# CCP/ICP Chamber-State Bridge Model Card

## Intended Use

This deterministic packet supports teaching, sensitivity comparison, evidence planning, and a normalized handoff to a separate feature-scale particle-transport lesson. It helps learners separate source sustainment, substrate bias, matching, gas-response, wall-loss, radial-delivery, IEDF/IADF, and wafer-facing flux questions.

## Model Class

Original, algebraic, dimensionless, precomputed browser replay over the complete locked 6,144-row Cartesian state space. It is not a plasma solver.

## Reference Archetypes

ccp_reference_archetype represents a bounded source/sheath-entangled teaching architecture. icp_reference_archetype represents a bounded coil-source plus substrate-bias teaching architecture with partial, not perfect, decoupling. Neither is universal and neither represents a named production chamber.

## Inputs

- Sixteen ordered normalized drive steps.
- CCP or ICP reference architecture.
- Low, middle, or high normalized pressure/collisionality regime.
- Low or high substrate-bias burden.
- Electropositive-like or attachment-prone-like gas response.
- Low or high wall-loss state.
- Nominal or detuned matching state.
- Center or edge radial zone.
- Teaching or replacement-slot evidence posture.

## Outputs

All outputs are dimensionless teaching proxies. The packet covers source coupling, absorbed drive, matching loss, bulk excitation, discharge support, density, electron-energy burden, attachment loss, wall loss, stored-state change, sheath burden, ion and neutral delivery, IEDF centroid and width, IADF spread, radial delivery, center-edge nonuniformity, source-bias entanglement, transition review, feature-bridge fields, role gates, and evidence readiness.

## Integer Teaching Ledgers

Three ledgers close exactly with integer tokens:

1. Delivered drive tokens split into bulk, sheath, matching/coupling loss, wall, and stored-state channels.
2. Generated charged tokens split into wafer ion delivery, charged wall loss, volume loss, and stored charge.
3. Generated neutral-reactive tokens split into wafer neutral delivery, wall recombination, exhaust loss, and stored neutral state.

Exact closure proves serialization and allocation consistency only. The tokens are not particles, charge, mass, energy, power, or species counts in a physical reactor.

## Determinism

The physical key excludes evidence posture. FNV-1a and xorshift32 provide deterministic tie-breaking for integer token allocation. Evidence-paired rows must be identical for every physical proxy, seed, token, and physical decision gate.

## Role Views

- Operator: sustained discharge, match sensitivity, wall-loss burden, and bounded-state review.
- Process engineer: source-bias entanglement, sheath, IEDF/IADF, attachment, and flux comparison.
- Integration: eligibility of normalized ion, neutral, energy-band, angular-spread, and radial-factor fields for a separate feature model.
- Yield: edge nonuniformity and attachment/wall-loss review without wafer disposition.

## Evidence Upgrade

Replacement evidence requires owned or rights-cleared geometry, chemistry, diagnostics, units, uncertainty, calibration range, provenance, independent verification, and accountable review. Selecting replacement_slot only opens an evidence task; it does not change physical values or upgrade model validity.

## Prohibited Use

Do not use this packet to choose a gas, pressure, source power, bias, frequency, matching action, recipe, endpoint action, maintenance action, equipment control, chamber purchase, supplier, lot disposition, qualification, safety action, or production signoff.

## Limitations

- No species chemistry, cross sections, EEDF, reaction set, RF waveform, electromagnetic field, sheath transient, stochastic particles, chamber geometry, coil, electrode, matching network, pump network, or thermal state.
- No physical units, calibration, measured row, benchmark row, PIC/MCC, Boltzmann, fluid, electromagnetic, eduPIC, PlasmaPy, LXCat, or vendor output.
- Center and edge are only two normalized delivery zones, not a radial profile or wafer uniformity prediction.
- Coefficients and thresholds are original teaching assumptions selected for bounded, auditable behavior rather than empirical fit.