The steady research into developing real-world applications for quantum computing is piling up interesting use cases. Today, IBM reported on work with Boeing to simulate corrosion processes to improve composites used in airframes. In this instance, battling rust or corrosion is the goal.
A paper on the work was published last month in Nature’s online npj Quantum, and IBM has posted an interview today with Nam Nguyen, an applied mathematician at Boeing; Kristen Williams, associate technical fellow and manager of the applied math computational methods research group at Boeing; and IBM Research senior research scientist Mario Motta to discuss their collaboration.
Here’s a quick excerpt from the Q&A:
Williams: Corrosion happens in the presence of a corrosive electrolyte. And for the aerospace industry, that is predominantly thin films that form on the surface of vehicles whenever they are operating in a humid environment. Anywhere you have high humidity, or instances of the humidity cycling between dry and wet conditions, you can form those films on the surfaces. And if the surface is unprotected, that can kick off corrosion.
Motta: Whether you’re talking about an aircraft, the hull of a ship or something else — sooner or later it’s going to be degraded by the interaction with the environment, and you won’t be able to use it anymore. So, the goal is for scientists to do calculations to characterize corrosion in existing materials, and eventually to propose a new material that is more resilient to corrosion than what we have today. But the first step is to understand what is happening in the existing materials.
The work was done on a combination of IBM quantum device and classical systems. As noted in the posted IBM article, the researchers developed two new techniques to perform quantum simulations of a key step in the corrosion process known as water reduction.
IBM reports, “[B]ecause quantum computation is particularly well-suited to modeling quantum-scale systems, the researchers were able to compute the energies involved in the water reduction reaction much more accurately than a leading classical method. The researchers also devised a potentially valuable new method for the exact and automated simplification of quantum circuits — significantly reducing the quantum resources required to run their simulations. They say their circuit simplification method may have a wide array of applications beyond their specific simulation experiments.”
Here’s a bit more from the interview:
Williams: Our aim was to develop a very precise description and understanding of the kinetic rates of the reactions that drive corrosion. So we’re really zooming in to the microscopic level, and looking fundamentally at how these water films react on the metal surface after they form. Our intent is to be able to model and understand those reactions very precisely with the quantum computer.
Motta: We wanted to take a step of the corrosion reaction — in this case, the splitting of a water molecule on a magnesium surface — and investigate whether it could be effectively simulated with quantum computers. This process, which we call water reduction, is the initiator of a chain of corrosion reactions, so simulating it is an important step towards the simulation of the full corrosion reaction.
The researchers propose a new algorithmic workflow to simulate reactions on surfaces on quantum computers. The proposed workflow comprises an embedding method specifically designed for reactions of molecules on surfaces, and a circuit simplification technique to facilitate experiments on near-term quantum devices. The figure below from the paper shows key pieces of the work which involved classical preprocessing and quantum simulation.
In describing the work, the researchers write, “First, we develop and compare two methods to rank and select active-space orbitals based on (i) their contribution to the difference between the DFT electronic density of the system and the superimposed DFT electronic densities of the constituent surface and adsorbate and (ii) their effect on the ground-state active-space energy. Second, we solve the Schrödinger equation in the active space using the variational quantum eigensolver.
“To achieve this goal, it was necessary to evaluate the expectation value of the active-space Hamiltonian over a quantum circuit. We simplify and economize this operation by employing the algebraic properties of Clifford transformations. This allows for the construction of an equivalent circuit with fewer qubits and gates, and lower depth compared to the original one. We illustrate the proposed workflow on a step in the corrosion reaction of magnesium by water. We discuss the underlying approximations and assess their impact on the accuracy of the computed properties. Finally, we demonstrate the proposed workflow using IBM’s quantum hardware.”
The Q&A provides a readily accessible overview and the paper, naturally, digs deeper.
In the interview, Williams said, “We were using the quantum devices to compute energy at a very fundamental level. For the most part, when chemists do these types of calculations today, they use tools like density functional theory (DFT), which require a lot of approximation so they can run on classical computing hardware. DFT is really a workhorse of the field, including in industry, but we’ve known for a long time that it is deficient in some areas — including accuracy in predicting chemical kinetics.
“In this paper, we actually showed that if we take this very fundamental equation involving water, and we recompute the energies with quantum hardware, we get energies that are more accurate than DFT. The DFT method has been used to study this same reaction in hundreds of other papers. So to me, that was very significant because it shows that you do need that quantum description to be able to study this reaction very precisely.”
Link to IBM article and Q&A, https://research.ibm.com/blog/boeing-quantum-corrosion
Link to npj Quantum paper, https://www.nature.com/articles/s41534-023-00753-1.epdf?sharing_token=BaGL3KAtVP7f2AUqc5YUt9RgN0jAjWel9jnR3ZoTv0MzxNIuj-XaUgEeODtnNwbH8eQ8KiuHwe9jF0E92UsLMTmq3czaRrlP5MEW3FfMjg-TfprHsLMhZJsHXrkqEMBoG5S27SIVjiPPJyV5C2hJ_qpZHVMSR8IqQvEgctlqy9s%3D