April 1, 2026 — An extremely high-effort April Fools' joke. The simulation below runs real 2-group neutron transport compiled to WebAssembly. All the physics is real.

The Nuclear LightCell

Candle in a Jar — a fission reactor that converts nuclear fire directly to electricity through light.

Interactive Neutronics

This simulation runs a 2-group neutron transport model calibrated to 35+ OpenMC Monte Carlo runs with ENDF/B-VIII.0 cross sections. Drag the sliders to change the reactor design and watch criticality respond in real time.

k = 1.0000
effective multiplication factor
k∞ = 1.0000
infinite lattice
M² = 130 cm²
migration area
void = +0 pcm/%
void coefficient
Neutron tracks:
gold = fission   blue = capture   red = escape   gray = scatter

What You're Seeing

Each colored line is a neutron random-walking through the reactor. Fast neutrons (born from fission) scatter through the salt, slowing down until they're thermal. Thermal neutrons that hit a fuel pin can cause another fission (gold dot) — sustaining the chain reaction.

Try this: Drag the salt density to zero. Watch k-eff. In infinite lattice mode, it goes UP (+5,369 pcm — the Chernobyl direction). Switch to finite core mode. Now it plummets to ~0.3. The candle cannot burn without the jar.

The Void Coefficient Drama

The same physics gives opposite answers depending on whether you assume an infinite or finite reactor. Drag the slider and watch both sides simultaneously.

Infinite Lattice (wrong question)
k∞ = 1.4305
+0 pcm
k goes UP as salt drains. Positive void coefficient.
Finite Core R=55cm (right question)
k = 1.0550
-0 pcm
k CRASHES as salt drains. Neutrons fly out. Safe.
Migration Area M² — why leakage wins
130 cm²
0full salt: 130voided: 1,500

The Four-Factor Formula

k = η × f × p × ε. Each factor responds differently to salt density. Watch which ones drive the positive void coefficient.

η (reproduction)
2.07
ν × Σf / Σa in fuel
f (thermal util.)
0.85
Σa_fuel / Σa_total
p (res. escape)
0.82
exp(-NΣ/ξΣs)
ε (fast fission)
1.03
fast fission bonus
k = 1.4305   →   keff = k / (1 + M²B²) = 1.0550

OpenMC Ground Truth

21-point void sweep from actual Monte Carlo simulation. 50,000 particles × 150 batches. ENDF/B-VIII.0 cross sections. The orange dots are OpenMC; the line is our 2-group model.

The Light: TPV Spectral Conversion

The reactor IS the light bulb. LiCl vapor emits at 670.8nm (Li D-line) — a discrete spectral line matched to III-V photovoltaic cells. Not blackbody radiation. Monochromatic nuclear light.

670.8 nm
Li 2p → 2s emission
589 nm
Na D-line (trace)
~55%
InGaP cell efficiency
at 670.8nm monochromatic

The Design Space

Same physics, same architecture, same candle. Four orders of magnitude.

Variant Enrichment Core U-235 W/kg
Open-source lantern0.72%~8 m~200 t
D₂O + table salt3%~1 m14 kg
HALEU compact19.75%~1.1 m~300 kg~6
Moon basketball93% HEU20 cm39 kg384

The Heartbeat

The reactor breathes. Fission heats salt, salt boils, void fraction rises, reactivity drops, power drops, salt condenses, reactivity rises, power rises. A coupled nonlinear oscillation. Insert reactivity and watch it stabilize.

Heartbeat Controls
POWER
1.000
VOID
0.0%
Tfuel
2000 K
ρ (net)
0 pcm

Try this: Push the reactivity slider to +200 pcm. Watch power spike, then self-stabilize as Doppler and void feedback pull it back. The reactor finds its own equilibrium. Now push to +500 pcm (near prompt critical at β = 650 pcm). The oscillation gets violent but still converges. The candle regulates itself.

The Stack

OpenMC
35+ Monte Carlo sims
ENDF/B-VIII.0
Rust → WASM
107KB neutronics engine
in your browser
Claude Opus 4.6
architecture + sim design
+ this explorable
Gemini 3.1
UX design
+ rendering
Codex 5.4
WebGPU architecture
consultation
42 Advisors
Feynman, Tao, Oppenheimer
Weinberg, Dirac, Wigner, Dyson
"The fission reactor IS the light bulb."
Danielle Fong · LightCell Energy · April 1, 2026
An extremely high-effort April Fools' joke. All the physics is real.