(6bd) Molecular Modeling and Machine Learning for Catalysis and Separations

Research Interests:

As earth’s population grows and the standard of living in developing nations improves, demand for energy and water will increase, furthering the strain on global resources. As production increases to meet demand, technologies that save energy and reduce pollution are essential to minimize environmental impact. Design of high-performance materials for catalysis and separations can reduce waste, conserve energy, and enable new green chemistries. Nanoporous adsorbents and catalysts are critical technologies with the prospect of controlling molecular-scale interactions through material and process design. Consequently, progress toward optimal materials and chemical processes can be accelerated with molecular modeling. Simulations of molecular-scale phenomena provide valuable insight into mechanisms of adsorption, transport, and reaction in complex environments. Computer-aided screening across the expansive space of materials and processes can focus experimentalists to synthesize the most promising materials and test them at optimal conditions. As vast libraries of materials are created in lab and in silico, extracting actionable information from these big data can be difficult. Techniques from machine learning and data science can be harnessed to identify critical trends and descriptors, as well as build predictive models for discovery.

Machine learning is experiencing a Renaissance in the chemical sciences, being used to make sense of the vast and diverse spaces of all possible molecules, materials, and reactions. Less attention is focused on characterizing process space – the space of industrially-relevant temperatures, pressures, and compositions – in which complex phase behavior, and non-ideal solvent/solute/surface interactions influence process performance. My research group will combine expertise in electronic structure calculations, Monte Carlo and molecular dynamics simulations, force field fitting, thermodynamics, and machine learning to investigate challenging problems in these complex chemical systems. Leveraging these techniques in collaboration with experimentalists will accelerate progress toward a more sustainable future.

Research Experience

During my Ph.D. with Prof. Dion Vlachos, I applied molecular dynamics and electronic structure calculations to investigate fundamental interactions between solvents, solutes, and catalysts in biomass catalysis. In my first project, I discovered how co-solvent interactions protect biomass-derived furans from side reactions in fructose dehydration [1] [2]. I developed structure-activity relationships for homogeneous Lewis acids to probe mechanistic details in glucose isomerization [3] [4]. I also exhaustively screened site geometries for Sn-Beta, identifying the most stable active geometry and proposing a source for its Brønsted acidity [5]. In my final project, which I continued during my post doc, I developed the reaction mechanism for fructose etherification on the hierarchical Sn-SPP zeolite and integrated it with multicomponent liquid-phase adsorption [7].

In my current post doc with Prof. Ilja Siepmann, I am developing and applying Monte Carlo and molecular dynamics simulations in five projects: 1) predicting solution-phase adsorption into hierarchical zeolites under reaction conditions [10], 2) investigating adsorption of organics at the air-water surface [6], 3) calculating the viscosities of novel lubricants [8], 4) predicting thermophysical properties of hydrocarbons [9], and 5) developing molecular modeling and machine learning techniques to predict adsorption isotherms [10].

Teaching Interests:

Excellence in teaching has the potential to multiply progress made toward a more sustainable future through the engineers who will come after us.

I have served as a teaching assistant in both undergraduate and graduate thermodynamics courses, and I have been a guest lecturer in kinetics and reaction engineering at 3 institutions. In addition, I organized a computational chemistry workshop for the graduate reaction kinetics course, to guide students through exercises using the same software I use in my research.

I am particularly excited about introducing machine learning and data science into undergraduate and graduate Chemical Engineering education. By broaching these topics in core courses and going deeper in electives, students will become acquainted with cutting-edge data analysis tools and equipped for careers in data science and its intersection with traditional chemical engineering careers.

Courses of expertise: thermodynamics and statistical mechanics, kinetics and reaction engineering, numerical methods and linear algebra, computational chemistry, and engineering statistics.

Postdoctoral Research Associate

Postdoctoral Project: Monte Carlo Simulations for Novel Materials and Process Design

Advisor: Prof. J. Ilja Siepmann, Department of Chemistry and Department of Chemical Engineering and Materials Science, University of Minnesota

PhD Dissertation: Leveraging First-Principles Simulations for Deeper Insights in Biomass Processing

Advisors: Prof. Dionisios Vlachos and Dr. Stavros Caratzoulas, Department of Chemical and Biomolecular Engineering, University of Delaware

AIChE Conference Talks:

[10] Josephson, T. R., Sun, Y., and Siepmann, J. I. “Molecular simulations and machine learning for multicomponent adsorption.” Session: Faculty Candidates in CoMSEF/Area 1a

[9] Josephson, T. R., Singh, R., Minkara, M. S. , and Siepmann, J. I., 2019, “Partial molar properties from molecular simulation using multiple linear regression.” Session: Recent Advances in Molecular Simulation Methods II

Selected Publications

[9] Josephson, T. R., Singh, R., Minkara, M. S. , and Siepmann, J. I., 2019, “Partial molar properties from molecular simulation using multiple linear regression.” In Review

[8] Liu, S., Josephson, T. R., Athaley, A., Chen, Q. P., Norton, A., Ierapetritou, M., Siepmann, J. I., Saha, B., Vlachos, D. G., 2019. “Renewable lubricants with tailored molecular architecture.” Science Advances. doi: 10.1126/sciadv.aav5487

[7] Josephson, T. R., DeJaco, R., Pahari, S., Ren, L., Guo, Q., Tsapatsis, M., Siepmann, J.I., Vlachos, D., Caratzoulas, S., 2018. “Cooperative Catalysis by Surface Lewis Acid/Silanol for Selective Fructose Etherification on Sn-SPP Zeolite.” ACS Catalysis. doi: 10.1021/acscatal.8b01615

[6] Minkara, M. S., Josephson, T. R., Venteicher, C. L., Chen, J. L., Stein, D. J., Peters, C. J., Siepmann, J. I., 2018. “Monte Carlo simulations probing the liquid/vapour interface of water/hexane mixtures: adsorption thermodynamics, hydrophobic effect, and structural analysis.” Molecular Physics. doi: https://doi.org/10.1080/00268976.2018.1471233

[5] Josephson, T. R., Jenness, G. R., Caratzoulas, S., Vlachos, D. G., 2017. Distribution of open sites in the Sn-Beta zeolite." Microporous and Mesoporous Materials. doi: 10.1016/j.micromeso.2017.02.065

[4] Josephson, T. R., Brand, S. K., Caratzoulas, S., Vlachos, D. G., 2016. 1,2-H versus 1,2-C-shift on Sn-silsesquioxane." ACS Catalysis, 7. doi: 10.1021/acscatal.6b03128

[3] Brand, S. K., Josephson, T. R., Labinger, J. A., Caratzoulas, S., Vlachos, D. G., Davis, M. E., 2016. Methyl-ligated tin silsesquioxane catalyzed reactions of glucose." Journal of Catalysis. doi: 10.1016/j.jcat.2016.06.013

[2] Josephson, T. R., Tsilomelekis, G., Bagia, C., Nikolakis, V., Vlachos, D. G., and Caratzoulas, S., 2014. Solvent-induced frequency shifts of 5-hydroxymethylfurfural deduced via infrared spectroscopy and ab initio calculations." Journal of Physical Chemistry A. doi: 10.1021/jp508340p

[1] Tsilomelekis, G., Josephson, T. R., Nikolakis, V., and Caratzoulas, S., 2013. Origin of 5-hydroxymethyl furfural stability in water/dimethyl sulfoxide mixtures." ChemSusChem. Cover article. doi: 10.1002/cssc.201300786