# HEX-CELL COMPONENTS — design · analysis · build · assembly · construction
2026-07-17. Companion to HEX-CELL.md. Sourcing discipline: [T] = textbook value, confirm against
spec sheet before release; [UNSOURCED] = estimate needing derivation; [OP] = operator datum.

## 1. Component list (one cell)

| # | Part | Material / process | Qty | Key spec | Cost est |
|---|------|-------------------|-----|----------|----------|
| C1 | Hex header, bottom (seat + salt bed + injector/nozzle block) | Alumina, Form 4 SLA + sinter | 1 | 150.7 mm flat-to-flat, ~76 mm tall incl. plenum | [OP] 4× below retail, target <10% BOM |
| C2 | Hex header, top (exhaust + tube retention + spring seat) | Alumina, Form 4 SLA + sinter | 1 | mirror of C1, sliding tube seat | same |
| C3 | Outer envelope tube | Fused quartz, commodity | 1 | 101.6 × 1257 mm, wall ~2.5 mm [T, confirm] | $49.62 retail [OP] → $15–25 vol |
| C4 | Inner emitter tube | PCA (polycrystalline alumina) or small quartz | 1 | ⌀ TBD ≤ 100 mm, sprung mount | [UNSOURCED] |
| C5 | Optional sleeves | vacuum jacket / ITO hot-mirror | 0–1 | slides in C1/C2 bore — tile is the ISA | [UNSOURCED] |
| C6 | Honeycomb column | Extruded alumina (catalyst-substrate industry) | shared | runs top-to-bottom through hex stack | cheapest ceramic form factor |
| C7 | Salt charge | Na reservoir salt, composition TBD | 1 | seat bed; cold-spot reservoir | small |
| C8 | Springs | High-temp alloy (Inconel-class) at COOL end only | 2–4 | preload ~10–20 N [UNSOURCED] | small |
| C9 | Gasketry | Compliant seal at cool ends only | 2 | — | small |

## 2. Analysis

### 2.1 Thermal expansion — the governing mechanical problem
CTE: fused quartz ~0.55 ppm/K [T]; alumina ~8 ppm/K [T]. Mismatch ~7.5 ppm/K.
- Radial, across 150 mm tile at ΔT ≈ 800 K: alumina grows ~0.9 mm, quartz ~0.07 mm →
  **never clamp quartz radially in alumina hot**. Seat = loose cradle + salt bed compliance.
- Axial, 1257 mm tube: quartz grows ~0.55 mm at ΔT 800 K; alumina honeycomb frame grows ~8 mm
  over the same span at its own ΔT. → **fixed seat at bottom, sliding seat at top**, spring
  preload from the COOL end only (C8). This is the mechanical meaning of "sprung".
- Operator's sprung-PCA note is the same principle applied to the inner tube.

### 2.2 Mass & loads
- C3 quartz: π·D·t·L·ρ ≈ 3.14·0.102·0.0025·1.257·2200 ≈ **2.2 kg** [T ρ]
- C1+C2 printed alumina: ~0.3–0.6 kg ea [UNSOURCED, print volume dependent]
- Cell total order **5–8 kg** incl. internals.
- 315 cells per 420 m column → ~2 t/column. **Cells must NOT be load-bearing in series** —
  glass in a 315-high compression stack invites buckling/stress-concentration failure.
  → cassette architecture (§4): cells hang in frames; frames carry load to tower structure.

### 2.3 Forced-air cooling (cool end)
Target: remove Q ≈ 1 kW/cell at cold end [UNSOURCED until wall-balance run against kernel]:
ṁ = Q/(cp·ΔT) = 1000/(1005·40) ≈ 0.025 kg/s ≈ **21 L/s ≈ 45 CFM per cell** — muffin-fan class,
or manifolded plenum per cassette. Cold-end air ALSO sets salt-seat temperature →
**the fan duty cycle is the P_Na density knob** (transport relation, not a limit —
see feedback_no_fabricated_limit_thresholds.md). Cooling and vapor-density control are one system.

### 2.4 Optics
- Alumina cavity walls: diffuse reflectance ~0.9+ visible [T, confirm for sintered SLA surface] —
  photon recycling between neighbor cells through shared walls.
- L/D 12.4: end-loss fraction per end ~(D/2L)²/4 ≈ sub-% [geometry].
- Silicon placement: along tube length (the "long solar length"), cool side of the air gap.

### 2.5 Chemistry boundaries (from HEX-CELL flags)
- Quartz sees moderated T only; Na chemistry lives on PCA/alumina inner. Quartz devitrification
  + alkali attack is the failure mode this architecture routes around.
- Salt composition (C7) unspecified — needs selection for vapor-pressure curve + quartz compat.

## 3. Build (per-part process)
- C1/C2: Form 4 print → burnout/sinter in-house ovens [OP: shrinkage factored] → seat-face lap
  if needed → salt charge. Print IS the tile; no machining of the hex outline.
- C3: buy. Incoming QC: dimensional + visual striae check.
- C6: purchase extrusion cut-to-length; ends interface into C1/C2 registration features.
- Tolerance chain: hex flat-to-flat sets cell pitch. SLA + sinter shrinkage scatter [UNSOURCED,
  measure on first article batch] — assembly must accept tile-to-tile variation via compliant
  inter-tile joints, not precision fit.

## 4. Assembly (cell → cassette)
Sequence per cell:
1. C1 bottom header: salt charge into seat bed; injector check (flow test, cold).
2. Drop C3 quartz into cradle; verify float (no radial bind).
3. Insert C4 inner emitter on sprung mount; optional C5 sleeve.
4. C2 top header on; C8 springs preload from cool end; C9 gaskets cool ends only.
5. Cold leak/flow test; log per-cell serial.
~~Cassette frame~~ **SUPERSEDED 2026-07-19 (operator): hexes clip into each other — the frame
was redundant.** Interlocked tile plate: hex tiles clip edge-to-edge via printed tabs on the hex
flats, forming a self-supporting diaphragm per level; shared plenum manifolds cooling air +
fuel/oxidizer through the tessellation. **Tile = field-replaceable unit** — unclip, lift out
vertically, refurb on bench. Uptime lives here.

## 5. Construction (cassette → tower)
- Tower structure supports tile plates at intervals; the clipped tessellation is the diaphragm — individual quartz cells never in the structural path.
- 315 tile-rows per column height; annular shell fill (~27% [OP notebook]) not solid pack.
- Service: tiles unclip and lift vertically; every cell reachable without disassembling neighbors.
- Commissioning: per-plate burn-in on ground before hoist (assembly line, not field debug).
- Scaling frame: everything above is bench-provable at n=1 cell and n=7 clipped plate BEFORE any
  tower exists. **The unit test is physical.**

## 6. Open engineering items
1. Wall balance from kernel → real per-cell Q for §2.3 (replace [UNSOURCED]).
2. Salt selection (C7): vapor-pressure curve + compat.
3. First-article shrinkage scatter measurement (§3 tolerance chain).
4. Spring alloy + preload at temperature (C8).
5. Clip-tab geometry: engagement depth, sinter-shrinkage tolerance, thermal-cycling retention. (Replaces superseded cassette-frame trade.)
6. Silicon module geometry vs tube (flat facets vs curved).