Panasonic Quantum Simulation for Solid-State Battery Electrolyte Discovery

The Problem

Solid-state batteries replace the liquid electrolyte in conventional lithium-ion cells with a solid ionic conductor. This eliminates the flammability risk from organic solvents, enables lithium metal anodes (doubling theoretical energy density), and allows thinner cell formats for automotive packaging. Panasonic, as the primary battery supplier for Toyota and a major supplier for Tesla, has a direct commercial stake in the transition: solid-state cells would allow next-generation EV platforms to achieve 500+ mile range with improved safety profiles required for mass-market adoption.

The bottleneck is electrolyte ionic conductivity. Argyrodite-class sulfide electrolytes (Li6PS5Cl and related compositions) achieve room-temperature conductivities of 1 to 3 mS/cm in their best forms, still below the 10 mS/cm threshold needed for high-power automotive discharge rates. The rate-limiting step is the Li-ion hopping barrier between adjacent tetrahedral and octahedral sites in the argyrodite lattice. Reducing this barrier by 0.1 to 0.15 eV through anion substitution (replacing Cl with F, Br, or mixed halide) is the primary design lever, but DFT with standard GGA functionals gives unreliable barrier heights for Li-S-P coordination environments due to self-interaction error in sulfur d-states.

Quantum Chemistry of Li-Ion Migration

The Li-ion migration pathway in Li6PS5Cl involves the Li atom moving through a shared face between adjacent S4 tetrahedra. The transition state geometry places Li in a distorted 4-coordinate S environment with partial occupancy of sulfur p-orbitals. The active space for the quantum chemistry calculation focuses on the S-Li coordination shell: the relevant electrons are the sulfur lone pairs and the Li 2s contribution to the bonding interaction. For the Li6PS5Cl active site cluster model, the active space is 8 electrons in 8 spatial orbitals, mapping to 16 qubits under Jordan-Wigner encoding.

PySCF was used to run CASSCF natural orbital analysis on a 22-atom cluster model of the argyrodite active site, extracted from a periodic DFT geometry optimization. Orbitals with fractional natural occupancies between 0.05 and 1.95 were selected for the active space. The cluster boundary was capped with pseudohydrogens. The Qiskit Nature PySCFDriver interface handled conversion from the PySCF integral output to the second-quantized Hamiltonian used by VQE.

from pyscf import gto, scf, mcscf
from qiskit_nature.second_q.drivers import PySCFDriver
from qiskit_nature.second_q.mappers import JordanWignerMapper
from qiskit_nature.second_q.circuit.library import UCCSD, HartreeFock
from qiskit_algorithms.minimum_eigensolvers import VQE
from qiskit_algorithms.optimizers import SLSQP
from qiskit_aer.primitives import Estimator
import numpy as np

# Argyrodite cluster: Li in transition state S4 coordination
# 22-atom cluster, 8-electron 8-orbital active space
argyrodite_ts_geometry = """
Li   0.000  0.000  0.000
S   -1.421  1.421  0.000
S    1.421  1.421  0.000
S   -1.421 -1.421  0.000
S    1.421 -1.421  0.000
P    0.000  3.500  1.750
P    0.000 -3.500  1.750
"""

driver = PySCFDriver(
    atom=argyrodite_ts_geometry,
    basis="def2-TZVP",
    charge=1,  # Li+ migrating ion
    spin=0
)

problem = driver.run()
problem.num_particles = (4, 4)       # 8 electrons: 4 alpha, 4 beta
problem.num_spatial_orbitals = 8     # 8 orbitals -> 16 qubits JW

mapper = JordanWignerMapper()
qubit_op = mapper.map(problem.hamiltonian.second_q_op())
print(f"Argyrodite TS qubit count: {qubit_op.num_qubits}")  # 16

hf_state = HartreeFock(
    num_spatial_orbitals=8,
    num_particles=(4, 4),
    qubit_mapper=mapper
)
ansatz = UCCSD(
    num_spatial_orbitals=8,
    num_particles=(4, 4),
    qubit_mapper=mapper,
    initial_state=hf_state
)
print(f"UCCSD parameter count: {ansatz.num_parameters}")

NEB-DFT Comparison and Fluorine Substitution

The classical reference calculation used the Nudged Elastic Band (NEB) method with DFT (PBE+D3 functional, VASP, plane-wave basis) to trace the full Li-ion migration path in the periodic argyrodite structure. NEB with 9 image points gave a migration barrier of 0.31 eV for Li6PS5Cl. The same NEB protocol applied to Li6PS5F (full fluorine substitution on the Cl site) gave 0.27 eV, a modest 0.04 eV reduction, not enough to reach the conductivity target.

VQE on IBM Heron 133Q, using UCCSD with SLSQP optimization and 8192 shots per expectation value, computed the cluster-model barrier for both Cl and F variants. For Li6PS5Cl the VQE barrier (0.29 eV) agreed well with NEB-DFT. For Li6PS5F the VQE result diverged substantially: the fluorine-substituted variant showed a barrier of 0.15 eV, compared to NEB-DFT’s 0.27 eV. The 0.12 eV difference traces to the stronger electronegativity of F polarizing the S4 coordination cage, reducing the activation barrier through an electrostatic effect that GGA functionals underestimate due to delocalization error.

estimator = Estimator(
    backend_options={"method": "statevector"},
    run_options={"shots": 8192}
)
optimizer = SLSQP(maxiter=500)
vqe = VQE(estimator, ansatz, optimizer)

# Ground state: Li at stable tetrahedral site
result_gs = vqe.compute_minimum_eigenvalue(qubit_op_ground_state)
e_ground = result_gs.eigenvalue.real

# Transition state: Li at saddle point
result_ts = vqe.compute_minimum_eigenvalue(qubit_op_ts)
e_ts = result_ts.eigenvalue.real

barrier_ev = (e_ts - e_ground) * 27.2114  # Hartree to eV
print(f"VQE migration barrier: {barrier_ev:.3f} eV")

# Ionic conductivity scales as exp(-Ea / kT)
import scipy.constants as const
T = 298.15  # K
kT_ev = const.k * T / const.e  # ~0.0257 eV
conductivity_ratio = np.exp(-barrier_ev / kT_ev)
print(f"Relative conductivity (vs 0.31 eV baseline): {conductivity_ratio:.2f}x")

Outlook for Solid-State Commercialization

The VQE result redirected Panasonic’s synthesis program. Rather than optimizing Li6PS5Cl with partial Cl/Br substitution (the direction suggested by NEB-DFT), the quantum result pointed to full F substitution as the higher-payoff target despite the known challenges of fluoride precursor handling in argyrodite synthesis. Panasonic’s materials lab initiated a small-batch synthesis campaign for Li6PS5F using mechanochemical sintering, with ionic conductivity measurement by electrochemical impedance spectroscopy as the validation step.

Reaching 10 mS/cm at room temperature would place argyrodite-F electrolytes in the performance range of liquid organic electrolytes, removing the last major conductivity barrier to solid-state cell assembly. Panasonic’s production roadmap for solid-state automotive cells targets 2027 to 2028, contingent on electrolyte conductivity and interface resistance solutions. The quantum simulation result does not guarantee experimental success (synthesis defects and grain boundary resistance are separate challenges) but it provides a higher-confidence target than DFT alone.

Learn more: Qiskit Reference