Modern computers rely on electronics and binary logic — the familiar ones and zeros. Baltussen: 'That works fine for today's applications, but with the rise of technologies like AI, our energy consumption is skyrocketing.’ While alternatives like quantum computers and neuromorphic computers (brain-inspired) are being explored, Baltussen proposed a different concept: building a computer based on chemical reactions.
Bacteria and molecules
‘That has been tried before’, Baltussen explains. ‘Scientists tried to mimic electronic computers and the transistors we use in them, i.e. the ones and zeros, with molecules. But this proved incredibly difficult: molecules don’t fit well into a transistor.’ The chemist therefore drew inspiration from nature, specifically how bacteria process information. ‘A bacterium searches for food by scanning tis surroundings for things like oxygen. It does through chemical reactions. That’s a form of information processing, a kind of calculation – and that is basically what happens in a computer too.’
Reservoir computers
Looking into chemical processes available in the lab, Baltussen focused on the Formose reaction: a chemical reaction that is thought to be at the basis of the origin of life on Earth. In this reaction, two sugar molecules react, creating a whole mixture of new, complex molecules.
He came up with the idea (‘It was a real Eureka moment’) of using a concept from neuromorphic computers: a reservoir computer. Put simply, a reservoir computer is a method that assumes that a complex system that performs complicated calculations already exists. All you have to do is extract the ‘correct calculations’. ‘My hypothesis was that the chaotic complexity of a Formose reaction could act a reservoir computer: a system that already performs complex processing. So it already exists, you don't have to do anything. You just have to learn how to interpret it.’
To test this, Baltussen simulated the behaviour of E. coli bacteria using a computer model. He altered the conditions of the bacteria, for example by adjusting available nutrients, and did the same with a Formose reaction. ‘By aligning the two, I trained the chemical system, the Formose reservoir, to predict how the bacteria would behave in different situations.’
It worked. Baltussen succeeded in predicting a basic weather patterns in this way. ‘It was amazing that my idea actually worked. It wasn't an obvious outcome, so that made it even more exciting.’
Potential for the future
‘The big advantage is that far less less energy and data are required. While chemical computers, won't replace digital ones, they may, like quantum computing and neuromorphic computing, serve a similar role in a data centres handling specialized tasks.’
Baltussen also sees future applications nanotechnology and biomedical science, such as smart body sensors. ‘But that's all still a long way off. A lot more research is needed first – and with it, a lot more funding – to move from concept to practical implementation.’