Professor Stefano Sanvito

Summary

The project aims at investigating the viability of a new spinout company that will develop a suit of natural language processing (NLP) algorithms for the precise extraction of materials properties from scientific literature. This will enable the creation of large datasets of materials properties and, most importantly, their interrelations, enabling the construction of large machine-learning (ML) models for materials design. The extracted data will be of a quality sufficient to enable downstream tasks, such as the construction of machine-learning models for materials design

Funding Agency

Enterprise Ireland

Programme

Feasibility Programme

Summary

Precise Information Extraction through Natural Language Processing for Materials Design and Materials Market Intelligence

Funding Agency

Enterprise Ireland

Programme

Manufacturing, Energy and Food Commercialisation

Summary

In "ML-transport: machine-learning accelerators for quantum transport" we propose to combine machine-learning methods for predicting the charge density in a density functional theory (DFT) calculations with quantum transport theory. This will allow us to achieve a thousand-fold computational speedup, thus making the evaluation of finite-temperature transport possible

Funding Agency

Research Ireland

Programme

Government of Ireland Postgraduate Scholarship

Summary

Spin Transport and Relaxation Dynamics in Single-Molecule Junctions (STARDSMJ) aims at developing a toolbox of computational methods to study spin relaxation of molecular nanomagnets in the presence of charge fluctuations and external bias. This will allow us to investigate the factors affecting the spin relaxation in molecular magnets in realistic environments, namely in the operational conditions of nanoscale devices. Density Functional Theory combined with non-equilibrium Green's functions (NEGF) theory is the ab initio workhorse method for electron transport calculations. This, however, is limited to scattering from a static potential, while dynamical effects are typically included through appropriate self-energies in a perturbative way. In contrast, spin relaxation can be investigated by propagating in time master equations with parameters extracted from accurate electronic structure theory. In this case the molecule remains in a single charging state and inter-molecular interactions are of dipole-dipole nature. The main goal of STARDAM is to construct a formalism that can capture spin relaxation in an environment with fluctuating number of electrons, namely going beyond and unifying the two approaches. I will achieve this goal by developing and applying a robust theoretical scheme to solve the master equation for the transport of single-molecule junctions, where various effective spin Hamiltonians, extracted from first-principles calculations, will be used to describe the molecule.

Funding Agency

EU

Programme

HORIZON-MSCA-2024-PF-01

Project Type

HORIZON TMA MSCA Postdoctoral Fellowships - European Fellowships

Summary

Friction between moving parts and the associated wear are estimated to be directly responsible for 25% of the world's energy consumption. SSLiP seeks to establish a radically new way to drastically reduce friction, with potentially enormous technological and societal impact. The driving concept is structural superlubricity, extremely low friction that takes place at a lattice misfit between clean, flat, rigid crystalline surfaces. Structural superlubricity is currently a lab curiosity limited to micrometer scale and laboratory times. SSLiP will bring this to the macroscale to impact real-life products. The key idea is the use of tribo-colloids: colloidal particles coated in 2D materials, that will produce a dynamic network of superlubric contacts. Structural incompatibility between arrays of colloids allows us to replicate the low friction on bigger length scales and overcome the statistical roughness of real surfaces. Through careful design of these coatings, carrier fluid, and the mechanical properties of the core particles, the chemistry of sliding and collective behaviour of the colloids can be controlled. Synthesis and experiments of individual contacts will be combined with visualisation of colloid dynamics during sliding on larger scales and in-site chemical characterisation. These will be combined with multiscale simulations and theory to bridge the different length scales. The developed ultra-low friction technology will drastically reduce loss of energy, for example in passenger cars (responsible for around 2 billion tonnes of CO2 per year) and increase the lifetime of parts. It will also enable radically new technologies that are impossible with current lubrication.

Funding Agency

EU

Programme

Echorizon