The charge is not a packed bed — it is a molten salt suspension
- At 3:1 KOH:biochar mass ratio, the system is 75% molten KOH by mass at activation temperature.
- The KOH is the continuous liquid phase; the biochar is the dispersed solid phase.
- This means the heat transfer problem is not 'heating a packed bed through walls' but 'maintaining temperature uniformity in a molten salt bath with suspended particles.' The molten salt processing industry has solved this at scales 10-100x larger than yours for over a century.
- The immediate practical implication: even without exploiting this reframing, the correct scaling strategy is to preserve the constrained dimension (bed thickness ≤5 cm) and scale the unconstrained dimensions (number of layers, area per layer).
- Kuraray has produced MAXSORB commercially at this scale since the 1990s using thin-layer Ni boat processing; the physics is proven and the engineering is catalog-level.
If you prioritize speed to revenue (production in 6 months), start with the tray furnace. If you're optimizing for long-term cost structure and competitive moat, run the $2K stirring experiment this week — it determines whether your 5-year architecture is a tray furnace or a stirred bath.
Ni Tray Stack in Off-the-Shelf Multi-Zone Box Furnace + Statistical Blending
Replicates proven MAXSORB thin-layer approach with catalog equipment; 144 kg/day in 4-6 months at $150-300K; blending tightens ±12% tray variation to ±1.5-3% product variation
Stirred Molten KOH Bath with Perforated Ni Basket Immersion
Reframes the charge as a molten salt suspension, achieving ±3-5°C uniformity via convective heat transfer; blocked by one unvalidated question — does gentle stirring preserve micropores?
If this were my project, I'd start two workstreams this week — one that generates revenue in 6 months, and one that could change the game in 12-18 months. For the revenue track, I'd call Nabertherm and Carbolite Gero today for furnace quotes, and simultaneously order Ni 200 sheet for 5 prototype trays. While waiting for the furnace (8-12 week lead time), I'd run the 5-tray validation in my existing lab furnace to confirm the thermal uniformity and blending math. I'd also order a V-blender and 20 thermocouples — the blending strategy is the highest-ROI element of the whole portfolio, converting ±12% reactor variation into ±1.5-3% product variation for $10K. The moment the production furnace arrives, I'm commissioning and producing. For the game-changer track, I'd set up the stirred melt experiment the same week. A Ni crucible, a Ni paddle fabricated in a machine shop, a variable-speed motor, and my existing tube furnace — total cost under $2K. Four experiments at 0, 10, 20, 30 RPM, send products for CO₂ adsorption. If micropore fraction holds at 20 RPM, I now know my 5-year architecture is a stirred bath, not a tray furnace. That's worth $2K and two weeks to find out. I'd also connect a $300 CO sensor to my furnace exhaust during those same experiments and log CO vs. time. If I see a clear inflection point in dCO/dt, I've validated the endpoint detection concept for free — and that's applicable to whatever reactor I build.
- Start PAT calibration in parallel: every activation experiment generates a data point for the feedstock model. By the time the production furnace is running, I'll have 15-20 calibration points — enough for a preliminary model.
- Don't commit >$500K to any single architecture until the stirring experiment results are in.
- The multi-hearth furnace is the fallback if I need >300 kg/day and the stirred bath doesn't validate — but I wouldn't order it until I've exhausted the cheaper options.
The key insight is that these aren't sequential decisions — they're parallel bets with different timelines and risk profiles. The tray furnace gets you to market. The stirred bath gets you a competitive moat. The PAT model and endpoint detection improve everything. Run them all.