An aerial view shows cooling vent fans on the roof of a Digital Realty data center in Ashburn, Virginia on November 12, 2025.
ANDREW CABALLERO-REYNOLDS/AFP via Getty Images
Every data center and factory on Earth discards heat it cannot use — warm exhaust streams sitting at around 40°C (104°F), too cool for any established cooling technology to do anything with. South Korea’s national machinery institute announced Wednesday that it has cracked that barrier, demonstrating a system that takes exactly that low-temperature waste and turns it directly into cooling energy. If the performance figures hold at scale, the technology could change the economics of data center cooling fundamentally.
The Korea Institute of Machinery and Materials (KIMM) unveiled a chemical adsorption heat pump integrated with a novel electrochemical compressor — a combination the institute says has no commercial equivalent. A 10 kW prototype, manufactured by Samjung Tech Co., is undergoing demonstration testing as of this announcement.
Why 40°C (104°F) Has Been a Wall for Cooling Technology
The challenge is thermodynamic. Adsorption cooling systems — a well-established technology for using waste heat to drive refrigeration instead of electricity — have a practical floor. Conventional adsorption cooling systems require a heat source above 70°C (158°F) to function. Below that temperature, the solid adsorbent material in the system cannot be regenerated effectively: it cannot release the refrigerant it has captured, so the cooling cycle stalls.
Most server halls, light-manufacturing exhaust lines, and commercial building plant rooms operate between 35°C (95°F) and 50°C (122°F) — a band that conventional adsorption cooling has always straddled but never been able to harvest. That waste heat has been vented to the atmosphere instead.
The scale of the missed opportunity is substantial. Data centers globally consumed roughly 415 TWh of electricity in 2024 — approximately 1.5% of worldwide electricity demand — and that figure is on track to nearly double to around 945 TWh by 2030 as AI workloads drive rack power density higher. Cooling accounts for 30–40% of total data center facility power. Nearly all of the electricity these facilities consume ultimately becomes heat, and most of that heat has historically been discharged to the air outside.
How KIMM Broke the Floor
The research was led by Dr. Young Kim, Principal Researcher at KIMM’s Heat Pump Research Center, in collaboration with Professor Minseong Kim and Professor Dongkyu Kim of Chung-Ang University, with manufacturing support from Samjung Tech. The full details appear in KIMM’s official announcement.
The system pairs two technologies. KIMM developed the adsorption heat pump side — the module that extracts energy from low-temperature exhaust and converts it into a cooling effect. Chung-Ang University developed the electrochemical compressor that drives the system’s pressure cycle.
The adsorption side works by cycling a solid chemical adsorbent through two phases: during adsorption, the material captures refrigerant vapor from the evaporator, creating a low-pressure zone that pulls the refrigerant to evaporate and generate cold; during desorption, the same material is heated by the waste heat source to release the captured refrigerant at elevated pressure, which then condenses and returns to the evaporator. In conventional systems, that desorption phase demands high-temperature driving heat — the adsorbent simply will not release its captured refrigerant at 40°C (104°F) with standard sorbent-refrigerant pairs. KIMM redesigned the adsorption bed to achieve stable desorption at that lower temperature, though the specific sorbent material used in the prototype has not been publicly disclosed.
The result: the new system successfully demonstrated cooling operation using waste heat at 40°C (104°F), approximately 30°C (54°F) below the threshold at which conventional adsorption cooling systems can function.
What Is an Electrochemical Compressor?
The other half of the system is where the technology departs most sharply from conventional refrigeration. Traditional compressors — whether in a home air conditioner or an industrial chiller — work mechanically: a piston, scroll, or rotary element compresses refrigerant gas, generating noise, vibration, heat, and wear that requires lubrication oil and scheduled maintenance. Electrochemical compressors do none of this.
