Joule Heist search

references · updated 2026-09-25

Adsorption Cooling — Performance, Working Pairs & Numbers

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

The quantitative deep-dive behind adsorption cooling: COP and SCP by working pair and driving temperature — the silica-gel/water benchmark band (COP ~0.45–0.6, SCP ~160–250 W·kg⁻¹ at 80–85 °C) and its 2nd-generation commercial baseline (SCP 246, COP 0.48); the ultralow-temperature champions EMM-8 (COP 0.85 at 63 °C) and MFM-300 (COP 0.8 at 62–64 °C with proven hydrothermal stability); MOF-era ceilings (cascaded COP 1.63, measured 827 W·kg⁻¹, MOF-303 predicted ~970 W·kg⁻¹); bed enhancements, the COP-fairness argument, hot-climate and fluctuating-source derating, economics, desalination cogeneration and waste-heat upgrading. The sizing reference for moving from concept to a real machine, with implications for 601 Delaware.

The quantitative deep-dive behind Adsorption Cooling: what COP and cooling power to expect, from which working pair, at what driving temperature, and what it costs — built from the full-text papers in the hothothot/Joule Heist adsorption set plus the 2025–2026 MOF, MOF-303, fluctuating-source and third-generation sources.

Provenance note. Figures marked (registry) — Dias 2022, Alahmer 2020, Motamedi 2022, Kalawa 2019, Greiner 2022 (CoolSkin), Tatlier 2023, the Jordan paper — were read from Zotero full texts in the June 2026 compile; only their registry metadata is in the raw set, so they are carried forward unverified. Everything else traces to a full raw source.

Two metrics run through everything. COP = cooling delivered ÷ driving heat — a thermal COP, not a compressor’s electrical COP (see the fairness note). SCP (specific cooling power, W·kg⁻¹ of adsorbent) = cold per kg of sorbent per unit time; SCP sets the machine’s size and cost, COP its running efficiency.

The benchmark operating point

The recurring test point is hot water 80–85 °C, cooling water 25–30 °C, chilled water 7–14 °C, where silica-gel/water converges on COP 0.45–0.6 and SCP ~160–250 W·kg⁻¹, sliding to COP 0.2–0.35 and SCP < 50 W·kg⁻¹ below 60 °C. Driving temperature is the dominant lever (Sosnowski’s three-bed sweep, Shahzad’s KAUST pilot), SorTech’s 7.5 kW prototype reached system COP 0.55 on 9 W of electricity (Elsheniti), and Gado/Frazzica cite the 2nd-generation commercial baseline as SCP 246 W·kg⁻¹ / COP 0.48. The classic flat-tube datapoint, COP 0.45 / SCP 176 at 80/30/14 °C, is surveyed in Adsorption Cooling (Kumar) and corroborated (registry) by Alahmer.

Source / casePairDriveCOPSCP (W·kg⁻¹)Notes
Three-bed chiller (Sosnowski)silica gel/water57 → 85 °C0.20 → 0.5827 → 160heating-water sweep
KAUST 4-bed solar pilot (Shahzad)silica gel/water85 °C, cooling 25 °C0.43–0.50—chilled 7 °C
2nd-gen commercial baseline (Gado)silica gel/water—0.48246SorTech/Fahrenheit class
Best conventional in ORNL tablessilica gel/water—0.61540enhanced lab bed
Cooler model (registry: Dias)silica gel/water70 °C0.5044cond/evap 30/15 °C
…same, very hot day (registry: Dias)silica gel/water70 °C0.3126condenser 42 °C
Mobile double-bed (Kumar survey)zeolite/water—0.40600compact automotive
CPO-27(Ni) vs SAPO-34 (Elsheniti)MOF / AlPO, water130 °C—440 vs 310900 s cycle
20-PPI metal-foam bed (ORNL)MOF/water—0.75827measured hardware
AQSOA-Z02 BACT coating (Jacobucci)zeolite-type/water90 °C—1875 (Z02) → 375 (bed)120 s cycle
MOF-303 two-bed model (Ben-Mansour)MOF/water75–90 °C0.62–0.63440 → 970predicted from measured properties
EMM-8 aluminophosphate (Liu)zeolite-like/water63 °C0.85—record COP at ultralow drive
MFM-300(Al/Fe/Cr/V) (JACS 2025)MOF/water62–64 °C0.80—proven hydrothermal stability
Cascaded Ni₂Cl₂BBTA + Co₂Cl₂BTDD (ORNL)MOF/water—1.63—record; single bed 0.70
Stratisorp, heating mode (see Limitations)zeolite Li-Y/water200 °C1.98 (heating)—sink 34 °C

