ZIF-8 Fe1/NC to Fe1/NCP Closed-Loop Baseline

This is RIGOR's reference two-Iteration scientific demonstration. It starts from one ASCEND request, composes two immutable PACE packages, pauses at operator and safety boundaries, analyzes raw ORR polarization data in PRISM, and stops from an evidence-bound ASCEND decision.

The checked-in baseline demonstrates system behavior and data lineage. ORR performance alone is not evidence that an atomic coordination structure was verified.

Demo Objective

Compare a ZIF-8-derived Fe1/NC control with a phosphorus post-treated Fe1/NCP child sample. The acceptance criterion is an ORR half-wave-potential improvement of at least 0.03 V after no more than two Iterations.

Iteration 1: sample_fe1_nc_baseline_001
  module: zif_fe_nc_baseline
  target output: ORR E1/2 and polarization-curve evidence

Iteration 2: sample_fe1_ncp_baseline_001
  module: zif_fe_ncp_baseline
  parent: sample_fe1_nc_baseline_001
  target output: ORR E1/2 and polarization-curve evidence

PRISM recalculates E1/2 from the delivered CSV instead of trusting only the device summary.

One-Request Entry

Use ASCEND chat to create a Coordination with this intent:

Compare a ZIF-8-derived Fe1/NC control with a secondary-sphere P-modified Fe1/NCP candidate using ORR RDE LSV. Run at most two Iterations. Start with zif_fe_nc_baseline and sample sample_fe1_nc_baseline_001; if its committed PRISM result is acceptable, run zif_fe_ncp_baseline with sample sample_fe1_ncp_baseline_001 and parent sample_fe1_nc_baseline_001. Stop when E1/2 improves by at least 0.03 V. Do not claim atomic coordination from ORR performance.

Recommended policy:

{
  "max_iterations": 2,
  "auto_decide": true,
  "auto_compose": true,
  "require_decision_approval": false,
  "require_package_approval": true,
  "parameter_bounds": {},
  "stop_conditions": [
    {
      "semantic_id": "electrochemistry.orr.half_wave_potential.delta",
      "min": 0.03
    }
  ]
}

Package review remains enabled. High-risk LabFlow and LabBridge actions retain their own approval or Human Task boundaries independently of ASCEND's decision policy.

Process Route

Loop 1 uses the bundled R002/R004 evidence model:

1. Prepare ZIF-8 from 1.29 g zinc nitrate and 1.62 g 2-methylimidazole, each in 30 mL of 4:1 DMF/methanol; stir for 24 h, separate, and vacuum dry at 80 C.
2. Disperse 200 mg ZIF-8 and 60 mg dopamine separately in 20 mL methanol, combine after sonication, stir for 12 h, separate, and dry.
3. Pyrolyze at 1000 C for 2 h under Ar at 5 C/min and apply 20 s low-RF Ar plasma treatment.
4. Use 50 mg FeCl3 and 80 mg N-doped carbon in operator-verified positions; treat at 700 C for 2 h under Ar at 5 C/min.
5. Prepare and document an O2-saturated 0.1 M KOH RDE cell, then acquire LSV at 10 mV/s and 1600 rpm.

Loop 2 takes 80 mg of the loop-1 Fe1/NC sample, uses 200 mg sodium hypophosphite at a separately verified source position, and applies the cited 900 C, 2 h, 2 C/min Ar post-treatment before the same RDE measurement.

Local SOP parameters are explicit package inputs where the cited route does not provide enough operating detail. They are not presented as literature facts.

Human Boundaries

The semantically bound reactive-furnace preparation capability requires a setup record and manual acceptance before reactive furnace work. The operator must confirm the method, source and sample positions, purge, hazard controls, and identity. Heating sodium hypophosphite can generate hazardous phosphorus-containing species; physical use requires an institution-approved SOP and engineering controls. This repository does not authorize that setup.

The semantically bound RDE-cell preparation capability records electrolyte, gas, electrodes, catalyst loading, operator identity, and a setup artifact. PACE remains waiting until the accepted submission completes the Gateway operation.

Evidence Flow

Composer zif-fe-ncp-baseline Skill
  -> semantic device binding and immutable package
  -> PACE / LabFlow execution
  -> LabBridge operations + Human Tasks + object-store artifacts
  -> PACE result contract + sample-lineage/v1
  -> <project>_loopN.tar.gz with small verified artifacts embedded
  -> PRISM orr-rde-analysis Skill
  -> E1/2, uncertainty, normalized CSV, SVG plot, analysis archive
  -> committed prism-analysis-result/v1
  -> ASCEND next_iteration or stop

PACE keeps object-store identifiers and links even when it embeds an artifact. Files larger than PACE_DELIVERY_EMBED_ARTIFACT_MAX_BYTES remain external and are marked with an explicit embedding status.

Scientific Boundary

The scenario models the modified candidate as Fe-N4 with P in the second coordination sphere. ORR LSV can show a performance difference but cannot prove FeN3P, a direct Fe-P bond, or any atomic coordination. A physical structural claim requires independent characterization such as appropriately interpreted XAFS/XPS evidence and a validated analysis method.

Verification

The repository tests cover:

• scenario versioning and deterministic separation between the two samples;
• Human Task schema and manual review for reactive furnace setup;
• Composer semantic feasibility, explicit module selection, and package build;
• PACE sample lineage, result metadata, and verified artifact embedding;
• PRISM curve recalculation, uncertainty, device cross-check, and plot output;
• ASCEND first-to-second Iteration lineage and evidence-bound stop.

The software test does not bypass package review, fabricate Human Task submissions, or execute a physical furnace.