Next Cafe

[cafes...@gmail.com]

Café Scientifique, Sheffield

3 April 2017 7.00 pm

 

From Bits to Batteries: Simulating Novel Materials

 

Details

 

As we increasingly require greener energy and more efficient power sources, much research is being done on photovoltaic materials and novel battery materials based upon solid oxide ion conductors, which do not rely on a liquid electrolyte to carry charge. In particular, the objective is to be able to make solar panels thinner and batteries smaller. Chris will talk about his work within computational materials science, where he develops next generation forcefields, in particular utilising ‘machine learning’, in which the computer program can learn from vast amounts of pre-generated quantum mechanical simulations. The end goal is the discovery of new and efficient materials using models that have quantum mechanical accuracy, but cost a fraction of the computer time. Chris will explain one method of ‘machine learning’ that is generating interesting results, artificial neural networks, whereby a program mimics the way our brain neurons interact to learn relationships between structure and properties.

Dr Christopher Michael Handley joined the University of Sheffield Department of Material Science and Engineering, under the SUbST grant, working with both the MESAS (Multiscale Engineering and Science Simulations at Sheffield) and the Functional Ceramics Group, in May 2015. His work focuses on the design of novel forcefields – a method of modelling chemistry using classical mechanics, rather than more expensive quantum mechanical simulations - for the simulation of photovoltaic materials, and solid oxide ion conductors, which have applications for fuel cells. Chris’ work, in conjunction with synthetic material scientists, aims at making solar energy cheaper and more available. Chris uses modern computing power to develop next generation forcefields and has focused primarily on the use of ‘machine learning’, in which the computer program can learn from vast amounts of pre-generated quantum mechanical simulations, with the end goal of discovering forcefields that give quantum mechanical accuracy, but for the fraction of the computer time. This is an example of “slow learning, fast emulation”, enabling an accurate model that can be used for a variety of simulation tasks that direct quantum mechanical simulations are not possible.
http://www.sheffield.ac.uk/materials/staff/research/chrishandley