|Funding for:||UK Students, EU Students, International Students|
|Funding amount:||A fully funded 4-year PhD project|
|Placed On:||24th July 2023|
|Closes:||1st March 2024|
Supervisor: Prof. Reinhard J. Maurer, Department of Chemistry & Department of Physics, University of Warwick
Funding: Home, EU, Overseas
Deadline: 01.03.2024, applications will be considered on a rolling basis please post for 90 days on FindAPhD or until the deadline (on jobs.ac.uk)
A fully funded 4-year PhD project in property-driven generative machine learning for tailored materials design is available with a flexible 2023-2024 start date. The project is open to candidates with a science Bachelor/Master degree (Chemistry, Physics, Computer Science) and includes a 4-year stipend with full tuition fees. Successful candidates will become members of the interdisciplinary computational research group led by Prof. Reinhard Maurer (www.warwick.ac.uk/maurergroup) based in the Departments of Physics and Chemistry at the University of Warwick, UK.
When chemists search for new functional molecules with tailored properties, they synthetically modify known structural motifs of molecules by trial and error. This empirical exploration of the space of possible chemical compounds forms the present-day bread-and-butter business of chemical innovation. In the last few decades, quantum chemistry has further aided this process by accurately predicting properties in silico, based on formal quantum theory, if provided with molecular structures and significant computational resources. However, certain applications require molecules to satisfy specific properties, for example the emission of light of a specific colour for novel organic light-emitting diode (OLED) materials. Here it might be secondary what the structure of the molecule is as long as it is easy to synthesise. Trial-and-error based on structural modification is unlikely to identify the best and most efficient material; at best, it would be very time consuming. Machine learning methods are ideal to address this inverse property-driven design task. Generative deep learning models are extremely powerful for image generation (deep fake images) and large language models (e.g. ChatGPT). They can also be used to generate new molecules.
The goal of this PhD project is to develop generative machine learning models that can create new undiscovered molecules with tailormade electronic and optical properties for specific applications in OLED materials, organic solar cells, and quantum sensing applications. The student will use existing and generate new data to train generative machine learning models and use advanced techniques to condition and bias such models so that the model predicts molecules with specific narrow properties such as light emission properties or ease of synthesizability. The project will also explore how this approach can be extended to more complex materials, such as molecular crystals, surface nanostructures, and metal nanoparticles.
Successful candidates will join a large, interdisciplinary research group that provides a collaborative and supportive environment. Projects will often involve teamwork and joint problem solving between colleagues with complementary skills. The successful candidate will be trained in state-of-the-art machine learning methodology, electronic structure theory, and molecular simulation methods. The student will acquire important transferable skills such as software development and project management. Substantial resources are available for group members to attend international workshops and conferences.
Interested candidates should contact Prof. Reinhard Maurer (firstname.lastname@example.org).
Type / Role: