Volkswagen: Quantum Simulation for Next-Gen Battery Electrolytes

The Challenge

Lithium-sulfur batteries promise two to three times the energy density of today’s lithium-ion cells, but their commercialization is blocked by a fundamental materials problem: the polysulfide shuttle. During cycling, soluble polysulfide intermediates dissolve into the liquid electrolyte, migrate to the lithium anode, and cause irreversible capacity loss. Solid-state electrolytes could physically block this shuttle, but designing a solid electrolyte with high enough lithium-ion conductivity while remaining chemically stable against both lithium metal and sulfur is an extraordinarily difficult multi-objective materials design challenge. Classical density functional theory (DFT) captures the ground-state geometry of candidate materials but struggles with the strongly correlated electronic structure of sulfur-containing compounds, leaving large uncertainties in computed ionic conductivity and electrochemical stability windows. Volkswagen’s quantum computing team partnered with IBM to investigate whether quantum simulation could resolve these electronic structure uncertainties.

The Quantum Approach

The team used Qiskit Nature to construct active-space Hamiltonians for lithium polysulfide clusters representative of proposed solid electrolyte interfaces. Variational quantum eigensolver (VQE) calculations on IBM Quantum processors targeted the ground and low-lying excited states needed to compute lithium-ion migration barriers, the key quantity governing ionic conductivity. An adaptive ansatz construction routine (ADAPT-VQE) was employed to keep circuit depth manageable while achieving near-exact active-space energies.

from qiskit_nature.second_q.drivers import PySCFDriver
from qiskit_nature.second_q.mappers import ParityMapper
from qiskit_nature.second_q.algorithms import AdaptVQE, VQEUCCFactory
from qiskit_algorithms.optimizers import L_BFGS_B
from qiskit_ibm_runtime import QiskitRuntimeService, Estimator

# Set up Li2S4 cluster for active-space VQE
driver = PySCFDriver(
    atom="Li 0 0 0; Li 2.9 0 0; S 1.45 1.2 0; S 1.45 -1.2 0; "
         "S 1.45 0 1.8; S 1.45 0 -1.8",
    basis="cc-pVDZ",
    charge=0,
    spin=0,
)
problem = driver.run()

# Reduce to (8e, 8o) active space
problem.active_space = (8, 8)

mapper = ParityMapper(num_particles=problem.num_particles)
qubit_op = mapper.map(problem.second_q_ops()[0])

# Run ADAPT-VQE for compact ansatz
service = QiskitRuntimeService()
backend = service.backend("ibm_sherbrooke")
estimator = Estimator(backend=backend, options={"shots": 8192})

vqe_factory = VQEUCCFactory(estimator=estimator, optimizer=L_BFGS_B(maxiter=500))
adapt_vqe = AdaptVQE(vqe_factory)
result = adapt_vqe.compute_minimum_eigenvalue(qubit_op)

print(f"Ground state energy: {result.eigenvalue.real:.6f} Ha")
print(f"Ansatz depth: {result.optimal_circuit.depth()} layers")

Migration barriers were computed by comparing VQE energies at the stable site and the transition-state geometry for lithium hopping through the electrolyte lattice, with classical nudged elastic band calculations providing the transition-state geometry while VQE supplied the high-accuracy energetics.

Results and Implications

Across eight candidate solid electrolyte compositions screened computationally, VQE calculations predicted two materials with ionic conductivity 15% above the current solid electrolyte state of the art (10 mS/cm at room temperature). Both candidates, lithium-rich argyrodite variants with tailored anion substitution, had previously been ranked low-priority by DFT screening due to uncertainties in the strongly correlated sulfur electronic structure that VQE resolved. The two compounds entered physical synthesis and characterisation at Volkswagen’s Salzgitter battery research centre.

The project demonstrated that quantum chemistry can overturn classical screening rankings in exactly the cases where classical theory is least reliable: strongly correlated, multi-reference electronic systems. For Volkswagen, the business case is compelling: finding even one viable solid-state electrolyte candidate faster than classical methods would accelerate the EV battery roadmap by years and represent billions in competitive advantage. The team is scaling the workflow to screen dozens of compositions per month as IBM Quantum hardware error rates continue to decline.