experiments/degradation.py:measure_frozen_ion_snapshot measures the electronic steady-state J-V curve at a supplied, fixed ionic configuration. This is a separation of electronic and ionic time scales, not proof of ionic equilibrium or a microscopic chemical degradation mechanism.

Both D_ion and D_ion_neg are set to zero. Layer composition, temperature, initial compensating backgrounds, site capacities, and the inherited positive and negative ion profiles are preserved. Optical substrates with no electrical material parameters are retained without attempting to replace those parameters.

A negative species with nonzero population remains in MaterialArrays even when its diffusivity is zero. The ordinary density state retains its negative ion block, including the Poisson background and site-limit arrays. A declared but initially empty negative block is also retained when measuring an inherited state. Freezing mobility must not remove a charged species from electrostatics.

Every requested voltage starts from the same snapshot. After transient seeding and after electronic steady-state solution, both ionic arrays must be exactly equal to their inherited values; this also preserves their integrals on the unchanged grid. Non-finite, negative, malformed or over-capacity states fail. The measurement and current observer use the same frozen material cache.

The requested finite-duration transient supplies an initial guess. Its success does not establish electronic steady state. The existing solve_steady_state then solves the same carrier continuity equations with frozen ions. It must return a finite, converged state and a finite nonnegative continuity-current bound within the requested limit, default 0.05 A/m2. The reported current is also checked for spatial spread over all faces using that limit. Frozen-ion current must vanish, and displacement current is omitted only for this electronically stationary measurement.

FrozenIonSnapshot retains the voltage, current, accepted states, carrier residual, continuity-current bound, spatial current spread and derived metrics. DegradationResult retains the maximum current bounds/spreads per snapshot and both ionic histories when profile storage is enabled. Failures propagate to the experiment; no prior state or efficiency is substituted. The empirical damage law and its lifetime mapping are not retuned by this repair.

These current bounds are numerical acceptance criteria, not measurement errors or universal accuracy guarantees. Spatial resolution and weak-current scales require their own study. A successful sparse voltage sweep does not establish that its interpolated Voc or fill factor is independent of voltage sampling.

• tests/unit/experiments/test_degradation_frozen_ions.py covers absent, positive, negative and dual ions; optical substrates; nonuniform profiles with nonzero flux before freezing; exact profile preservation and zero per-species frozen current; independence from the original diffusivities; and rejection of missing or invalid electronic convergence evidence.
• tests/integration/test_degradation_snapshot_refinement.py fixes the initial nonuniform dual-ion state and physical history, varies internal seed steps by factors 1/2/4, then separately uses voltage steps 20/10/5 mV. The last two levels must agree within 1 mV in Voc, 0.2% in Jsc and 0.5% in FF. SOLARLAB_F4_EVIDENCE_PATH optionally exports raw curves, accepted states, parameter/mesh data, numerical-library versions and source hashes as JSON. Its paired JUnit result supplies the pass/fail record.
• Existing degradation tests retain their original checks on the empirical performance trend, and the componentwise-tolerance forwarding test remains.
• TPV frozen-dual-ion refinement now asserts actual negative-population retention before solving; earlier passing runs without that assertion did not verify this species when the old material builder discarded it.