top of page
Gradient Background

From Laboratory Proof to the 2036 Thermal Diode

Part A: The Future Device · Part B: What we built to prove the concept

PART A — THE MODEL: WHAT THE 2036 DEVICE LOOKS LIKE

The Roadmap: From Proof to the 2036 Device

Four defined engineering steps — each a solved problem, not a new scientific question.
Step 1 — Better materials

Replace steel-glass (4.5× contrast) with gold-coated glass vs high-emissivity ceramic.
→ ~47× emissivity contrast · theoretically ~10× more rectification

Step 2 — Deeper vacuum

KF-flanged vacuum system targeting below 1 Torr.
→ Radiation fraction rises above 99%

Step 3 — Smaller gaps

Push below 10 µm to access near-field evanescent regime.
→ Heat transfer scales as 1/d² · rectification amplifies

Step 4 — 2036: Integrated arrays

Vacuum-sealed MEMS-fabricated micro-gap arrays embedded in chip packages and building panels.
→ Passive, solid-state thermal management with zero power input

PART B — THE PROOF OF CONCEPT

The physics experiment that proved directional heat flow is real

Every component of this apparatus was assembled by hand — in a home lab, with off-the-shelf parts, across two complete design phases. The final system ran overnight experiments logging 30,000 data points per run.

photo_apparatus_full.jpg
photo_vacuum_chamber_closeup.jpg

Solution Components

Component
Role

Aluminum 6061 carrier plates (75×75 mm)

Structural frame + thermal guard ring

Kapton film heaters (12V, 7W each)

Bidirectional heating — one per side

TEG modules (SP1848-27145)

Heat flux sensors — Vtop and Vbot

Graphite thermal interface material

Consistent thermal contact

Kapton shim spacers

Set exact gap: 8.5, 25.4, 254, 508 µm

Swappable sample plates

Glass (ε≈0.9) or Steel (ε≈0.2)

Cork insulation

Minimize edge heat losses

Arduino Mega 2560 + ADS1115 ADC

1 Hz, 16-bit, <3 µV noise

Vacuum chamber + mechanical pump

~38 Torr for vacuum tests

photo_phase1_failed.jpg
Phase 1 — Failed ✗
  • ✗ 150×150 mm plates — heat spread laterally
  • ✗ Saturated at 50°C, never reached steady state
  • ✗ Microsphere spacers clumped and crushed
  • ✗ Screw clamping introduced tilt and gap variation
Phase 2 — Working ✓
  • ✓ 75×75 mm plates — steady state in 10–12 min
  • ✓ Reached 70–92°C with stable plateaus
  • ✓ Kapton shim spacers — exact, repeatable gaps
  • ✓ Kinematic guides — parallel alignment every run
  • ✓ Overnight stable runs — 30,000 data points each
photo_phase2_exploded.jpg
photo_phase2_assembled.jpg
photo_phase2_side_layers.jpg

“A reliable measurement of a smaller effect is worth more than an unreliable measurement of a larger one.”

EXPERIMENT 1 — GAP SWEEP

What happens to heat transfer as the gap grows from 8.5 µm to 508 µm?

chart_gap_sweep_vbot.png
chart_thermal_current_divider.png

Discovery: The Thermal Current Divider

Metric
Change across 60× gap increase

V_bot (through-gap)

−32%

V_top + V_bot (sum)

only −2%

Sensitivity advantage

16× in favor of V_bot

The conventional measurement metric used in prior studies is insensitive by 16×. V_bot is the correct metric for gap physics.

EXPERIMENT 2 — ATMOSPHERIC NULL

Is thermal rectification detectable at atmospheric pressure?

chart_atmospheric_null.png

Experiment Results Summary

Result
Value

Corrected rectification ratio η

1.0006 ± 0.0008

p-value

0.754

Conclusion

No rectification detected

NULL RESULT — confirms air masking hypothesis ✓

A measurement system that correctly returns zero when no signal exists is a measurement system you can trust.

EXPERIMENT 3 — VACUUM RECTIFICATION ⭐ KEY RESULT

What happens when air is removed from the equation?

chart_vacuum_4panel_result.png
anim_abba_timeseries.gif
anim_atmospheric_null_vs_vacuum.gif
anim_rectification_result.gif
η_corrected below 1.0 means the glass-heated direction transferred more heat through the gap than the steel-heated direction - the asymmetry is the rectification signal.
 
chart_air_vs_vacuum_prediction.png
chart_allan_deviation.png

Setup

Parameter
Configuration

Pressure

~38 Torr (~5% atmosphere)

Air-to-radiation

7:1 → 0.4:1

Materials

Glass (ε=0.9) vs Steel (ε=0.2)

Gap

508 µm

Protocol

ABBA

Overnight run

~30,000 data points

Result

Phase
Configuration
V_bot (mV)
Vacuum

A1

Steel heated (forward)

350.7

28.5 inHg ✓

B1

Glass heated (reverse)

307.8

28.5 inHg ✓

B2

Glass heated (repeat)

306.2

Late drift ✓

A2

Steel heated (repeat)

306.3

29.5 inHg ✗ excluded

η corrected = 0.9325 ± 0.0016 · 6.7% thermal rectification · p < 0.001

95% CI: [6.4%, 7.1%]

“The same apparatus and protocol that produced a perfect null in air produced a clear, statistically significant positive when air conduction was removed. The atmospheric null makes the vacuum result credible.”

bottom of page