surya.narada.systems

Research directions, full treatment

Each direction is defined by two things: which documented constraint it escapes, and what must be true for it to win. Founding document of the Surya program; adapted from theoriginating articleat space.narada.systems.

R1Space-ified terrestrial silicon

The workhorse. Near-term, highest confidence.

Ultrathin crystalline silicon with polymer encapsulation, no coverglass, radiation self-annealing — the route Solestial and Source Energy are already flying (~$5/W modules, flight-proven). Nothing here needs inventing; it needs characterizing.

  • Annealing kinetics at real operating temperatures as a function of wafer thickness and doping — the mechanism exists in the literature; the cliff is processing/fluence/doping-dependent, not a thickness wall.
  • Handling yield below 100 µm: the cited 10%-at-100 µm breakage figure traces to a 2016 CEA-INES pilot-line snapshot, since mitigated by tooling and edge reinforcement. Carriers, kerfless methods.
  • Atomic-oxygen protection without glass: ETFE, SiO₂ nanolaminates, sacrificial coatings.

Wins if: end-of-life degradation becomes predictable and boring — an actuarial input rather than a research question.

Design point, resolved August 2026: Ga-doped p-type SHJ at 60–90 µm, with a tracked path to 20–40 µm; the earlier 100 µm n-type assumption is superseded. CEA/INES demonstrated 90 µm and 60 µm Ga-SHJ recovering 97% after 1 MeV electrons at 10¹⁴ e/cm² with an 80 °C cure (ISFH-certified). Full record at/r1-design-point/ (graph eventev-000000000009, established-with-gap).

Annealing baseline, August 2026 (three research sweeps, 25 sources): the mechanism is real and defect-resolved, and ultra-thin defect-engineered SHJ cells fully recover at 65–100 °C under light — inside the realistic sunlit LEO band. Recovery is doping- and fluence-gated rather than purely thickness-gated. Full record at /r1-annealing/ (graph event ev-000000000008, established-with-gap).

R2Perovskite and perovskite/Si tandems

The upside bet. Mid-term.

Self-healing under irradiation, specific power above 1 kW/kg, projected $5–15/W at maturity. The tandem escapes every materials chokepoint at once: no germanium, no gallium worth counting, no glass.

  • Multi-year stability under combined UV, ATOX, and thermal cycling — single-stressor data exists; the coupled dataset does not, anywhere.
  • Encapsulation as the actual product — the cell is easy; keeping it alive is the company.
  • Qualification protocols: none exist for this class. Writing the test standard is itself a research direction with unusual leverage — whoever defines the protocol defines everyone's market entry cost.

Wins if: a credible 5-year LEO degradation model survives independent replication.

Readiness baseline, August 2026 (two independent research sweeps, corroborated): flight heritage exists and is recent and thin — MAPHEUS-8 (2019, first perovskites in space), a ~10-month ISS MISSE exposure whose degradation was >90% reversible, ~100 days on an RIT CubeSat with no observable degradation, plus Big Red Sat-1 and RHOK-SAT. Readiness sits at TRL 4–5, trending 6. The gap is duration and coupled stressors, not existence. Recorded as proposalep_1ca635b3 in the family epistemic graph; narrative record in the space ledger,Entry 888.

Coupled-stressor gap map, August 2026: the full single-stressor inventory, the III-V qualification benchmark, and the four datasets that would close the gap are published at/r2-gap-map/ (graph eventev-000000000007, assessment established-with-gap).

R3Substrate liberation: ELO/IMM

The geopolitical hedge. Mid-term.

Epitaxial lift-off and inverted metamorphic growth separate the III-V junction from the germanium wafer. The wafer becomes a reusable template instead of a consumable — the only direction that attacks the germanium chokepoint while keeping III-V performance.

  • ELO throughput and yield at industrial scale — best published etch rates imply hours per wafer; no yield figures are published anywhere.
  • Substrate reuse cycle count: the ~10-cycle threshold traces to a modeling assumption; demonstrated reuse is 4 cycles (GaAs) and 1 (Ge). Per-cycle reconditioning cost, not cycle count, is the binding constraint.
  • IMM mechanical reliability without the handle wafer: metal carriers and polymer laminates demonstrated; no space qualification of substrate-free product exists.