Working pairs — the real trade

  • Silica gel / water — the workhorse. 2–3.5 nm pores, 400–700 m²·g⁻¹, 2500–2800 kJ·kg⁻¹, regeneration typically 85 °C (multi-stage to 50 °C; destroyed above ~120 °C). Equilibrium uptake 0.35–0.40 kg·kg⁻¹ but a usable swing ≤ 0.1 kg·kg⁻¹ — the root of its bulk (Elsheniti). The default unless there is a reason to move.
  • Zeolite / water — higher uptake, stability and SCP (600 W·kg⁻¹ in the mobile unit), but 3300–4200 kJ·kg⁻¹ heat of adsorption lowers COP and classic zeolites want the hot end of a 70–250 °C desorption band — disqualifying most low-grade heat (Elsheniti; Zeolite).
  • Aluminophosphates (AQSOA-Z02, SAPO-34) — the commercial low-temperature zeolites. AQSOA-Z02/water is Jacobucci’s automotive pair (1.5 kW over a 400 s cycle from 90 °C engine coolant; 150 W at 1:10 scale) and plateaus at COP ~0.5 as a heat-transformer pair at 80 °C (Frazzica). SAPO-34: 310 W·kg⁻¹ at 130 °C (Elsheniti).
  • EMM-8 (SFO aluminophosphate) — cooling COP 0.85 at 63 °C drive (evap 5 / cond 30 °C), regeneration 65 °C, uptake 0.283 g·g⁻¹ with an S-shaped step at P/P₀ 0.15–0.17, isosteric heat 46.8 kJ·mol⁻¹ (barely above water’s 44 kJ·mol⁻¹ latent heat — why COP is high), heat-pump COP 1.75 at 82 °C, and 5–15 °C lower driving temperature than reference materials. Stable over 30 cycles and 24 h of boiling; cheap precursors, ~100 g batches (Liu 2022, Nature Communications).
  • MFM-300(Al/Fe/Cr/V) — the 2025 MOF answer: COP 0.8 at 62–64 °C in both A/C and refrigeration modes, uptake 0.31–0.34 g·g⁻¹, heat of adsorption ~49 kJ·mol⁻¹, working capacity 0.27–0.35 mL·mL⁻¹ at 64 °C, full capacity after a one-month soak, five days boiling and 20 cycles (JACS 2025). Source-vs-source note: the raw summary says it “beats EMM-8 on driving temperature,” but as ingested both sit at ~63 °C and EMM-8’s COP is higher; MFM-300’s real edge is demonstrated hydrothermal stability — the property the 2026 MOF-durability verdict says most MOFs lack.
  • Other MOFs (MIL-101(Cr), Al-fumarate, UiO-66, MOF-801, MOF-303, CPO-27(Ni)) — uptakes far beyond silica gel (MIL-101(Cr) ~1 g·g⁻¹, MIL-101@GO 1.58, Al-fumarate > 0.5 regenerating at 60 °C; Elsheniti). The ORNL/DOE review tabulates MOF–water COP 0.3–1.66 vs conventional 0.1–0.8 at 57–127 °C desorption and concludes MOFs now consistently beat conventional adsorbents on COP — yet remain pre-commercial behind four barriers: hydrothermal cycling stability, scalable synthesis, heat-exchanger coating, cost. MOF-303: ~7× silica gel’s uptake under cooling conditions; a standard two-bed model on its measured kinetics predicts COP 0.62–0.63 with SCP 440 → 970 W·kg⁻¹ over 75–90 °C, high-COP and high-SCP points coinciding — but predicted, not a tested prototype; ORNL’s 827 W·kg⁻¹ is the best measured hardware (Ben-Mansour).
  • Activated carbon / methanol, ethanol, ammonia, CO₂ — low heats of adsorption (1800–2000 kJ·kg⁻¹) favour COP; AC–ammonia runs at ~1600 kPa for ≥ 200 °C sources; AC–methanol regenerates ~100 °C, decomposes above 150 °C (Elsheniti). Carbon/CO₂ for sub-zero or pressurised duty — CO₂ (R744).
  • Chemical and composite sorbents — metal chlorides bind ammonia at ~1 kg·kg⁻¹ with 40–80 °C desorption but swell; salt-in-matrix composites fix that and dominate Frazzica’s screening (below) — Composite Salt Sorbents.

