references · updated 2026-09-25
Cooling Technologies — Master Comparison
confidence: high volatility: warm verified: 2026-09-25fresh
A master feature-by-feature comparison of all cooling-generating technologies — drive source, working fluid/pair, COP, SCP, drive temperature, maturity, and best-fit application — with drill-downs on sorbent working pairs and not-in-kind alternatives, a COP-fairness note, a note on geothermal as sink versus source, and a quick decision guide. Figures are reconciled to the wiki's deep articles; where they come from a single vendor or a model rather than a measurement, the row says so.
Every cooling system moves heat against the gradient, so something must drive the pump. The cleanest way to compare cooling technologies is by what drives them — low-grade heat, electric work, sunlight, radiant emission, or the ground. This article compares the cooling-generating technologies across drive source, working fluid/pair, efficiency (COP), power density (SCP), drive temperature, maturity, and best-fit application, then drills into the sorbent pairs and not-in-kind alternatives. It is the side-by-side companion to the Overview; each row links to the article that carries the sources behind it.
COP for thermally-driven machines is a thermal COP (cooling ÷ heat in, typically < 1); for electric machines it is an electrical COP (cooling ÷ electricity, typically > 1) — the two are not directly comparable (see the COP-fairness note). Adsorption chillers also have an electrical COP (cooling ÷ pump and valve electricity) that Elsheniti’s review puts as high as 10, which is the number to set against a compressor’s 3–5.
The cooling generators
| Technology | Drive source | Working fluid / pair | COP | SCP / capacity | Drive temp | Maturity | Best-fit application |
|---|---|---|---|---|---|---|---|
| Vapor compression | Electric work (compressor) | HFC/HFO/CO₂/NH₃ refrigerants | 3–5 (electrical); ideal cycle 70–80 % of Carnot, large water-cooled centrifugals ~60 %, small air-cooled units ~40 % | high | — | Dominant / mature | Default firm-load cooling; anywhere with cheap electricity |
| Absorption | Low-grade heat (+ small solution pump) | LiBr/water (AC); ammonia/water (refrigeration) | ~0.7 single-effect (0.73–0.87 on 100–120 °C geothermal water); 1.2–1.4 double-effect | — | 70–200 °C; practical single-effect floor ~90 °C; Yazaki water-fired units run on 70–95 °C | Mature commercial (Yazaki water-fired 5–100 t, direct-fired 30–200 t) | Large waste-heat / CCHP chilled-water plants; below 0 °C with ammonia |
| Adsorption | Low-grade heat | silica gel/water (workhorse); zeolite, aluminophosphate, MOF, salt composites | 0.4–0.6 thermal (silica-gel benchmark ~0.45–0.6 at 80–85 °C); EMM-8 0.85 at 63 °C; MFM-300 0.8 at 62–64 °C; cascaded MOF pair 1.63 (lab) | silica gel ~160–250 W·kg⁻¹ (2nd-gen commercial 246); MOF bed 827 measured, ~970 predicted; coated AQSOA-Z02 1875 per kg of sorbent (375 whole-bed) | 50–95 °C (ultralow-drive materials 62–65 °C) | Commercial niche (~USD 215 M, 3.6 %/yr); every shipping unit still uses silica gel or zeolite | Waste-heat / solar sites with low-grade heat, benign refrigerant, and no moving or corrosive parts wanted |
| CO₂ (R744) | Electric or heat | CO₂ transcritical; activated-carbon/CO₂ adsorption pair | varies; CCHP trigeneration | — | — | Growing (natural refrigerant); the CCHP entry is held at title level | Natural-refrigerant systems; sub-zero; cogeneration |
| Desiccant & free cooling | Air-side economizer / low-grade heat | desiccant + air; heat-wheel economizer | n/a (economizer); sorbent handles latent load | — | regeneration roughly 50–80 °C | Mature in data centers; vendor entry held at title level | Latent-heavy loads; economize first, then run a machine |
| Radiative / black-body | Passive — emission to the cold sky | none (engineered surface) | passive (no work) | ~117 W·m⁻² (BaSO₄ paint, measured); CoolSkin façade 54–150 W·m⁻² (simulated) | none | Emerging: paint is factory-compatible, façade adsorber is a design study | Passive roof/façade heat rejection; supplement, not standalone |
| Heat pumps (alt cycles) | Electric work | water (R718) turbo/ejector; solid-state claims | varies; R718 commercial | — | — | Mixed (R718 commercial; solid-state at trade-news level) | Water-as-refrigerant chillers; two-way heating/cooling |
| Solid-state / caloric | Field on a solid (magnetic, electric, stress, pressure) | Gd / La-Fe-Si; ferroelectrics; NiTi; plastic crystals | magnetocaloric prototype ~1.2 at 12 °C span; elastocaloric material COP up to ~52 in the lab; best realized system ~60 % exergetic | elastocaloric now demonstrated at kW scale | none | ≥ 25 magnetocaloric prototypes, none commercial; elastocaloric has the best near-term path | Future compressor-free cooling; watch elastocaloric |
| Thermoacoustic | Electric (acoustic wave) | inert gas | ~20 % of Carnot achieved (60–100 % ceiling) | — | none | Many prototypes; PNNL “good” prospect | Future; highest theoretical ceiling |
| Thermoelectric (Peltier) | Electric (solid-state) | Bi₂Te₃ (solid) | 10–15 % of Carnot | very high flux | none | Commercial (spot) | Spot / electronics cooling only — not whole-space |
Geothermal: a sink, and only sometimes a source
Geothermal doesn’t make cold — it provides a stable ground-temperature heat-rejection sink for any of the above (especially adsorption, whose hot-day COP collapses with condenser temperature). It belongs in every comparison as the rejection partner that fights derating. At the other end of the spectrum, EGS produce 175–200 °C brine, but their natural cooling pathway is electricity to an electric chiller; Geothermal-Driven Sorption Cooling only wins when the heat is ≥ 90 °C, would otherwise be reinjected, and runs at high flat utilization.
