Use this URL to cite or link to this record in EThOS:
Title: Simulating self-assembly of organosulfur species on gold nanoparticles
Author: Serapian, Stefano
ISNI:       0000 0005 0732 542X
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Access from Institution:
This Thesis aims to establish an accurate but computationally effective method for simulating self-assembly of organosulfurs on gold nanoparticles (AuNPs), a process resulting in their functionalisation. A second gold rush is currently rekindling chemists' interest in the synthesis of novel functionalised AuNPs: these can bear the most chemically diverse functional groups, making them employable in a wide variety of applications, from optoelectronics to catalysis. Some aspects of self-assembly remain experimentally unclear at the mechanistic and electronic levels: achieving its accurate reproduction in silico would indeed represent an important contribution in the synthesis of functionalised AuNPs. This task, however, has so far proven difficult to achieve. In this work, I set out and review four fundamental challenges facing the computational chemist aiming to simulate self-assembly, and describe the strategy chosen to overcome them, using thiols (RSH) as the reference organosulfur. These challenges involve proper reproduction of: I) gold's relativistic effects and aurophilicity; II) the extensive surface reconstruction occurring upon self-assembly, with formation of RS-Au-SR staples and hydrogen loss; III) the large scale ligands involved in the process and their interactions; and IV) the fluctuating solvent environment in which it occurs. Confined to the AuNP core and RSH headgroups, challenges I and II involve complex electronic properties and entail electronic change, with bonds being cleaved (S-H) and reformed (S-Au, possibly H-H): overcoming them requires explicit simulation of electrons with a QM method (DFT). Challenges III and IV involve the entire RSH-AuNP system, including RSH tails of typically ~10^2 atoms: QM methods become impracticable at these system sizes, and a less costly classical forcefield treatment (MM) is necessary in this case, at least in part. The work presented here then proceeds towards the stated aim by attempting to resolve each of these challenges I-IV. The eventually devised solution proposes a combination of classical molecular dynamics (MD), followed by the hybrid QM/MM method ONIOM, which allows to combine the 'best of the QM and MM worlds' and is well established for other systems. To overcome challenge I, various effective core potentials (ECPs); basis sets; and density functionals are evaluated based on their ability to predict properties and geometries of several pristine AuNPs. These properties and geometries are either derived experimentally, or from high-level ab initio calculations. The chosen QM method PBE/LANL2DZ is then further tested on various systems, assessing its ability (challenge II) to reproduce hydrogen loss and staple formation. Upon proposing to tackle challenge III using ONIOM (with the OPLS-AA forcefield for the MM part), the method's performance is first compared to that of full QM (PBE/LANL2DZ) in terms of accuracy and efficiency, and in a variety of contexts, including on AuNPs featuring a 38-atom gold core. Once these calculations confirm the considerable time gains afforded by the introduction of ONIOM, I then demonstrate its full applicability in the optimisation of a large, experimentally plausible functionalised AuNP. Finally, I propose to tackle challenge IV by introducing a classical MD simulation stage to precede QM/MM optimisation. As a test, MD is used to generate statistically significant sets of 8-atom AuNPs coated with alkylthiols of different chain lengths, which are then optimised, thereby successfully reproducing the early stages of reconstruction. I then conclude by successfully testing this MD + ONIOM approach on two much larger functionalised AuNPs, having 20-atom gold cores and sixteen or seventeen 64-atom ligands. My Thesis highlights both the strengths and limitations of the ONIOM approach in simulating such a complex process as organosulfur self-assembly on AuNPs. Nonetheless, the chosen MD + ONIOM strategy can indeed reproduce key aspects of self-assembly with increased CPU-efficiency, and, importantly, makes electronically plausible predictions: it therefore represents a viable route for the in silico investigation of this process, and an encouraging fulfilment of my initial aim.
Supervisor: Bearpark, Michael; Bresme, Fernando Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral