Quantum researchers at Ford have just published a new preprint study that modeled key electric vehicle (EV) battery materials using a quantum computer. While the results reveal nothing new about lithium-ion batteries, they do show how more powerful quantum computers could be used to accurately simulate complex chemical reactions in the future.
To discover and test new materials with computers, scientists must break the process into many separate calculations: One set for all the relevant properties of each individual molecule, another for how those properties are affected by the smallest environmental changes such as fluctuating temperatures, another for all possible ways that two molecules can interact together, and on and on. Even something as simple as two hydrogen molecules binding together requires incredibly deep calculations.
But developing materials with the help of computers has a big advantage: the researchers don’t have to carry out all the possible experiments physically, which can be incredibly time-consuming. Tools like AI and machine learning have been able to speed up the research process to develop new materials, but quantum computing offers the potential to do it even faster. For electric cars, finding better materials can lead to longer durability, faster charging and more powerful batteries.
Traditional computers use binary bits – which can be a zero or a one – to perform all their calculations. While they’re capable of incredible things, there are some problems like highly accurate molecular modeling that they simply don’t have the power to handle—and, because of the kinds of calculations involved, may never be able to. Once researchers model more than a few atoms, the calculations become too large and time-consuming, so they must rely on approximations that reduce the accuracy of the simulation.
Instead of ordinary bits, quantum computers use qubits, which can be a zero, a one, or both at the same time. Qubits can also be entangled, rotated and manipulated in other wild quantum ways to carry more information. This gives them the power to solve problems that are intractable with traditional computers – including accurate modeling of molecular reactions. Additionally, molecules are quantum by nature and therefore map more precisely to qubits, which are represented as waveforms.
Unfortunately, much of this is still theoretical. Quantum computers are not yet powerful or reliable enough to be widely commercially viable. There is also a knowledge gap – because quantum computers work in a completely different way than traditional computers, researchers still need to learn how best to use them.
[Related: Scientists use quantum computing to create glass that cuts the need for AC by a third]
This is where Ford’s research comes in. Ford is interested in making batteries that are safer, more energy- and power-dense, and easier to recycle. To do so, they need to understand chemical properties of potential new materials such as charge and discharge mechanisms, as well as electrochemical and thermal stability.
The team wanted to calculate the ground state energy (or normal atomic energy state) of LiCoO2, a material that could potentially be used in lithium-ion batteries. They did so using an algorithm called the variational quantum eigensolver (VQE) to simulate the Li2Co2O4 and Co2O4 gas-phase models (basically the simplest possible form of chemical reaction) that represent the charging and discharging of the battery. VQE uses a hybrid quantum-classical method with the quantum computer (in this case 20 qubits in an IBM state vector simulator) used to solve only those parts of the molecular simulation that benefit most from its unique attributes. Everything else is handled by traditional computers.
As this was a proof-of-concept for quantum computing, the team tested three methods with VQE: unitary coupled-cluster singles and doubles (UCCSD), unitary coupled-cluster generalized singles and doubles (UCCGSD), and k-unitary pair coupled-cluster generalized singles and doubles (k-UpCCGSD). In addition to comparing the quantitative results, they compared the quantum resources required to perform the calculations accurately with classical wavefunction-based approaches. They found that k-UpCCGSD produced similar results to UCCSD at lower cost, and that the results from the VQE methods were consistent with those obtained with classical methods—such as coupled cluster singles and doubles (CCSD) and complete active space configuration interaction (CASCI).
Although it’s not quite there yet, the researchers concluded that quantum-based computational chemistry on the types of quantum computers that will be available in the near term will play “an important role in finding potential materials that can improve battery performance and robustness.” While they used a 20-qubit simulator, they suggest that a 400-qubit quantum computer (soon to be available) would be necessary to fully model the Li2Co2O4 and Co2O4 system they considered.
This is all part of Ford’s attempt to become a dominant electric car manufacturer. Trucks like its F-150 Lightning are pushing the limits of current battery technology, so further advances — likely using quantum chemistry — will be increasingly necessary as the world moves away from gas-guzzling cars. And Ford isn’t the only player looking to use quantum to up the ante in the battery chemistry game. IBM is also working with Mercedes and Mitsubishi to use quantum computers to reinvent the electric car battery.