Selection is not obvious. Tatlier 2023 (registry) argues the “best” sorbent depends on the exact drive/sink/chilled temperatures — isotherm shape against the operating window matters more than headline uptake, the lesson the pore-engineered silica study repeats in Limitations & Mitigations.

Squeezing more out: bed & heat-transfer enhancements

Adsorbents transfer heat and mass badly, so much of the gain comes from the bed and cycle, not the material:

  • Cycle-level recovery: heat recovery lifts COP up to 25 %, adding mass recovery > 10 % more (Elsheniti); cycle time has an optimum — longer raises COP while SCP peaks then falls (Sosnowski).
  • Granular-bed additives: 15 wt% copper in silica gel raised both COP and SCP; CNTs were most promising (smallest uptake penalty, fastest sorption), aluminium cut sorption time, copper delaminated in another study (Sosnowski). Hua’s chiller used a silica-gel/copper-chip composite — a cheap middle path between granular and coated beds.
  • Steam pre-heating cut desorption time from ~300 s to ~30 s (Sosnowski); fluidised beds raise the bed-to-exchanger coefficient (LSTM-modelled at > 0.95 agreement).
  • Coated and foam beds: 20-PPI metal foam COP 0.75 / 827 W·kg⁻¹; CPO-27(Ni) on copper-foam tubes 1922–5130 W·kg⁻¹ specific heating power (ORNL). Jacobucci’s boiling-assisted channel templating (BACT) coating of AQSOA-Z02 on copper foam reached 1875 W·kg⁻¹ of Z02 (120 s cycle), projecting 375 W·kg⁻¹ for the whole bed vs 200 for immersion-dried coatings. Functionalisation compounds it: (CH₃)₂-MOF-801 +40 % SCP / +30 % COP; MIL-101(Cr)/CaCl₂ +68 % SCP with desorption cut 110 → 90 °C (ORNL).
  • Evaporator turbulators (registry: Kalawa): twisted tape COP +10.5 % / SCP +9 %; Z-type +41 % / +47 %; coated vs bare tubes +20 % / +47.6 %. Modelling caveat (ORNL): constant-conductivity assumptions can give ~30 % SCP error.

The architectural ceiling is the third-generation adsorption heat pump (Gado, Palomba, Mittelbach, Frazzica, Renewable Energy 2026): evaporation and condensation directly at the adsorber, liquid-phase refrigerant distribution, no large vacuum valves, targeting thermal COP 0.65, electric COP > 20, > 17 kW·m⁻³ at 75–90 °C against the 246 W·kg⁻¹ / 0.48 baseline — field sites in Commercial Adsorption Chillers.

The COP-fairness argument (why 0.5 isn’t as bad as it looks)

A thermal COP of ~0.5 looks poor beside a compressor’s 3–5, but the currencies differ. Elsheniti: thermal COPs of ~0.6 are “low,” yet the electrical COP of an adsorption chiller can reach ~10 (9 W for 7.5 kW). Dias 2022 (registry) makes the primary-energy correction explicit — at ~40 % grid thermal-to-electric efficiency, vapor-compression’s primary-energy COP drops far below nameplate, so “AC systems’ COP is not a real disadvantage.” With waste or solar heat (primary-energy cost ≈ 0) the comparison tilts further. Adsorption wins on free heat, loses on first cost and footprint.

Derating: hot climates and fluctuating sources (read this for San Antonio)

  • Hot condenser. The modelled machine giving COP 0.50 / SCP 44 at a 30 °C condenser falls to 0.31 / 26 at 42 °C (registry: Dias) — ~40 % capacity loss on the hottest afternoons. Hua’s measured sensitivity ranking is chilled-water > hot-water > cooling-water temperature, so the chilled setpoint is the cheapest lever and a good rejection sink the next.
  • Fluctuating drive. With emulated fuel-cell heat at 70–76 °C swinging ±2.5–4 °C, a built double-bed chiller lost up to ~7 % capacity and ~19 % COP versus steady ratings; keep fluctuation within ±3 °C, and a stepped ramp (70 → 73 → 76 °C) recovers ~4.6 % COP (Hua 2025; abstract-level figures, flagged for full-text check). Solar and waste heat — the sources that justify adsorption — are fluctuating sources.

Implication: in San Antonio pair the chiller with a geothermal loop or a well-sized wet tower, buffer the drive heat, and size for the design-day condenser, not the average. Curves and the mitigation stack: Limitations & Mitigations.

