Perfect for designing new molecules

Even a small quantum computer should be able to simulate exactly the properties and behaviour of new molecules. This would take chemistry to an entirely new level. Better solar panels, more powerful batteries, saving lots of energy in the chemical industry: the applications have the potential to transform society.

Only some bacteria, not plants, have mastered the trick of nitrogen fixation: taking nitrogen gas (N2) out of the air and, with water, turning it into ammonia (NH3). This is a basic nutrient, directly or indirectly, for all life on Earth. Like all biological processes, nitrogen fixation is made possible by a specific enzyme, the core of which is a molecule called FeMoco.

Simulation

Humans make 175 million tonnes of ammonia each year from atmospheric nitrogen and natural gas, mainly to produce fertiliser. But this chemical process only works at very high pressure and temperature. If we understood exactly how FeMoco worked, we could design our own version of that molecule to do industrial nitrogen fixation at atmospheric pressure and room temperature, saving lots of energy and CO2-emissions.

That is why, in 2016, Swiss quantum chemists proposed FeMoco as a first example of a molecule to be simulated with a quantum computer. Thomas O'Brien, assistent professor of theoretical physics at LION (Leiden Institute of Physics) is now working on that: ‘This application alone would be worth billions of euros. It's the 'killer application' for quantum computing.’

Simulating a large molecule means calculating the dynamical interaction between all the electrons that have freedom of movement between the atoms that make up the molecule. In fact, the molecule sticks together because of this interaction, and its chemical properties are determined by it.  

This socalled electronic structure problem is impossibly complicated for classical computers, even when the molecule consists of only a few atoms.

Ideally suited

But this kind of calculation is ideally suited for quantum computation. According to O'Brien, you need, loosely speaking, about as much qubits as there are interacting electrons in the molecule. For FeMoco that number is around 120. Protoype quantum devices with a few dozen qubits are in operation already. O' Brien and his group are presently collaborating with QuTech, a large consortium of quantum physicists at Delft Technical university, to explore how such small quantum systems can be used for quantum chemistry. 

These small quantum devices don't have enough qubits yet to function as a universal quantumcomputer. That would take many thousands of qubits, and the first one probably will be built in about ten years. But because of the special properties of the electronic structure problem, O'Brien and his collaborators at the Leiden Institute for Chemistry and the VU Amsterdam expect to make progress much faster: 'We have already designed prototype algorithms and implemented them on real quantum hardware with collaborators at TU Delft.'

Prototypes

If they succeed in simulating the FeMoco-molecule, there's no reason why they couldn't simulate many other molecules which are extremely relevant to society. Once you know exactly how sunlight makes electrons run through silicon and other semiconductors, there's a fair chance you can design more efficient solar cells. The same goes for catalysts that direct chemical reactions, or useful biomolecules. O'Brien: ‘I'm sure other things will appear as well, the ones that nobody thought about yet.’