Deep learning has begun a renaissance in chemistry and materials. We can devise and fit models to predict molecular properties in a few hours and deploy them in a web browser. We can create novel generative models that were previously PhD theses in an afternoon. In my group, we’re exploring deep learning in soft materials and molecules. We are focused on two major problems: interpretability and data scarcity. Now that we can make deep learning models to predict any molecular property ad naseum, what can we learn? I will discuss our recent efforts on interpreting deep learning models through symbolic regression and counterfactuals. Data scarcity is a common problem in chemistry: how can we learn new properties without significant expense of experiments? One method is in judicious choice of experiments, which can be done with active learning. Another approach is self-supervised learning and constraining symmetries, which both try to exploit structure in data. I will cover recent progress in these areas. Finally, one consequence of the state of deep learning is that you can just make cool things in chemistry with minimal effort. I’ll review a few fun projects, including making molecules by banging on the keyboard, doing math with emojis, and doing molecular dynamics constrained on space groups.

Join Zoom Meeting Here. 

Meeting ID: 936 2999 1597 

Passcode: 830219 

Scott Cushing is an Assistant Professor at Caltech. His expertise lies in fundamental research in developing new spectroscopy, such as table-top XUV / soft x-ray pulses with down-to-attosecond time resolution. His lab is developing few-femtosecond, few-photon spectroscopy with entangled photons that can probe rare and intermediate events in the excited state. His lab also uses spectroscopy to explore ultrafast light-matter interactions, entangled-classical interactions, and multi-component spectroscopy of working devices. For more information go to cushinglab.caltech.edu Before joining Caltech, Dr. Cushing was a post-doctoral researcher at UC Berkeley. He holds a Ph.D. from West Virginia University.

Sijia Dong is an assistant professor in the Department of Chemistry and Chemical Biology at Northeastern University. Sijia is passionate about accelerating science using computation and automation. She received her Ph.D. in Chemistry from the California Institute of Technology in 2017, advised by Prof. William A. Goddard III, with whom and Dr. Ravinder Abrol she developed a first-principles-based and data-driven computational method to predict the structures of proteins that are crucial drug targets for many diseases. She carried out her postdoctoral research at the University of Minnesota with Prof. Donald G. Truhlar and Prof. Laura Gagliardi, and then at Argonne National Laboratory with Prof. Giulia Galli. Her postdoctoral work was to use and develop quantum chemical methods and workflows to study the photochemistry of molecules and materials in light-harvesting systems and to use machine learning to accelerate quantum chemical methods. Research in the Dong Lab focuses on developing and applying physics-based and data-driven computational methods to understand multiscale processes, from electronic structures to emergent properties, and to design molecules, materials, and processes for renewable energy, biomedicine, and beyond.

Join via Zoom here.

Meeting ID: 942 4035 1473 

Password: 303240

Dr. Brenda M. Rubenstein is a Joukowsky Family Professor of Chemistry at Brown University. She completed her Ph.D. from Columbia University and held postdoctoral positions at Lawrence Livermore before joining Brown University. The Rubenstein Group is a theoretical chemistry group that applies analytical and computational tools to explore a variety of problems in quantum chemistry, quantum physics, and biophysics. They currently focus on developing new methodologies for studying problems in quantum and statistical mechanics, molecular and quantum computing, and biophysical simulation.

Join ZOOM Here.

Meeting ID: 996 6445 3011

Passcode: 525657

Heather J. Kulik is an American computational materials scientist and chemist who is an associate professor at the Massachusetts Institute of Technology. Her research considers the computational design of new materials and the use of artificial intelligence to predict material properties. She received her B.E. in Chemical Engineering from Cooper Union in 2004 and her Ph.D. in Materials Science and Engineering from MIT in 2009. She completed postdocs at Lawrence Livermore (2010) and Stanford (2010−2013), prior to returning to MIT as a faculty member in 2013 and receiving tenure in 2021. In this talk, she will give us a broad overview of her research and some interesting developments in the field.

