###AI-powered molecular thermodynamics of solvated magnetite nanoparticles for the development of new materials
Project ID: 2228bd1145 (You will need this ID for your application)
Research Theme: Advanced Materials
UCL Lead department: Chemical Engineering
Lead Supervisor: Marcello Sega
Project Summary:
Magnetite nanoparticles have numerous environmental and biomedical applications, including water decontamination and hyperthermia cancer treatment, with the potential to improve the quality of life for future generations.
A better understanding of the properties of the water/magnetite interface at the microscopic level is crucial for improving the stability of nanoparticle suspensions in water, which is essential for its use in biomedical applications. However, modelling the hydrated surface of magnetite using traditional methods at the atomistic level is challenging due to the presence of chemisorbed and physisorbed water molecules forming a hydrogen-bonded surface network. To date, there is no accurate molecular-mechanical model that can describe the interface region realistically, and accurate quantum chemical calculations are computationally too expensive to simulate the fully hydrated system in thermodynamic equilibrium. This is where deep learning comes in.
In this project, you will work with a team of experts at the UCL chemical engineering department to generate predictive models of atomic interactions using deep neural networks. This approach has proven successful in providing accurate results for various substances, including large interfacial systems. Your specific tasks will include extending the quantum mechanical calculations of the magnetite surface already performed by the group and creating a database to train the neural network with high-accuracy forces and energies. After training the model, you will use it to run molecular dynamics simulations and gain valuable insights into the interfacial water structure, dynamics, and thermodynamics, learning skills that will be helpful far beyond the field.
We are seeking candidates with an interest in becoming experts in AI-based methods for molecular thermodynamics. The successful applicant must hold a minimum 2:1 MEng/MSc or equivalent degree in Chemical Engineering, or other areas pertinent to the project. Prior experience in molecular dynamics, quantum chemistry, or machine learning is desirable.