimizer. In this demonstration, the optimizer was configured to minimize the density of the electrolyte solutions while maximizing the viscosity by exploiting experimental and computational results. The tenants (the units connected to FINALES in the MAP) are shortly described in the following:

- Autonomous Synthesis and Analysis of Battery electrolytes (ASAB) setup: an experimental tenant providing density and viscosity data using a densimeter of the type DMA 4100M and a viscometer of type Lovis 2000 both by Anton Paar Germany
- Molecular dynamics tenant: a computational tenant capable of providing radial distribution functions, diffusion coefficients, ionic conductivity, transference numbers, heat capacity and density data to the MAP
- Optimizer: the tenant guiding the optimization by processing the available data and generating requests for electrolyte formulations to be tested subsequently

This dataset accompanies the publication:
Vogler, M., Busk, J., Hajiyani, H., Jørgensen, P. B., Safaei, N., Castelli, I. E., Ramirez, F. F., Carlsson, J., Pizzi, G., Clark, S., Hanke, F., Bhowmik, A. & Stein, H. S. Brokering between tenants for an international materials acceleration platform. Matter 6, 2647–2665 (2023). DOI: https://doi.org/10.1016/j.matt.2023.07.016

本研究展示了快速意图无关学习服务器（Fast INtention-Agnostic LEarning Server，简称FINALES）的初步实现，并在分布式材料加速平台（Materials Acceleration Platform，简称MAP）的演示中，结合了实验和计算方法以及基于机器学习（Machine Learning，简称ML）的优化器。在该演示中，优化器被配置为通过利用实验和计算结果，在最小化电解质溶液密度的同时最大化其粘度。以下简要介绍了连接至FINALES的MAP中的各个租户（单元）：

- 电池电解质自主合成与分析设置（Autonomous Synthesis and Analysis of Battery electrolytes，简称ASAB）：由德国 Anton Paar 公司提供的DMA 4100M 密度计和 Lovis 2000 型粘度计进行实验，以提供密度和粘度数据
- 分子动力学租户：一种计算租户，能够为MAP提供径向分布函数、扩散系数、离子电导率、迁移数、热容和密度数据
- 优化器：通过处理可用数据并生成后续需测试的电解质配方请求，指导优化的租户

本数据集伴随以下出版物：
Vogler, M.，Busk, J.，Hajiyani, H.，Jørgensen, P. B.，Safaei, N.，Castelli, I. E.，Ramirez, F. F.，Carlsson, J.，Pizzi, G.，Clark, S.，Hanke, F.，Bhowmik, A. & Stein, H. S. 为国际材料加速平台进行租户间的经纪。Matter 6, 2647–2665（2023）。DOI：https://doi.org/10.1016/j.matt.2023.07.016