Computational prediction of hot-electron chemistry: Towards electronic control of catalysis

Lead Research Organisation: University of Warwick
Department Name: Chemistry


Higher living standards and a growing world population are the drivers behind continuous increases in greenhouse gas emission and industrial energy use. This provides growing pressure on chemical industries to develop more sustainable and efficient chemical transformations based on innovative new technologies. Light-driven plasmonic catalysis offers a promising route to more sustainable and energy efficient chemical transformations than conventional industrial-scale catalysis by replacing petrochemical reactants and energy sources with abundant feedstocks such as carbon dioxide from the atmosphere and renewable energy from sunlight. In addition, light energy can selectively be transferred via excited electrons in metal nanoparticles, so-called "hot" electrons, to molecules and enables more specific chemical reactions than conventional catalysis, potentially increasing yield and decreasing unwanted side products.

Underlying this unconventional form of chemistry is the intricate coupling of light, hot electrons, and reactant molecules, the lack of understanding of which has inhibited systematic design and study of reaction parameters such as particle size, shape, and optimal light exposure. A predictive theory of hot-electron chemistry will support the adaptation of this technology in the chemical industry, which holds the potential to significantly reduce the industry's carbon footprint.

The aim of this project is to develop and exploit a computational simulation framework to understand, predict, and design light-driven chemical reactions on light-sensitive metallic nanoparticles and surfaces, so-called plasmonic nanocatalysts. The vision behind this fellowship is to provide quantum theoretical methods that fill a conceptual and methodological gap by providing accurate and feasible computational prediction of experimentally measurable chemical reaction rates as a function of catalyst design parameters relevant to the real-world application of this technology.

In synergy with experimental project partners, the fellow will lead a research team of 2 postdoctoral researchers to develop highly efficient computational chemistry methodology, which will be applied to scrutinize mechanistic proposals, support and guide experimental efforts on light-driven plasmonic carbon dioxide reduction chemistry, and to construct reaction rate models relevant to improve the industrial viability of this technology. The aim is to provide a step-change in the mechanistic understanding of light-driven plasmonic reduction catalysis on the example of carbon monoxide and carbon dioxide transformation to enable rational design of catalyst materials with wide implications for continuous photochemistry and electrochemistry applications in industry. These applications will be explored by continuous engagement efforts of the fellow with leading chemical and petrochemical companies. With this project, the fellow will establish an international track record by fostering existing and establishing new collaborations with the goal to become a recognized researcher in this comparably young field.

Planned Impact

Continuing fossil fuel depletion and accumulation of greenhouse gases in the atmosphere will become the defining threats to living standards, a healthy society, and energy security for more than 10 billion people inhabiting the planet in the second half of the 21st century. To significantly reduce these threats, while maintaining the important socioeconomic role of industrial catalysis, innovative catalytic pathways need to be identified that are both efficient and environmentally sustainable. This extraordinary challenge amounts to nothing less than transforming the national and global petrochemicals and commodity chemicals industry to build on abundant feedstocks such as carbon dioxide, water, and renewable energy from sunlight with product yields that push beyond the thermodynamic limitations of conventional industrial catalysis. Our proposed research on computationally predicting light-enhancement of catalysis will directly impact the realization of a technology that can contribute to such a transformation.
1. Academic Impact
Hot-electron chemistry, the subject of this project, touches on many disciplines and is driven by plasmonic light-matter interaction with numerous proposed applications beyond catalysis that include sensors, mobile fuel generation devices, and energy harvesting materials. These areas are of specific interest for defence applications developed by academics, the UK Defence Science and Technology Laboratory, and the US Department of Defense Research Offices. The methods we develop and the models we will construct will impact our Academic Beneficiaries and the above stated research areas via a dissemination strategy targeted to actively engage a wide range of scientific communities.
2. Economic and Environmental Impact via Innovation in Industrial Catalysis
This project will develop models to predict catalyst structure, cost, and efficiency relations that support the adaptation of hot-electron-enhanced chemistry in industrial catalysis. This has the potential to provide drastic gains in energy efficiency, product selectivity, and a reduction in carbon footprint on a planet-wide scale. This will provide an important know-how advantage for the UK petrochemical and commodity chemicals industry represented by companies such as BP and Johnson-Matthey against low-tech competitors in emerging markets. The Fellow is new to the UK research and innovation landscape and this fellowship will enable him to identify industrial contacts and actively engage with them to effectively communicate the commercial implications of our findings.
3. Societal Impact via Maintained and Improved Living Standards
Long-term industrial adaptation of plasmonic catalysis technology will benefit environmental protection and energy efficiency on a scale that will affect individual UK consumers and households. The research we propose aims to impact this technology by providing theoretical rate models and structure-function relationships that support catalyst design. We propose several outreach efforts to inform policy makers and the wider public of the importance and potential benefit of innovation in industrial catalysis and the important role of computational modelling as a driver for such innovation.
4. Impact on People and Skills
We will use and develop methods of computational science, artificial intelligence, and data-driven design, which are key to further develop the UK's digital economy. The two postdoctoral researchers funded by this project will develop skills that are in high demand in academia and industry. They have the potential to become leading innovators in the UK's growing data science and computational modelling sector. The fellow will fully utilize this fellowship to establish himself as an internationally recognized leader in the field of plasmonic catalysis and computational chemistry. This will be supported by our dissemination plans designed to engage a diverse range of academics and industrial stakeholders
Description The activities associated with this award can be broken down into: (A) machine-learning-based model creation, (B) computational method development incl. software development, (C) simulation of key quantities for hot-electron catalysis. In the first year of this award, the research team has been assembled and important preliminary method developments (B) have started. Initial progress in (A) and (C) have led to scientific publications (either published or currently being drafted). The main outcome (A) was the creation of a machine learning algorithm that can predict molecular wave functions, which will be highly useful for a broad range of applications in computational simulation and for the next steps in this project.
Exploitation Route Research outcomes in the three activity streams of this project are geared to generate outcomes that can be widely used and easily taken forward by others. In particular, the developed machine-learning algorithm to predict molecular wave functions (SchNOrb) has been made public and has the potential to be useful for a wide range of researchers in academia and industry. This will include its application and repurposing to address research problems in synthetic chemistry and catalysis.
Sectors Chemicals,Energy,Environment