Wins if: reuse economics close at constellation volumes. Watch MicroLink and the Rocket Lab/AZUR process disclosures.

Gap map, August 2026 (four research sweeps, 32 sources):substrate liberation is physically demonstrated at cell level — 13 µm epitaxial stacks at ~31% AM0 and >3,000 W/kg, film-laminated flight heritage since 2009 — but economically undemonstrated: no published ELO yield, production throughput, multi-cycle Ge reuse, or space qualification of substrate-free product. The volume player (Rocket Lab/SolAero) removes substrates without reusing them; AZUR recycles Ge instead of lifting off. Full record at/r3-gap-map/ (graph eventev-000000000010, established-with-gap).

R4III-V on alternate lattices

A watching brief, not a program.

Metamorphic III-V on silicon, GaAs substrates (recyclable but still gallium), sapphire. Honest assessment: trades a structural chokepoint for a defect-physics problem. Threading dislocations and CTE mismatch over 30,000 thermal cycles are soluble in principle and unrewarding in practice so far.

Wins if: someone else's metamorphic-buffer breakthrough makes it cheap to revisit. Check annually, fund never.

R5In-space cell manufacturing

The long game. Far-term, and uniquely aligned with this program.

Deposit thin-film cells directly onto deployed membranes in orbit. Vacuum is free up there; launch mass collapses to feedstock; coverglass becomes optional when nothing has to survive the fairing. The only direction that escapes substrate, coverglass, andlaunch mass in one move.

  • Deposition rate and film quality in uncontrolled vacuum, on non-ideal substrates, with thermal drift.
  • Contamination of the host platform — the coater and the array must not poison each other.
  • Repair and re-coat as a service model: a coating pod visiting a membrane is an ISAM mission profile, and composes with on-orbit attachment architectures.

Wins if: deposited specific power beats launched specific power after accounting for the coater's own mass and operations. Nobody else is motivated to build this.

The gap map (August 2026) verified the empty chair: ~100× mass leverage of feedstock over arrays, no funded in-space PV program anywhere, and a critical path that starts with a ground deposition demo, not a flight. Full record at/r5-gap-map/ (graph eventev-000000000011, established-with-gap).

R6System architecture as research

Cross-cutting. Cheap to do now, compounding returns.

The ISCR analysis showed the move: integrate solar, radiator, and compute so the dedicated panel structure disappears (~500 W/kg at array level).

  • High-voltage array design to cut harness mass.
  • Mass-production panel standardization — the Starlink lesson: the cost was never the cell.
  • Qualification economics: cheap, fast radiation-test campaigns that let new materials accumulate credible heritage without a decade-long dossier. Attacks the filing-cabinet moat directly and benefits every other direction.

The gap map (August 2026) grades each sub-direction by evidence class: ISCR's 500 W/kg is a design target from a single unrefereed preprint, not a demonstrated number; high voltage is established heritage fenced out of LEO by plasma arcing; the Starlink standardization lesson survives independent scrutiny; the qualification moat now has a price list ($4.5M campaign, $800/h beam time, 33% of facility requests unsatisfied). Full record at/r6-gap-map/ (graph eventev-000000000012, established-with-gap).

R7Non-photovoltaic solar: the closing argument

Closed, August 2026.

Solar-thermal Stirling, thermionics, thermophotovoltaics with concentrators. The sweep was executed on 18 August 2026 and the answer is in: on measured system-level numbers, no non-PV technology beats PV on specific power, cost, or readiness — solar-dynamic closed at 3.4 W/kg engineered, thermionics at ~14 W/kg in a 1978 study, solar TPV at 0.8–3.2% measured system efficiency. The full closing argument with sourced numbers, and the four triggers that would re-open it, are at /r7-closing/. The space-unique variants (permanent source+sink, eclipse-cycle harvesting, thermoradiative laminates, tethers, solar-pumped lasers) were audited and closed earlier; see theclosed table.