Solid-state cooling eliminates fans from electronics thermal management. Compare thermoelectric, electrocaloric, and ionic wind technologies.
TL;DR: Solid-state cooling removes mechanical fans from thermal management, using electrical and physical effects to move heat with no moving parts. As TDPs climb past passive heatsink limits, technologies like thermoelectric, electrocaloric and ionic wind are becoming production-ready. This guide compares each on efficiency, capacity, noise and readiness.
Heat flux, not total power, is what breaks modern thermal design. AI accelerators now exceed 300 W/cm² at the die, a regime where the limiting resistance is the boundary layer at the surface rather than the heatsink behind it.
Fans remain the default answer, and they carry a fixed set of costs: audible noise, vibration coupled into sensitive optics and sensors, bearing wear, and dust ingestion that degrades performance over the deployment life. Worse, an entire class of hardware cannot use them at all. Sealed enclosures — edge AI boxes, outdoor telecom cabinets, automotive control units, industrial IP67 equipment — are designed specifically to keep air out.
The market has responded. Solid-state cooling reached roughly $960M in 2025 and is projected to exceed $2B by 2033, an 8–10% CAGR driven mostly by electronics thermal management rather than the refrigeration applications that historically defined the category.
The most mature option by a wide margin. Bismuth telluride (Bi2Te3) semiconductor couples pump heat when current flows across the junction. Coefficient of performance sits at 0.4–0.7 in practical electronics use, meaning you spend more electrical power than the thermal power you move — acceptable when the goal is precision spot cooling below about 50 W, unacceptable at system level. Recent thin-film Bi2Te3 work has improved COP by 15–20%, and TECs remain the only solid-state technology that can drive a surface below ambient. See our direct comparison with plasma actuators for the trade-offs in detail.
Applying an electric field to a polar dielectric changes its dipole ordering and therefore its temperature. Prototypes have demonstrated temperature spans around 10 K, and because the active material is a thin film, chip-scale integration is genuinely plausible. The technology is not yet broadly commercial: cycling endurance, dielectric breakdown and heat-transfer plumbing at the film scale are all open engineering problems.
Gadolinium and gadolinium alloys heat and cool as they magnetise and demagnetise. The effect is strong — ΔT above 20 K is routine — but it requires field strengths above 1 T, which in practice means heavy rare-earth permanent magnets or superconducting coils. That mass and volume penalty rules it out for board-level electronics; magnetocaloric is a credible HVAC and refrigeration technology, not an electronics one.
Nickel-titanium shape-memory alloys release and absorb latent heat as they transform under mechanical stress. Reported temperature changes reach a record 38.5 K, the largest of any caloric effect. The catch is in the name: it needs mechanical actuation, which reintroduces the moving parts that solid-state cooling exists to eliminate. Promising for future HVAC, not yet viable for electronics.
A dielectric barrier discharge ionises a thin layer of air at an exposed electrode; the electric field accelerates those ions, which drag neutral air with them into a wall jet. There are no blades, no bearings and no actuation — it is solid-state in the strict sense.
Benard and Moreau measured induced velocities up to roughly 7 m/s in optimized surface DBD configurations, the same order as a small axial fan, from a device under a millimetre thick.
The practical numbers are what make it interesting for electronics: devices under 3 mm thick, orientation-independent operation, and no mechanical wear-out mechanism, since there are no bearings or blades. Independent coverage of our CES 2026 laptop cooler reported operation at 17 dBA. Convective heat transfer coefficients approach 50 W/m²·K, which is enough to break up the stagnant boundary layer that dominates thermal resistance in confined spaces. Best fit: sealed enclosures, edge AI, telecom cabinets and drones. Our ionic wind vs fans comparison covers the crossover in depth.
| Criterion | Fans | Solid-state |
|---|---|---|
| Noise | 25–45 dBA | Below 20 dBA |
| Reliability (MTBF) | 30,000–70,000 h | Above 100,000 h |
| Form factor | 10–40 mm thick | Ionic wind under 3 mm |
| Sealed enclosures | Incompatible with IP67 | EHD works inside sealed volumes |
| Efficiency | Wins at high thermal load | Crossover below roughly 15–20 W |
The efficiency crossover is the honest summary: above about 20 W of thermal load per module, a well-designed fan still moves more air per watt. Below that, and in any design where noise, sealing, thickness or maintenance dominate the requirement set, solid-state wins on system terms rather than on raw thermal terms.
The edge case is the sharpest one; see edge AI thermal management for how the constraints stack up in practice.
Match the technology to the constraint, not to the headline number:
If you are designing a sealed, silent or space-constrained thermal path, we can help you evaluate whether solid-state airflow closes the gap. Book a demo with the YPlasma team.