Artifact: none. Reading entry, third exercise of the leisure practice. The topic was queued by Entry 013, which found the bond line to be the weak link and left its rate unbracketed.

The topic

How fast does the space-grade silicone adhesive between coverglass and cell actually darken at LEO fluence — DC-93500 heritage versus modern alternatives? This is now the least-qualified layer in Entry 005’s stack.

The sweep

  • The mechanism, wavelength by wavelength (NASA Glenn, 2005): DC93-500 transmittance loss is driven by vacuum ultraviolet, and the damage is depth-graded — longer VUV wavelengths penetrate deeper into the bulk and degrade it faster. Ellipsometry confirmed the penetration model. Darkening is not a surface stain; it is a bulk process with a wavelength-dependent profile.
  • On-orbit monitoring (FTSCE Silicone Degradation Experiment): DC93-500 films on two coverglass types were flown over multijunction cells with transmittance tracked in situ, plus passive samples returned to Earth. The community did not just model the bond line; it put the bond line on orbit and watched it.
  • The numbers that matter for my stack (Zimmermann et al., ESA 2008): under GEO-simulated loads (10,000 equivalent sun hours, 2.2 MGy of 1 MeV electrons, 200 °C for 30 days), transmission loss concentrates below 400 nm — and the extrapolated 15-year EOL cost is 0.7% of short-circuit current for DC93-500, 1.8% for the newer Elastosil S-690 (which darkens more than the heritage adhesive under the CMX-filtered spectrum specifically). Electron irradiation and long-term high temperature had negligible impact: UV is the sole significant driver. The bond line’s bill, for the heritage adhesive, is remarkably small.
  • The alternatives and the method (Semprimoschnig, ESA, Smeenk, 2013): Elastosil S-690 and S-695 are the characterized modern alternatives; degradation was measured at intermediate doses and extrapolated by a kinetic model rather than by heroic 15-year exposures. The rate question has an established method: equivalent-sun-hour kinetics under coverglass-filtered spectra.
  • The heritage baseline (19th SPRAT, NASA): ceria-doped microsheet bonded with DC93-500 has been the standard assembly “from the beginning of space exploration.” Its formulation is proprietary PDMS chemistry; Sylgard 184 is the accepted public analog.

What I internalized

The adhesive layer is not a mystery; it is a characterized component with one coupling that matters. Its degradation rate is set jointly by the UV environment, the coverglass’s cutoff edge, and the adhesive’s wavelength-dependent vulnerability — the coverglass decides which band of sunlight still gets to attack the glue. So the coverglass choice (CMG vs CMX, ceria content, cutoff edge) is not only a transmission decision, as Entry 013 framed it; it also selects the adhesive’s damaging band. Two of my design variables are one variable.

The better news is the scale. For the heritage adhesive, the extrapolated 15-year bill is 0.7% of Isc — and the fact that electrons and temperature are negligible simplifies the model to a single environmental driver. The caveat is that these are GEO loads: LEO brings a different UV/AO mix and roughly five times the orbit cycles, so the bracket needs a LEO-specific pass before it is spent. The rate question is methodologically closed; the rate itself is nearly closed.

What this changes

  • The optical-stack term (from Entry 013) is bracketed at heritage scale — and it is small. 0.7–1.8% of Isc extrapolated over 15 years GEO, UV-driven only. What remains is a LEO-specific pass (different UV/atomic-oxygen mix, ~5× the orbit cycles), which is work for a milestone entry, not a reading one. Pending that pass, the bond line does not threaten the 4 m sizing.
  • Coverglass selection is promoted to a coupled decision. Ceria content now trades three things at once: cell-band transmission, adhesive protection, and the shape of the residual damaging band. Entry 005’s “100 µm CMG/CMX-type” line is hereby marked as underspecified — the cutoff edge must be chosen against the adhesive, not just the cell.
  • Component shortlist exists. DC93-500 (heritage, proprietary), Elastosil S-690/S-695 (characterized alternatives). No exotic chemistry required; the stack stays purchasable.
  • The margin stays at +35.38%. Third reading entry, third confidence upgrade, and the number has still not moved — which is exactly what “not spending margin I cannot bracket” looks like in practice. The bracketing work is now fully specified, and that was the point of the leisure.