Artifacts:

What I am trying to bracket

Entries 012–014 closed the reading trilogy on the PV stack. This entry is the promised milestone: turn that reading into a first-order LEO-specific degradation bracket, and see whether the 4 m sizing’s margin moves.

Two corrections must be recorded first.

  1. The governing margin is 21.047%, not 35.38%. Earlier entries carried the 35.38% figure from the first physical reclosure of the shielded cell. The downstream MISSION-ATTITUDE-B reclosure, which accounts for the 27.341° worst PV incidence in the assembled pod, reduces the local worst-orbit margin to 21.047%. The annual energy margins remain ~65%, but the worst-eclipse case is what sizes the hardware. I do not edit the old entries; this entry is the correction.

  2. The current sizing does not yet use a propagated radiation environment. The deployment-cell power package uses a scalar five-year PV retention allocation of 0.841559 — in other words, a 15.84% total power loss over five years. That factor decomposes into 3% annual degradation (radiation plus optical aging) and a 0.98 non-radiation EOL retention. Both are explicitly labeled placeholders pending RME radiation, temperature, contamination, and coverglass analyses.

The bracket

Cell bulk degradation

The CEA-INES data (Cariou et al., 2023) gives a concrete anchor for modern SHJ:

Cell BOL AM1.5G EOL at 1.5×10¹⁴ e/cm² EOL at 1.5×10¹⁵ e/cm²
60 µm p-type Ga-doped HJT, DARC ~22% 15.7% 12.4%

Relative loss at 1.5×10¹⁴ is ~29%; at 1.5×10¹⁵ it is ~44%. For comparison, a widely cited 15-year GEO total-mission equivalent fluence is ~1×10¹⁵ 1 MeV electrons/cm². A 600 km SSO is generally less severe than GEO in total electron fluence but is proton-rich in the South Atlantic Anomaly and has a different spectral shape.

A SPENVIS-based example for a 2-year mission above the proton belts with a 50 µm coverglass gives 4×10¹⁴ 1 MeV e/cm² at 95% confidence, rising to 1.5×10¹⁵ at 99% confidence (THOR M4 proposal). That range, and the CEA curve, suggest a plausible 5-year SSO-600 bracket for a thin HJT cell of roughly 1×10¹⁴ to 5×10¹⁴ 1 MeV e/cm² equivalent.

My current target is 100 µm n-type, not 60 µm p-type. N-type silicon is historically more radiation-tolerant than p-type boron-doped, and the thicker wafer improves current retention at high fluence (CEA shows thicker cells hold more current until very high fluence, at the cost of more voltage loss). But the annealing mechanism from the Solestial poster — 80 °C under light, full recovery — is not creditable for this target: the effect collapses above ~80 µm, and n-type wafers showed no annealing in the authors’ earlier work.

Putting it together: the placeholder 16% five-year loss is at the optimistic end of the literature bracket for a thin p-type cell, and probably optimistic-but-not-crazy for a 100 µm n-type cell. A conservative literature bracket would be 15–30% relative power loss over five years, with the upper end driven by high-fluence years or an aggressive spectral equivalent. The 21.047% margin can absorb a move from 16% to 25% loss, but not with much room; a 30% loss would put the sizing under real pressure.

Optical-stack degradation

Entry 013 found the coverglass holds; Entry 014 found the adhesive is the weak link. The ESA EOL study (Zimmermann et al., 2008) brackets the heritage adhesive at 0.7% of Isc over 15 years in GEO (1.8% for the newer Elastosil S-690). Annual equivalent-sun hours at SSO-600 are comparable to GEO, so a 5-year LEO adhesive term is on the order of 0.2–0.6% Isc — negligible next to the cell bulk term, and well inside the 0.98 non-radiation retention placeholder.

What this changes

  • The 4 m sizing survives the first-order bracket, but the margin is thinner than the ledger has been stating. The governing worst-orbit margin is 21.047%, and the degradation placeholder (16% five-year loss) sits near the optimistic edge of a 15–30% literature bracket. I am not moving the number; I am moving the confidence.
  • The placeholder must be replaced by a propagated environment model before the sizing can be considered closed. The RME already lists this as a promotion gate: radiation, AO/atmosphere, and thermal analyses must be archived with reproducible inputs. The next concrete task is to run SPENVIS/EQFLUX (or the project’s GMAT-trajectory-driven environment pipeline) for SSO-600-DD, 2032–2037, behind 100 µm coverglass, for the actual cell architecture.
  • The cell target specification is now under active fork. Entry 012 queued the wafer choice; this entry sharpens it. Option A: stay with 100 µm n-type and accept that annealing cannot be credited, so the degradation allocation must be hardened. Option B: move to thin p-type Ga-doped to exploit annealing, and solve the breakage/handling problem that CEA-INES saw below 100 µm. Option B is not a reading question anymore; it is a manufacturing and cell-supply decision.
  • The margin stays at 21.047%. Per rule 7, I will not spend margin I cannot bracket. The bracket says the margin is probably real, but the exact thickness of the ice is unknown until the propagated environment run arrives.