Pembroke College Cambridge

Chemistry, computers and commerce

‘I vowed three things in my life’ says Professor Mike Payne.

‘I said I’d never become an academic, I’d never work with computers and I’d never do physics.’ Today, he is one of the world’s leading academics in the field of computational physics.

Mike PayneMike is widely recognised as ‘an outstanding scientist and innovator’. In 1996 he won the Maxwell Medal. In 1998 he gave the prestigious Mott Lecture. In 2008 he was elected a Fellow of the Royal Society. He now holds the Chair of Computational Physics at the University of Cambridge. Just last month the Institute of Physics recognised his ‘outstanding contribution’ to the field by awarding him the Swan Medal.

When he arrived at Pembroke in 1978 he had no idea that all of this lay ahead. By 1985 he had completed a BA and an MSc in Natural Sciences, a PhD in condensed matter physics and become a Draper Research Fellow at the College. He says: ‘I liked being at Pembroke, but there’s always a view that if you want to get onto the academic career ladder you can’t just stay in one place.’ He set his sights on a position at Bell Labs, which was ‘the place everyone wanted to be’, but when they wouldn’t take him he ‘ended up’ at MIT instead.

At MIT, Mike was set to work on a computer programme that could simulate systems of atoms. Based on the Schrödinger equation, the aim was to create a piece of software that could simulate the behaviour of particle systems. Importantly, it should be able calculate the responses of different atoms based only on their atomic number or symbol. This had been achieved in 1981, but the programme could only predict the behaviour of silicon (and one or two other elements) and could only handle structures comprising a few atoms. Mike anticipated that his role would involve continuing to gradually increase the software’s capabilities by implementing the well-established method.

In fact, he was to be a part of something far more exciting. Mike still remembers the day his boss returned from a visit to IBM, having heard about a new method of performing quantum mechanical calculations. He believed that this new Car–Parrinello method was the way of the future and called a group meeting to describe the new methodology. Mike was sitting in the back row, only half concentrating. At the end of the meeting his boss declaring: ‘And you’re going to implement it.’ Mike glanced up to see who was being the given the job and, he says, ‘I realised he was pointing at me.’

He continues: ‘So as someone who had vowed never to have anything to do with computers, I was suddenly given the job of writing a complete set of computer codes to implement the new methodology.’ He did just that and wrote a ‘first principles total energy pseudopotential code’, now known as CASTEP. Mike’s work meant that rather than simply making predictions for a few atoms of silicon, the computer could simulate the behaviour of tens of atoms of silicon and many other elements. This proved to be the tipping point.

Mike says: ‘Within a decade we were simulating 400 atoms of silicon and hundreds of atoms of whatever you like. It was mainly my work that did that. That’s when people really started to take notice.’ Mike also managed to simplify the calculations themselves so that he could give the software to a chemist or an experimental geologist who knew nothing about condensed matter physics. Crucially, they no longer needed to be an expert in the maths behind the software to make the calculation work. All they needed to do was enter the atomic numbers of the structures they were working on and the computer would perform a predictive simulation.

By the late 1980s, Mike was running a large computational group at the University of Cambridge. This inevitably meant that he stepped back from research in order to manage the administrative work. However, he also had other new responsibilities; the next major step for CASTEP was commercialisation. A local company now known as BIOVIA approached Mike with an offer. In 1995, after 18 months of negotiation, they licensed the code and linked it to their user-friendly graphic interface. This made the software even easier to use. Mike explains: ‘What they did was develop the interface so that you could just click with the mouse – an oxygen atom here, a silicon atom there – and get a predictive quantum mechanical calculation done. It was so simple. ’ The rest, as Mike says, is history.

CASTEP

At the time they acquired the code, BIOVIA anticipated that the resulting software would have a shelf life of ten years and a market size of around £300,000 per year. Their powers of prediction were not as powerful as those of CASTEP.

By 1998, the product had exceeded sales of £1 million per year. Mike himself played an important role in the sales process. He says: ‘I spent a lot of time going and talking to customers in industry and doing sales tours. We all had to put in a huge amount of effort.’

The software was later re-written in such a way that it could be more easily updated and expanded. From 2000-2003, five students and post-doctoral colleagues in Mike’s group worked with some of his collaborators to re-engineer the code from scratch during their spare time. He explains: ‘They’ve completely rewritten it – so I did not write a single line of code that is in the version of the software currently being sold.’ Since then, the technology has steadily increased in its range and capability. This has made it invaluable for many branches of academic and industrial science. Today it has cumulative sales of over £20 million. ‘The sales are not going down,’ Mike adds, ‘and the platform is still going strong. It’s an absolute triumph.’

Mike has achieved industrial and academic success simultaneously. So many papers are written each year that make reference to the underlying methodology Mike developed that they have to be measured in a new unit: ‘kilopapers’. This is partly because Mike has worked hard to show its relevance for a range of scientific disciplines including chemistry, geology and biology. He explains: ‘The good thing about this tool is that you can do so many different things with it. To perform multiple experiments on a material you’d need a lot of different machines. Our software is a one-size-fits-all solution.’

Recent reports estimate that companies who invest in CASTEP software receive a seven-fold return on their investement. ‘If you believe those reports,’ says Mike, ‘that means that our computer code has had over £100 million impact worldwide.’ It’s an incredibly achievement – and all the more for a man who vowed never to work with computers.

Latest tweets

Pembroke College Cambridge