Drill-down: adsorption & absorption working pairs
The two heat-driven families reward a closer look at the sorbent, since that sets COP, drive temperature, and cost.
| Pair | Family | Drive temp | COP / SCP | Note |
|---|---|---|---|---|
| Silica gel / water | Adsorption | 60–95 °C (typical regeneration 85 °C; destroyed above 120 °C) | COP ~0.45–0.6, SCP ~160–250 W·kg⁻¹ | The cheap, benign workhorse; net swing ≤ 0.1 kg/kg despite 0.35–0.4 equilibrium uptake |
| Zeolite / water | Adsorption | classic grades 150–250 °C (Elsheniti’s band 70–250 °C) | SCP up to ~600 W·kg⁻¹ (compact double-bed) | High uptake but high heat of adsorption (3300–4400 kJ·kg⁻¹) lowers COP |
| EMM-8 aluminophosphate | Adsorption | 63 °C | COP 0.85 (cooling); COP 1.75 heating at 82 °C | Record COP at ultralow drive; 0.28 g/g uptake, isosteric heat only ~47 kJ/mol; 30-cycle and boiling-water stable |
| MOF (MFM-300, CPO-27(Ni), MOF-303, MIL-101) | Adsorption | 62–130 °C | MFM-300 COP 0.8 at 62–64 °C; CPO-27(Ni) SCP 440 vs 310 for SAPO-34 at 130 °C; MOF-303 ~970 W·kg⁻¹ predicted | Frontier; MOF water uptake ~0.3–1.6 g/g but cycle-life at chiller duty is the open question |
| Salt-in-matrix (CaCl₂/LiCl/LiBr composites) | Adsorption | 90–100 °C desorption; AdHT pairs work on 40–80 °C waste heat | uptake up to ~4–5× the bare matrix; up to ~1700 kJ/kg specific energy for heat upgrading at COP ~0.5 | Deliquescence, salt-solution leakage, corrosion and pore blockage bound the gains |
| Activated carbon / methanol, ethanol, ammonia | Adsorption | ~100 °C (methanol, must stay < 120 °C); ≥ 200 °C sources for ammonia | low heat of adsorption (1800–2000 kJ·kg⁻¹) helps COP | Methanol is the classic solar ice-maker pair; ammonia runs at ~1600 kPa |
| LiBr / water | Absorption | 70–200 °C; floor ~90 °C single-effect | COP 0.7–1.4 | AC workhorse; crystallization risk, needs ≥ ~20 °C condenser water |
| Water / ammonia | Absorption | ≥ 90 °C single-effect (half-effect ≥ 80 °C at COP ~0.3) | COP ~0.6; reaches < 0 °C | Refrigeration; needs a rectifier; incompatible with copper and brass |
See Performance & Numbers, Zeolite, Composite Salt Sorbents, and Absorption Cooling.
Drill-down: not-in-kind alternatives scorecard
| Technology | Prospect vs vapor compression (PNNL 2010) | Best Carnot achieved | Theoretical max | Status |
|---|---|---|---|---|
| Thermoacoustic | Good | ~20 % | 60–100 % | Many prototypes |
| Magnetocaloric | Good | ~20 % | 50–60 % | ≥ 25 prototypes, none commercial |
| Thermotunneling | Average | (no data) | 50–80 % | Nanometer-gap fabrication problem |
| Thermoelectric | Fair | 10–15 % | 25–35 % | Niche spot cooling |
| Thermionic | Poor | < 10 % | 20–30 % | Backward heat conduction kills COP |
Baseline: vapor compression already reaches 60–80 % of Carnot in its best equipment. The four solid-refrigerant options gain a 10–20 % COP bonus from direct-contact heat transfer, which is why the ceilings look attractive on paper. The caloric family that emerged after this 2010 assessment (elastocaloric, barocaloric, ionocaloric) is covered in Solid-State & Caloric Cooling; the PNNL ranking itself is in Alternative Cooling Technologies.
