Dynamic nanocluster – biomolecule interfaces

Introduction

Metal nanoclusters (MNC) are atomically precise metal nanoparticles with definite mass, structure, and chemical composition. Their metal core of 1-3 nm in diameter exhibits a quantized electronic structure, and they are chemically stabilized by a molecular surface layer which is modifiable for functionalization and optimized biocompatibility. This makes them promising materials for novel applications in bioimaging, biosensing, and nanomedicine as fluorescent markers, sensors, and targeting drug carriers, paving the way to personalized medicine and therapeutics.

Significance of Size

Their ultrasmall size makes them amenable to atom-scale modeling, which may greatly help experimental efforts to design their properties for applications. However, consolidated simulation strategies capable of dealing with phenomena over a wide range of dynamical processes at the nanocluster – biomolecule interfaces are missing.

Project Objectives

To address this need, DYNANOINT will develop new multiscale simulation strategies, assessing critically graph theory and machine learning (ML) methods as potential accelerators for the discovery of structure-function relationships. This will be combined with traditional electronic structure methods and force-field based dynamical simulations.

Methodology

The methodology is applied to:

1. Weak chemical interactions between MNCs and proteins.
2. Electronic excitations and charge-transfer interactions at the MNC – environment interface.
3. Chiral MNC – environment interfaces.

Collaboration and Impact

The Principal Investigator’s extensive collaboration network with three key experiments around the world ensures efficient spread of the impact of this theoretical-computational project to real-life applications.

Broader Impact

The project has a broad impact on computational nanoscience since the developed open-source software will provide new tools to study structure-function relations in low-dimensional, low-symmetry nanostructures. The combination of robust, transferable, and interpretable ML models with traditional simulation methods poses a significant contemporary challenge.