Joule Heist search

topics · updated 2026-09-25

Solid-State & Caloric Cooling — Refrigeration Without a Compressor

confidence: high volatility: warm verified: 2026-09-25fresh

Solid-state cooling replaces the vapor-compression refrigerant with a solid (or near-solid) whose entropy changes under a field — pumping heat with no compressor and no high-GWP gas. The caloric family is organized by driving field: magnetocaloric (magnetic), electrocaloric (electric), elastocaloric (stress), barocaloric (pressure), plus the emerging ionocaloric (electrochemical) route and the commercial outlier thermoelectric (Peltier). Caloric is theoretically efficient (magneto- and barocaloric ceilings exceed ideal vapor compression) but practically gated by cost, fatigue, hysteresis, and power density; the best realized caloric system reaches only ~60% exergetic efficiency, and only thermoelectric ships today, in precision/silent niches. Elastocaloric — now demonstrated at kilowatt scale — has the best near-term path.

Solid-state cooling swaps the vapor-compression refrigerant gas for a solid refrigerant whose entropy changes when a field is applied or removed — pumping heat with no compressor and no high-GWP gas. It is the most prominent class of not-in-kind cooling, and the one most often pitched as the eventual successor to the vapor-compression baseline. It hasn’t happened yet — and the honest reason is economics and implementation, not thermodynamics.

The caloric family — one mechanism, four fields

A caloric material sits near a ferroic phase transition; applying a field reversibly changes its entropy, so it heats up adiabatically when the field is applied and cools when it’s removed. Run in a regenerator, that cycle pumps heat. All members are characterized the same two ways — isothermal entropy change ΔS and adiabatic temperature change ΔT_ad — and the family is organized by which field drives it (Takeuchi & Sandeman’s frame):

EffectDriving fieldTypical ΔT_adSignature materialHeadline number
Magnetocaloricmagnetic (~1 T)~3 K (Gd, 1 T)Gd, La-Fe-Si, Mn-Fe-P-SiΔS ~10-15 J/kg·K at 1 T
Electrocaloricelectric (100-1000 kV/cm)~0.003-2.5 Kferroelectric ceramics/polymerslowest ΔT; least developed
Elastocaloricuniaxial stress (~400 MPa tension / ~600 MPa compression)~17 K (largest single-field); >20 °C reportedNiTi shape-memory alloyΔS 10-80 J/kg·K; material COP up to ~52
Barocalorichydrostatic pressure (0.1-0.6 GPa)~30 Kplastic crystals (neopentylglycol)ΔS ~445-510 J/kg·K (colossal)

The shared appeal is no refrigerant gas, no compressor. Multicaloric materials combine two fields for cross-coupled gains beyond the sum of the parts (e.g. Ni-Mn-In: +200% cyclic effect when 55 MPa of stress is added to a 1.9 T field). The shared liability is hysteresis: first-order transitions dissipate energy on every cycle, which both erodes efficiency and drives fatigue.

The members, in depth

Magnetocaloric — most-developed, cost-trapped

The historical leader, and the caloric candidate the DOE/PNNL screening judged most developed (the prototype history is in the alternatives scorecard). It is theoretically the most efficient caloric route (~100% maximum exergetic efficiency in an active-regenerator configuration) — but it needs a ~1 T field from expensive rare-earth permanent magnets, whose cost dominates the system, and its modest ΔS per cycle forces elaborate regenerators. Cost, not physics, is the wall.

Elastocaloric — the best near-term path

A stress-induced martensitic transformation in shape-memory alloys (mainly NiTi/Nitinol) releases and absorbs latent heat, giving the largest single-field ΔT_ad (~17 K; >20 °C under strain) and the highest material COPs (~52 reported at ΔT_ad ~11 K). The 2014 US DOE study of non-vapor-compression HVAC ranked it the most promising not-in-kind technology (~790 TWh/yr potential savings).

  • Performance ceiling: a 2025 HKUST multi-cell system (Nature) crossed kilowatt scale — 1,284 W at zero temperature lift, sustained over 500,000 cycles, 12.3 W/g specific cooling power at 3.5 Hz (prior devices ≤300 W). It uses compressive tubular NiTi, a graphene nanofluid heat-transfer medium, and an “SMAs in series, fluid in parallel” architecture that scales refrigerant mass at low system pressure. The prior US benchmark was UMD’s 260 W / 22.5 K-span multimode demonstrator (Science 2023).
  • The gating barrier is fatigue, not efficiency: tension-mode NiTi fails in hundreds to ~5,000-6,000 cycles against a >1,000,000-cycle (~20,000 h) commercial requirement. The fixes are loading mode (NiTi survives >10,000 cycles in compression vs ~165 in tension, hence the shift to compression/bending designs) and fatigue-resistant alloys (NiTiCu >10⁷ transformation cycles; additively-manufactured NiTiCuV >10⁶ cycles with small hysteresis). Durability today still needs specialized manufacturing.

Barocaloric — colossal entropy, high pressure

Hydrostatic pressure drives a plastic-crystal → ordered-crystal transition in cheap, abundant materials like neopentylglycol (NPG, a paint/lubricant/cosmetics feedstock). Because the molecules lose extensive orientational disorder (~60 configurations collapsing to 1), the entropy change is the largest of any caloric class — ΔS ~445-510 J/kg·K, ΔT_ad ~30 K at 0.57 GPa, transition near 41 °C, latent heat 110-121 kJ/kg, ~4.9% volume change — rivaling commercial HFC refrigerants. The catch is hardware: pressures of 0.25-0.57 GPa (2,500-5,700 bar) vs ~0.001 GPa for fluid refrigerants, ~10 K thermal hysteresis, and poor thermal conductivity (graphite matrices are the proposed fix). Lower-pressure salts such as ammonium sulphate (|ΔS| ~60 J/kg·K at only ~0.1 GPa, inverse barocaloric) are the active push.