Economics

  • Solar residential (registry: Motamedi, Alahmer): optimal collector ~38 m², feasible below $700/m², life-cycle savings ~$3,500, payback ~11 years; the Shanghai Green Building case gives 5–8 years, ~2 with domestic hot-water credit (Adsorption Cooling).
  • Jordan (registry): ~$5,000/yr operating savings, $20–30k/yr for larger plants.
  • Industrial waste heat: RE-WITCH’s 80 kW pharma site projects 110 MWh/yr saved (29 % of cooling electricity) — Commercial Adsorption Chillers.
  • First cost is the barrier — low SCP means adsorbent mass and exchanger area per kW; ORNL’s four MOF barriers and the registry’s cost of manufacturing adsorption chillers study target it. Free driving heat is what closes the payback.

Cogeneration: cooling + fresh water, and upgrading heat

Adsorption desalination + cooling (ADC). Conventional AD runs on 55–85 °C heat, recovers > 85 % of feed via sub-ambient evaporation and yields < 50 ppm TDS water (Shahzad). Specific daily water production (registry, prior compile): silica gel ~4 kg·kg⁻¹, UiO-66 24 → 40 with functionalisation, CPO-27(Ni) 6.8 kg·kg⁻¹ potable alongside SCP 200 W·kg⁻¹; the pore-engineered silica figures are in Limitations & Mitigations. Pressure-driven AD (PDAD) regenerates the beds with 2–5 bar motive steam through a thermal vapour compressor instead of hot water — silica gel dumped in a tank, no bed heat exchangers — and coupling its 65 °C discharge steam to multi-effect distillation lets MED run from 65 °C down to 7 °C, doubling its effects: +22 % water over the earlier MEDAD hybrid, > 2× over plain MED, ~20 % thermodynamic efficiency vs 10–13 % for RO/MED/MSF (Shahzad 2023, KAUST). MIL-160 harvested 0.31–0.33 g·g⁻¹ water from air per cycle at 80 °C regeneration (Sosnowski).

Adsorption heat transformers (AdHT). Run the cycle to upgrade 40–80 °C waste heat instead of chilling. Frazzica’s ten-pair screening: two-bed heat recovery always beats one-bed; a realistic 5 K exchanger approach makes 40 °C heat nearly useless (~10 K lift at COP 0.3) while 80 °C heat lifts up to ~52 K; and composite salt sorbents dominate — MWCNT-LiCl/methanol ~1660 kJ·kg⁻¹ at COP 0.50 for a 10 K lift from 50 °C heat, SGST-LiCl/water ~700, ammonia/carbon pairs never above ~380, bare adsorbents too weak above 0 °C ambient.

Building integration: the CoolSkin façade

CoolSkin (SFB1244 Stuttgart; registry: Greiner 2022) builds a closed low-pressure adsorption system into the façade: solar heat regenerates the adsorber to ~100 °C and the panel delivers 100–150 W of cooling per m² for ~12 hours — the envelope as cooling machine, bridging to the radiative/façade and envelope threads.

Commercial reality check (absorption vs adsorption)

Much of what vendors sell into “waste-heat cooling” is absorption (Thermax, Johnson Controls/YORK LiBr machines), not adsorption. Absorption has higher COP and power density and owns large commercial waste-heat cooling; adsorption wins on lower drive temperature, no crystallisation or corrosion, no moving liquid, solid-state simplicity at smaller scale. For 601 it is a real down-select — absorption if scale and a steady 80–90 °C stream justify it, adsorption for lower-grade or intermittent heat. See Absorption Cooling and Commercial Adsorption Chillers.

What this means for 601 Delaware

  • Expect COP ~0.45–0.55, SCP ~160–250 W·kg⁻¹ from silica-gel/water on ~80 °C roaster heat; EMM-8- or MFM-300-class sorbents keep COP near 0.8 if the heat is only 60–70 °C.
  • Size adsorbent mass from SCP and the roaster’s actual exhaust enthalpy and duty cycle, derate ~40 % for the August condenser and a further ~7 % capacity / ~19 % COP for a fluctuating roaster schedule unless the heat is buffered.
  • Treat MOF-303 and cascaded-MOF figures as the ceiling the working pair can buy, not a procurement spec, until chiller-duty cycle life is shown.
  • Desalination cogeneration, heat upgrading and the CoolSkin façade are on-brand feasibility options, not lab curiosities.

See also

Sources