Description Artificial and Augmented Intelligence for Automated Scientific Discovery
Amount £1,014,318 (GBP)
Funding ID EP/S000356/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 06/2018 
End 06/2021
Title SchNOrb machine learning model 
Description The SchNOrb machine learning model is a deep tensor neural network that reads in molecular geometries (atom positions and elemental composition) and predicts molecular wave functions and other molecular electronic properties (such as the total energy). 
Type Of Material Computer model/algorithm 
Year Produced 2019 
Provided To Others? Yes  
Impact The SchNOrb model is open source and publicly available and can be used to generate a machine-learning based surrogate models of the electronic structure of a molecule. This model can be extended in many different ways to benefit research in molecule and materials design with applications in pharmaceutical and catalysis research. 
Description David Duncan, I09, Diamond Light Source 
Organisation Diamond Light Source
Country United Kingdom 
Sector Private 
PI Contribution I have contributed computational simulation expertise to research projects by Dr David Duncan and Dr. Tien-Lin Lee at Diamond Light Source. Team members from my research group will provide simulation data to support the design and characterization of novel metal catalysts.
Collaborator Contribution Diamond Light Source (via direct collaboration with Dr David Duncan and Dr. Tien-Lin Lee) have contributed funding to support 50% of the cost of a PhD studentship (£51,184) and funding for 3 months of salary for a postdoctoral fellow (£11,681) to support my research efforts in the wider context of this project. The corresponding staff members will perform experimental measurements at Diamond Light Source which will support the efforts in this award.
Impact No outcomes have yet resulted from this collaboration.
Start Year 2019
Description Jiang, USTC, Hefei 
Organisation University of Science and Technology of China USTC
Country China 
Sector Academic/University 
PI Contribution Within this research collaboration, we provide expertise on nonadiabatic electron-nuclear coupling effects during gas-surface dynamics in heterogeneous catalysis simulations. We have provided the collaborators with a large amount of simulation data for several relevant systems incl. molecular hydrogen dissociation on Ag(111) surfaces and nitrous oxide scattering on Au(111).
Collaborator Contribution The collaboration partners have provided us with their expertise in machine learning-based potential energy surface interpolation and have given us access to their extensive software stack for gas-surface dynamics simulations.
Impact One publication has resulted from this collaboration: 10.1021/acs.jpcc.9b09965 This collaboration is not multi-disciplinary
Start Year 2019
Description Klaus-Robert Mueller, TU Berlin 
Organisation Technical University Berlin
Country Germany 
Sector Academic/University 
PI Contribution I have contributed my expertise in electronic structure theory and computational chemistry to this collaboration. I and my team members have performed a large nuber of quantum chemical calculations to generate training data sets. These data sets are used to train deep learning models.
Collaborator Contribution The collaborator and his team have provided us with crucial expertise and software to develop deep machine learning models that support the efforts in this award. The contribution consists of time for regular discussions, joint manuscript preparation, and hosting members of my team at the TU Berlin.
Impact One publication has resulted from this collaboration: 10.1038/s41467-019-12875-2
Start Year 2019