Nano-confined phase change materials
Materials properties can change significantly, if the dimensionality is reduced. In this project, we are studying crystals encapsulated in single-wall carbon nanotubes, which are only a few nanometers across. This is essentially a one-dimensional crystal. Recent work at the University of Warwick (J. Sloan) has demonstrated the possibility of these wires undergoing transitions between nano-crystalline structures with markedly different properties, in response to bending strain in the CNT. These "phase change" properties open the way for nanoscale electromechanical switches and non-volatile memory, as well as providing a playground for fundamental studies of phase changes at the smallest length scale possible in a material.
In an EPSRC-funded project, we are aiming to enhance modelling capabilities for those materials, to aid in interpretation of experiments, understand the origin of the phase change behaviour, and guide our experimental colleagues toward compounds with potentially advantageous properties. I am particularly interested in the long-term evolution of metastable states, which is essential if those structures are to be used in memory devices.
Further information about this project can be found on the project website.
Development of effective interaction potentials from ab-initio data
In principle, the atomic forces in a system can be determined with ab-initio methods to an impressive level of accuracy from the positions and atomic number alone, however these methods are limited to a few hundred particles. Larger simulations require effective potentials. With the force matching method, the power of ab-initio calculations can be used to determine those effective potentials: The parameters of the interaction potential are adjusted to optimally reproduce forces, energies and stresses from first-principles calculations.
The program potfit has been developed to determine pair and EAM potentials from ab-initio data. This program is enhanced within the framework of the collaborative research centre to include oxide potentials (see below). Additional improvements are the optimisation of analytic potentials, angular dependent potentials and electron-temperature dependent potentials.
The functionality for potfit has been extended to operate on interatomic potentials form the Open Knowledgebase of Interatomic Models (OpenKIM), a cyberinfrastructure for improving the reliability of molecular and multiscale simulations of materials. With the potfit/OpenKIM integration, parameters of interatomic models from OpenKIM can be optimised in potfit, and the resulting potentials validated using tests provided by OpenKIM. OpenKIM then facilitates the integration for these potentials in major simulation codes. For further details visit the OpenKIM Website.
Simulation of graphene growth on metal substrates
Graphene growth by chemical vapour deposition (CVD) on copper foil has emerged as one of the most promising routes for large-scale production of high-quality graphene films. The electronic properties of the graphene sheet can then be modified in a controlled fashion via oxygen or nitrogen functionalization in ultra-high vacuum. Modelling the electronic properties of this system is however challenging: Low impurity atom doses (a few percent) and a large lattice mismatch between graphene and substrate require large unit cells in first-principles calculations.
We developed an add-on tool called bs_sc2pc to the density functional theory code CASTEP, which allows to extract meaningful information about the effective band structure from those large supercell calculations. We use this capability to study the effects of oxygen and nitrogen functionalisation and the interaction between graphene and copper substrate.
- P. Brommer and D. Quigley, Automated effective band structures for defective and mismatched supercells. J. Phys.: Condens. Matter 26, 485501 (2014).
- A.J. Marsden et al., Effect of oxygen and nitrogen functionalization on the physical and electronic structure of graphene. Accepted for publication in Nano Research (2015).
Kinetic ART simulations of bcc iron
Many structural relaxation processes (like for example radiation damage recovery) happen on a timescale that is inaccessible to molecular dynamics simulations. When the dynamics is dominated by rare atomic diffusion events associated with high-energy barriers (compared to temperature), kinetic Monte Carlo (KMC) simulations can be used reach the relevant simulated times. In complex systems, it is however challenging to assemble a catalog of possible events a priori, which has hindered a wide-spread adaptation. The off-lattice, self-learning KMC code k-ART (kinetic Activation-Relaxation technique) uses a topological classification scheme with an ART nouveau, an unbiased saddle point search method to locate the relevant barriers.
KMC simulations are plagued by curse of low barriers. If barriers exist between a group of states that are significantly lower than those leading to other states, then the method spends most of its time uselessly jumping between those states over and over again. We developed the basin-autoconstructing mean rate method, which efficiently groups such states together and allows the simulation to explore other areas of configuration space. We used k-ART to study vacancy diffusion in iron, where we showed that vacancies cluster to form nanovoids on timescales of under a second at room temperature.
- L.K. Béland, P. Brommer, F. El-Mellouhi, J.-F. Joly, and N. Mousseau, Kinetic activation-relaxation technique. Phys. Rev. E 84, 046704 (2011).
- P. Brommer, L.K. Béland, J.-F. Joly, and N. Mousseau, Understanding long-time vacancy aggregation in iron: A kinetic activation-relaxation technique study. Phys. Rev. B 90, 134109 (2014).
Scalable Molecular Dynamics simulation of long range interactions in oxides
Oxidic materials are dominated by Coulomb and dipolar forces, which decay only slowly with distance. This makes Molecular Dynamics (MD) simulations complicated and tedious. Traditional methods, like the Ewald summation scale as N3/2. For very large particle numbers N, this method cannot keep up. Other techniques (multipole or grid methods) do scale better, but they cannot easily be incorporated into existing MD codes. Additionally, the charge distribution in oxides might induce dipoles, whose strength must be determined in some way.
This problem was addressed in subproject B.1 of the collaborative research centre 716 at the University of Stuttgart. To this end, the so-called Wolf summation is used, which scales proprortional to the number of particles. The dipoles are determined self-consistently according to the model introduced by Tangney and Scandolo. This allows the treatment of system with large particle numbers. Later, this model was implemented in the force matching code potfit, which allows the determination of potentials for polarisable materials.
- P. Brommer, P. Beck, A. Chatzopoulos, F. Gähler, J. Roth and H.-R. Trebin, Ab initio based polarizable force field generation and application to liquid silica and magnesia. J. Chem. Phys. 132, 194109 (2010).
- P. Beck, P. Brommer, J. Roth and H.-R. Trebin, Direct Wolf summation of a polarizable force field for silica. J. Chem. Phys. 135, 234512 (2011).
Dynamic processes in complex metallic alloys
Some complex metallic alloys (CMAs) have interesting material properties, like hardness combined with low density, or low thermal conductivity together with high electric conductivity. Molecular dynamics (MD) simulations can be used to study and understand these properties. Especially phonon dispersion relations might help here.
In the Mg–(Al,Zn) system, there are alloys in various degrees of complexity up to quasicrystals. This system is therefore ideal to study the influence of increasing complexity on the vibrational properties. For the relatively simple Laves phase MgZn2 the dynamical structure factor (from which the phonon dispersion can be determined) was calculated using MD methods and compared to data from neutron scattering experiments.
- P Brommer, M. de Boissieu, H. Euchner, S. Francoual, F. Gähler, M. Johnson, K. Parlinski and K. Schmalzl, Vibrational properties of MgZn2. Z. Kristall. 224, 97–100 (2009).