Join via Zoom here. 

Meeting ID: 947 6889 8894

Passcode: 247347

Jawwad Darr is Professor of Materials Chemistry at University College London.  He has published over 200 academic publications in advanced functional materials including catalysts and batteries (h-index 50, 9k citations, google scholar), has three patents and has graduated over 45 PhD students (as of 2021).  From 2016 to 2021, he was Vice Dean of Enterprise in the Maths and Physical Sciences faculty at UCL in, which he was supporting UCL’s faculty engagement with industry. He has strong research collaborations with industry and is a regular speaker at industrial forums and academic events in a wide range of topics.  As well as his own latest research, he teaches on topics related to green chemistry, applications of advanced functional materials (including batteries) and at outreach lectures in the UK and overseas. 

In 2017, he co-founded the UPSIGN charity (UK-Pakistan Science and Innovation Global Network; www.upsign.org.uk) that works to support, educate and connect British Pakistani and Pakistani academics.  In his UPSIGN work, he is involved in public outreach workshops and lectures, training Pakistani academics and students, and developing training and workshops to support underprivileged undergraduate black and minority Asian students in research. In his spare time, he enjoys dabbling in interior design, reading about the origins of life and evolution, watching/playing association football. 

Evaluation Committee:

Dr. Basit Yameen (Supervisor) 

Dr. Irshad Hussain (Thesis Committee Member) 

Dr. Nauman Zafar Butt (Thesis Committee Member) 

Dr. Salman Noshear Arshad (Thesis Committee Member)

Meeting ID: 919 0187 5994

Passcode: 624794

Hydrogen gas is regarded as a sustainable, environment-friendly, and promising source of energy for the future. Water electrolysis is a very attractive process for the production of H2 in which two half-reactions are involved: oxygen evolution reaction (OER) and hydrogen evolution reaction (HER) [1]. Electrocatalytic water splitting is a well-established process, but there are still several limitations for the large-scale production of hydrogen. Firstly, the most important factor is the OER reaction that requires large electrical energy and it is a kinetically slow process. Secondly, the unavoidable mixture of oxygen gas with hydrogen gas may propel the reaction to explosion in the electrolyzer. Several approaches have been suggested to reduce energy consumption and avoid the possibility of explosion. Chemical-assisted electrocatalytic water splitting is a cost-effective approach for replacing OER with a thermodynamically favorable anodic oxidation reaction in water. Electrochemical oxidation of methanol, some biomass-derived alcohols (ethyl alcohol, benzyl alcohol, and furfuryl alcohol), and carbohydrates (glucose, fructose, maltose, and sucrose) show not only high performance for hydrogen production but also produce high-value by-products. To date, noble metals such as Pt, Pd, Ir, Ru, and their alloys are the most efficient catalysts for the chemical oxidation in the chemical-assisted hydrogen evolution reaction (CAHER). Several critical challenges in the development of the CAHER system still need to be addressed such as the issue of low catalytic selectivity and efficiency toward the most valuable products in alcohol electrooxidation reactions etc. Recently, we reported 3D hierarchical porous NiCo bimetallic foams on Cu foil electrodes, prepared using a single-step bubble templating electrodeposition process, as efficient electrocatalysts for overall methanol-assisted energy-saving hydrogen production with the co-generation of valuable formate product [3]. To investigate more efficient electrocatalysts, we also synthesized nanostructured NiCu porous bimetallic foams on Cu foil electrodes by a simple, ultrafast, and single-step bubbles templating electrodeposition method. Benefiting from the excellent electronic conductivity, high internal reactive surface areas, profiting electrolyte accessibility, and effective mass transfer at the electrolyte/electrode interface, porous NiCu electrode required a cell voltage of 1.47 V to attain a current density of 10 mA cm-2 for integrating selective methanol oxidation to formate and energy-saving hydrogen production with continuous 18 h of operation. This study also emphasizes the critical role of Cu atoms in modifying the charge distribution of the Ni site and improving selectivity and instruct activity for MOR and HER.