Counting qubits for better batteries

The widespread adoption of electric vehicles rests on developing faster charging, longer lasting battery technology – the critical enabler for transitioning away from internal combustion engines. PsiQuantum has been working with Mercedes Benz[1] to assess just how advanced a quantum computer must be to revolutionize Lithium-ion (Li-ion) battery design. By drawing on its unique expertise in Fault Tolerance[2], PsiQuantum has identified novel ways to relax the resources needed by a quantum computer when tackling challenging problems in battery chemistry.

Understanding and improving battery chemistry requires molecular simulation. Modern Li-ion batteries contain a charge-carrying electrolyte, and one goal of battery development is to find additive chemicals that can enhance the battery current provided by the electrolyte. Evaluating potential additives requires accurately simulating how their presence affects electrolyte molecules. The calculations involved in such simulations are forever impossible on a conventional computer, even for modestly sized additive molecules.

Assessing larger molecules with fault tolerant quantum computing
PsiQuantum's team investigated quantum algorithms for simulating effects of the ubiquitous electrolyte additive fluoroethylene carbonate. This meant assessing the physical resources needed to simulate one of the largest molecules (in terms of electronic orbitals) ever considered for quantum computing. The team found that, naively, the problem would require a quantum computer with 16,382 logical qubits[3], able to execute a circuit containing 232 billion T-gates[4]. This is clearly impractical.

Such huge numbers may seem daunting (NISQ[5]-era quantum computers are yet to exhibit a single logical qubit and can only perform hundreds of non-fault-tolerant gates). But the PsiQuantum team found ways to optimize these requirements by compiling the application to a specific hardware architecture - photonic Fusion Based Quantum Computing (FBQC). In this architecture the fundamental hardware units are so-called Resource State Generators (RSGs) - silicon photonic gadgets able to produce small collections of entangled photons on demand. The work with Mercedes Benz found that, without further optimization, an FBQC machine could simulate the effect of fluoroethylene carbonate on battery performance in under a day.

Decidedly not NISQ
This is in stark contrast to most other work evaluating quantum computing’s application to battery chemistry, which focuses exclusively on the performance of current NISQ technology. PsiQuantum’s assessment leveraged their expertise in the million-qubit regime of utility-scale, fault-tolerant quantum computing architectures. This not only allowed the production of realistic resource estimates for problems of practical importance, but also led to surprising ideas for reducing the bottom line in hardware requirements. These ideas required thinking entirely orthogonal to a NISQ-focused mindset.

For example, molecular simulation problems previously assessed for quantum computing have typically involved much smaller molecules. In these cases, a certain fault-tolerance sub-protocol known as magic state distillation was assumed to be the most expensive computational step. However, PsiQuantum showed that this no longer applies in the large molecule regimes relevant to battery chemistry - justifying the parallelization of magic state distillation and suggesting further savings in computational runtime – with possible reductions of 10x or more.

The unique advantages of photonic quantum computing
The study also demonstrated one of the great advantages of photonic quantum computing - how a technique specific to this approach, known as interleaving[6], can be leveraged to trade-off runtime and hardware resource requirements. Interleaving reuses single pieces of hardware multiple times - storing photons output from the hardware in optical fiber until they're needed. By increasing algorithm runtime to 2.5 weeks, interleaving allows fluoroethylene carbonate to be simulated with 20x fewer hardware resources.

This means that PsiQuantum now knows with mathematical conviction that they can enable breakthroughs in Li-ion battery design, by running this optimized algorithm on its utility scale quantum computing architecture.
As the automotive industry ends its reliance on fossil fuels, society will become increasingly reliant on improvements in battery technology. The design of electrolyte additives is just one area of battery design accelerated by fault tolerant quantum computing. From developing novel cathode materials, to multi-scale battery simulation, quantum computing can revolutionize broad areas of battery development. In all cases, only optimized fault-tolerant algorithms running on utility-scale quantum computing devices can enable the breakthroughs sought by automotive manufacturers.

_______________________

[1] I. H. Kim, Y.-H. Liu, S. Pallister, W. Pol, S. Roberts, and E. Lee, “Fault-tolerant resource estimate for quantum chemical simulations: Case study on Li-ion battery electrolyte molecules,” Phys. Rev. Research, vol. 4, no. 2, p. 023019, Apr. 2022, doi: 10.1103/PhysRevResearch.4.023019.

[2] Fault tolerance enables suppression of errors so that a quantum computer can run billions of operations. Understanding the nuances of fault-tolerance is critical in determining how quantum algorithms will really perform when solving commercially useful problems.

[3] Logical qubits are error-corrected, and a single such qubit is built from hundreds or thousands of the kinds of physical qubits available in current generations of quantum computers.

[4] T-gates are one of several universal gates a quantum computer must perform to be able to implement general computations. T-gates are the most expensive gate to perform fault-tolerantly, and so the 'depths' of circuits are often specified only in terms of the number of T-gates needed.

[5] Noisy Intermediate-Scale Quantum (NISQ) computers are devices with qubits too noisy and too few to perform fault tolerant quantum computations. The many attempts to develop useful algorithms under the limitations of the noisy shallow circuits in NISQ devices have proved unsuccessful.

[6] H. Bombin et al., “Interleaving: Modular architectures for fault-tolerant photonic quantum computing.” arXiv, 2021. doi: 10.48550/ARXIV.2103.08612.