Supporting layers (enable cooling, don’t generate it)
| Layer | Role | Article |
|---|---|---|
| Solar thermal | Supplies desorption heat for sorption cooling; ~3.5 m² of collector per kW of adsorption cooling | [[topics/solar-thermal |
| PV | Supplies electricity for the heat-pump route that now usually beats solar-thermal sorption | [[topics/solar-cooling-pv-vs-thermal |
| Heat battery | Buffers intermittent drive heat (load-shift) | [[topics/heat-battery-thermal-storage |
| Hydronics | Source-agnostic chilled/hot water distribution | [[topics/hydronics |
| Envelope, glazing & passive form | Cuts the cooling load before the machine | [[topics/envelope-and-glazing |
| Adsorbent bed engineering | Raises SCP within a given sorbent (coatings, additives, TPMS) | [[topics/adsorbent-bed-engineering |
How to read COP across the table
A vapor-compression “COP 4” and an adsorption “COP 0.5” are different currencies: the first is cooling per unit electricity, the second cooling per unit heat. Correct vapor compression to primary energy (grid efficiency ≈ 40 %) and its advantage shrinks; when the driving heat is waste or solar (≈ free), the heat-driven machines win on operating cost despite low thermal COP — at the price of higher first cost, larger footprint, and roughly double the heat rejection per unit of cooling. Yazaki’s CCHP arithmetic makes the same point at plant scale: separate power and boiler deliver ~56 useful units per 100 of fuel, a CHP plant with an absorption chiller on its waste heat ~80. The exergy view explains why thermal COP is low and not a tuning problem: a basic adsorption cycle’s exergy efficiency is ~10 %, with ~77 % of the destruction in adsorber thermal coupling; mass and heat recovery recover part of it (+24 % COP and +37.5 % SCP from mass recovery at 60 °C; +12.6 % COP but −10.8 % SCP from heat recovery). This is the central trade behind this wiki’s bias toward heat-driven cooling.
Quick decision guide
- Cheap grid power, firm load, smallest box → vapor compression.
- Roof space and no free heat → PV + heat pump usually beats solar-thermal sorption on cost, COP and simplicity.
- Abundant waste heat ≥ 90 °C, large chilled-water plant → absorption (higher COP; single-effect floor ~90 °C).
- Low-grade heat 50–90 °C, want no moving/corrosive parts, off-grid or solar → adsorption (EMM-8 or MFM-300-class sorbents if the drive is cool; silica gel is what ships).
- Geothermal brine → electricity to an electric chiller unless the heat is ≥ 90 °C, otherwise wasted, and the load is flat.
- Latent-heavy or data-center air → desiccant / free cooling first.
- Passive supplement, hot-sky climate → radiative roof paint or façade.
- Need a rejection sink that survives hot days → geothermal-coupled condenser.
- Spot-cool a chip → thermoelectric. Watching the future → elastocaloric first, then magnetocaloric / thermoacoustic.
See Also
- Overview — the narrative map this table condenses
- Getting Started — reading paths for new visitors; this table is the “compare options” entry point
- Performance & Numbers — the adsorption figures behind the table
- Absorption Cooling — the absorption column in depth
- Alternative Cooling Technologies — the not-in-kind column in depth (PNNL ranking)
- Solid-State & Caloric Cooling — the caloric family and thermoelectric in depth
- Solar Cooling: PV vs Thermal — PV-driven vs solar-thermal sorption, the cost/COP decision
- Geothermal-Driven Sorption Cooling — drive-temperature floors and COPs by cycle when the heat is geothermal brine
- Adsorption vs Absorption — the surface-vs-bulk distinction behind the two working-pair families in the table
Sources
Synthesized from this wiki’s compiled cooling articles (compiled-from: mixed); the key raw figures in this table trace to:
- Zotero cooling source registry — the theme list and the metadata-only entries (CO₂ CCHP, KyotoCooling, Rondo, heat pumps)
- Brown et al., PNNL-19259 (2010) — vapor-compression Carnot fractions and the not-in-kind scorecard
- Liu et al. — EMM-8 aluminophosphate — COP 0.85 at 63 °C, uptake, isosteric heat, stability
- Frazzica et al. — AdHT working pairs — composite salt sorbents for 40–80 °C waste-heat upgrading, ~1700 kJ/kg at COP ~0.5
- Elsheniti et al. — Adsorption Refrigeration Technologies — thermal vs electrical COP, silica gel / zeolite / carbon / MOF pair properties, CPO-27(Ni) vs SAPO-34
- Yazaki — off-grid cooling / absorption chillers — 70–95 °C water-fired drive, tonnage ranges, CCHP 56 vs 80 useful units
- Exergy analysis of advanced adsorption cycles — 10 % exergy efficiency, 77 % adsorber-coupling loss, mass/heat-recovery gains
- Aristov — CSPM salt-in-matrix review — up to 4× uptake, deliquescence, leakage, corrosion and pore-blockage limits