(147s) Data Analytics Enabled High Throughput Material Discovery and Computational Analyses for Membrane Innovations

Xinhong Liu is a passionate PhD candidate in chemical engineering with a minor in computational science and engineering at University of Notre Dame. Supervised by Prof. Alexander Dowling and Prof. William Phillip, her research centers on data analytics-enabled high-throughput material discovery and computational analyses of copolymer membranes for challenging separations. Xinhong has demonstrated her expertise through internships at NREL (National Renewable Energy Laboratory) with the ALIS (AI, Learning & Intelligent Systems) group, where she contributed to problem formulations with improved computational performance for two python packages, ideas-pse (Institute for the Design of Advanced Energy Systems Process Systems Engineering Framework) and waterTAP (Water treatment Technoeconomic Assessment Platform). Enjoying numerical and simulation-based research, she is actively seeking full-time employment starting in 2024 summer.

Research Interests

Xinhong’s expertise lies in process modeling & simulation, process control, nonlinear optimization, uncertainty analysis, design of Experiments, statistical learning, and membrane separation.

Abstract

Most of the research in membrane separation science is focused on developing better membrane materials, yet very few of these materials are being used in commercial applications. As such, there

are significant opportunities to broaden pathways from materials to industrial systems paved by Edisonian research for membrane technology. We bring up data analytics to elucidate the structure-property-performance relationships, which provide the fundamental knowledge to guide both the inverse material design and the large-scale process design for novel membrane applications.

Chemically patterned membranes with engendered useful characteristics can offer improved selective transport of electrolytes and material durability. Chemical patterning across the membrane surface via a physical inkjet deposition process requires precise control of the reactive-ink formulation, which enables the introduction of functionalities to the membrane. However, the “click” reaction for additive manufacturing of charged membranes involves an excess load of harmful components (i.e., copper), which hinders this type of membrane from being commercialized. The first application of our work seeks the optimal reactive ink formulations that enable precise controllable and environmental-friendly membrane functionalization process. We recently developed a dynamic mathematical model for the primary step of the batch reactive-ink formulation considering an ink mixture of copper sulfate and ascorbic acid, and we argued that pH measurements are insufficient to identify all the rate constants. In this talk, we demonstrate our investigation and experimental design process to elucidate the Copper(I)-Catalyzed Azide-Alkyne Cycloaddition (CuAAC) reaction mechanism. We verify the presence of two side reactions for 3-dimethylamino-1-propyne (DMA) through experimental design and quantify their effects in ink formulation preparation. Informed by Fourier Transform Infrared Spectroscopy (FTIR), we develop a dynamic model describing both the diffusion of the dinuclear copper-alkyne complex (DNCuAC) and the “click” reaction between the DNCuAC and azide moieties of the copolymer membrane. Building off the knowledge of the reaction network, we can optimize the reactive ink formulation that follows the fast reaction regime while minimizing the amount of copper based on specific coating time requirements. The workflow for ink formulation optimization here is applicable to other alkyne systems, which will not only promote the innovations of the chemically patterned membrane, but also develop the potential of CuAAC-based processes in other applications.

Membrane characterization provides essential information for the scale-up, design, and optimization of new separation systems. In the second part of our work, we proposed the Diafiltration Apparatus for high-Throughput Analysis (DATA) framework, which enables a 10-times reduction in the time necessary to characterize neutral membrane transport properties by integrating experiments, a new sensor, dynamic modeling, and parameter estimation. Recently, we extended the DATA framework to consider charged membranes. We postulate different physics-informed models to capture the concentration-dependent membrane performance. Using the tools of data science, we intelligently compare these model alternatives, and show that the solute permeability coefficient of NF270 membranes exhibits quadratic behaviors as a function of upstream conditions. Moreover, we extended the modeling framework to consider experiment start-up to leverage additional data to elucidate the physical system and improve the parameter precision. Using Fisher information matrix (FIM) analysis, we quantitatively compare the information gained for different experimental operating modes, i.e., “lag” or “overflow” startup. Additionally, a time correction for permeate product collected is introduced to improve the model predictions. Finally, we use model-based design of experiments (MBDoE) techniques to contemplate the benefits of modulating the applied pressure during experiments.

Select References

• Liu, X., De, R., Pérez, A., Hoffman, J. R., Phillip, W. A., & Dowling, A. W. (2022). Mathematical Modelling of Reactive Inks for Additive Manufacturing of Charged Membranes. In Computer Aided Chemical Engineering (Vol. 49, pp. 1063-1068). Elsevier.
• Ouimet, J. A., Liu, X., Brown, D. J., Eugene, E. A., Popps, T., Muetzel, Z. W., ... & Phillip, W. A. (2022). DATA: Diafiltration Apparatus for high-Throughput Analysis. Journal of Membrane Science, 641, 119743.
• Liu, X., Wang, J., Ouimet, J. A., Phillip, W. A., & Dowling, A. W. (2022). Membrane Characterization with Model-Based Design of Experiments. In Computer Aided Chemical Engineering (Vol. 49, pp. 859-864). Elsevier.