Electrocaloric — compact but weakest

Electric-field-driven entropy change in ferroelectrics; attractive for on-chip/compact cooling (solid-state, no moving parts, easily addressed with voltage) but the lowest ΔT_ad and material COP of the four, the highest driving fields (100-1000 kV/cm), and the least developed. Its theoretical exergetic ceiling (93%) is nonetheless above vapor compression’s. Noted as a frontier in alternative heat-pump cycles.

Ionocaloric — the pumpable newcomer

The newest class (Science 2023, Lilley & Prasher, Berkeley Lab): electrochemically adding or removing ions from a salt shifts a material’s melting point, driving a solid↔liquid phase change. Demonstrated ΔT = 25 °C at <1 V using sodium iodide in ethylene carbonate (a lithium-ion battery solvent) — a larger lift than the other caloric routes. Its trick is that the liquid phase is pumpable, easing heat transport and sidestepping both barocaloric’s high-pressure hardware and the heat-transfer bottleneck of fully-solid caloric methods. Calculated to potentially match or exceed today’s gaseous refrigerants, with zero high-GWP gas.

Thermoelectric (Peltier) — the commercial outlier

The one solid-state cooling that actually ships at scale: a current through a thermocouple pumps heat (Peltier effect). Phononic alone cites 35M+ modules in the field — hyperscaler GPU/HBM/optical-transceiver chip cooling, medical-grade refrigerators, beverage coolers, cold-chain totes — alongside Coherent (ex-II-VI), Ferrotec, KELK, Kyocera, and Tark (ex-Laird), in a market growing ~8.7% CAGR. It wins where its weaknesses don’t matter: spot, precise, silent cooling with no moving parts, no refrigerant, compact form, millisecond response. But its efficiency is poor for bulk cooling: the figure of merit ZT has been stuck near ~1 for decades (lab ~1.7-3), which caps it at 10-15% of Carnot; DOE’s screening says replacing thermoelectrics into mainstream cooling needs ZT > 3. Vendor-cited real-device COP is <0.5 at a 20-25 °C lift, while the DOE/PNNL screening quotes COP ~1-1.5 — the two are not contradictory, since COP collapses as the lift grows, but both sit far below vapor compression’s 3-5. It is a niche technology, not a vapor-compression replacement.

Why solid-state hasn’t won — economics and implementation, not physics

The thermodynamic ceiling actually favors caloric. Schipper et al. (2023) put the maximum exergetic efficiency of an active-regenerator caloric cycle at ~100% (magnetocaloric), 96% (barocaloric), 93% (electrocaloric), 82% (elastocaloric) — all above vapor compression’s <80% even with a perfectly isentropic compressor. So in principle magneto- and barocaloric could beat an ideal vapor-compression system. The problem is the gap between ceiling and reality:

  • The best experimentally realized caloric system reaches only ~60% exergetic efficiency; real vapor-compression systems achieve ~50% (compressor isentropic efficiencies of 40-80%), or 40-60% of Carnot in the DOE/PNNL framing.
  • Vapor compression is a cheap, mass-produced, high-power-density, century-mature target at COP 3-5 (Takeuchi & Sandeman use ~3.6; the Carnot limit for a 10 K lift at 298 K is ~28.8). Their bar: a complete caloric system must reach COP ≥ 4 to compete, and “it remains to be seen how price-competitive complete systems will be.”
  • Each rival has its own deal-breaker: thermoelectric ~10-15% Carnot; magnetocaloric rare-earth magnet cost; elastocaloric fatigue plus 400-600 MPa actuation forces; barocaloric thousands of bar plus ~10 K hysteresis. Universal caloric problems: hysteresis losses (energy dissipated every cycle, driving fatigue), low power density vs compact compressors, and regenerator complexity that erodes the theoretical efficiency edge.

Verdict: Elastocaloric has the best path — no rare-earth field source, demonstrated >10⁶-cycle fatigue-resistant alloys, kW-scale prototypes, and lab-competitive COP — if fatigue-durable alloys can be made cheaply at scale. Thermoelectric owns the precision/silent niche today. Ionocaloric is the wildcard worth watching (pumpable fluid, sub-volt drive). For now, none has cleared the COP-4, low-cost-per-kW bar that vapor compression sets.

Source note on the DOE benchmark. The raw note for the DOE/PNNL screening (PNNL-19259) attributes it to Goetzler et al., 2014, while the same PDF as ingested for the alternatives article is attributed to Brown et al., March 2010 (five technologies ranked). The “2014 US DOE report” that ranked elastocaloric most promising (cited by the elastocaloric review) is a separate DOE non-vapor-compression HVAC study; this article does not conflate the two, and the Carnot/ZT figures above are consistent across both raw notes.

Where it fits this KB

Solid-state cooling is electric-work-driven (or stress/pressure-driven), so it sits with the vapor-compression baseline rather than the heat-driven adsorption/absorption thread this KB centers on. Its relevance here is as the other way off high-GWP refrigerants: where heat-driven cooling exploits free waste heat, solid-state exploits better materials — and like PV-driven cooling, it competes on electrical efficiency, not on harvesting an existing thermal stream.

See also

Sources