The electrochemical approach, which has been studied since General Electric pioneered the concept in the 1980s and which has seen substantial refinement for ammonia-based systems in recent years, uses an electrical potential across an ion-exchange membrane to move refrigerant molecules without any mechanical action. In KIMM’s ammonia-based system, the key steps are:
At the anode side of the membrane electrode assembly, a DC voltage ionizes the ammonia-hydrogen working fluid into ammonium ions (NH₄⁺). Those ions are conducted through a perfluorosulfonic acid (PFSA) membrane — the same basic membrane architecture used in hydrogen fuel cells — from the low-pressure side to the high-pressure side. At the cathode, the ions are reduced back to ammonia gas at elevated pressure. The net effect is identical to mechanical compression — refrigerant gas moved from low to high pressure — but accomplished entirely by electrochemical work, with no moving parts, no lubrication, and no significant noise or vibration.
A comprehensive 2025 electrochemical heat pump review published in November 2025 confirmed that these systems can achieve 10% to 30% higher energy efficiency than conventional vapor compression systems, with cooling coefficients of performance ranging from 3.5 to 14.3 under standard conditions. The same review identified the field’s main barriers: membrane degradation over time, electrode fouling, and capital costs that remain higher than mechanical alternatives. It also noted a documented lack of prolonged closed-loop system testing — a gap that applies to the KIMM prototype as well.
One key scale-up challenge for any electrochemical compressor is membrane area: more flow requires larger membranes. KIMM’s announcement specifically highlighted progress on a large-area ammonia compressor through the Chung-Ang collaboration, describing it as a step toward proof-of-concept validation for commercial scale.
How the Adsorption Bed Performance Compares
Beyond the temperature threshold, KIMM reports a significant advance in the adsorption bed itself, claiming a specific cooling power (SCP) of 346.5 watts per kilogram — a figure the team describes as more than twice the performance of comparable international technologies and as a world-leading result.
SCP measures how much cooling output a given mass of adsorbent material can produce. It matters because the adsorbent bed is the heaviest, bulkiest component in an adsorption system; a higher SCP means the same cooling capacity can be achieved with a physically smaller and lighter adsorption module — a critical factor for retrofitting into existing data center facilities or factory footprints where space is constrained.
Commercial silica gel-water adsorption chillers typically achieve SCP figures in the range of 100 to 170 W/kg; advanced composite adsorbent systems in academic research have reported up to approximately 200 W/kg. KIMM’s claimed 346.5 W/kg, if confirmed by independent testing, would represent roughly two to three times the performance of current commercial systems. That independent confirmation has not yet occurred — the figure is taken from KIMM’s own press release, and third-party laboratory validation will be needed before the claim can be treated as an established benchmark. The team has produced 74 SCI papers and holds 29 registered patents through the project, indicating substantial prior research output from which the prototype emerges.
The absence of mechanical moving parts also makes the system well-suited to locations where conventional compressors would be impractical, including hospitals, schools, and residential buildings.
What a Closed Loop Could Mean for Data Centers
The most significant architectural implication of the KIMM system — one that the announcement does not explicitly name — is the possibility of a partially self-sustaining cooling loop. A data center consumes electricity, which becomes heat. That heat is currently cooled by air handling units that themselves consume more electricity. A system that converts 40°C (104°F) exhaust from the cooling plant into additional cooling energy creates a feedback path: the more heat the cooling system rejects, the more driving energy it has available for the adsorption cycle.
This is not a perpetual motion claim — thermodynamics does not bend. The adsorption cooling cycle’s coefficient of performance is well below one in electricity terms; external electrical input for the electrochemical compressor is still required. But the system’s driving energy is waste heat rather than purchased electricity, meaning that a meaningful fraction of the cooling load could be powered by the facility’s own thermal exhaust rather than the grid. For data centers in warm climates where cooling runs year-round at high load, that fraction could represent significant energy cost savings and a direct contribution to regulatory compliance.
Germany’s Energy Efficiency Act, which now mandates a minimum Energy Reuse Factor of 10% for new data centers commissioned from July 2026, rising to 15% from July 2027 and 20% from July 2028, has made exactly this kind of self-referential heat reuse a legal requirement rather than an optional sustainability measure. Data center operators in Germany designing facilities today must demonstrate a plan for waste heat reuse before going online — and the dominant pathway currently available (piping warm water into district heating networks) requires heat temperatures that many facilities cannot reliably achieve at scale. A system that converts low-grade exhaust directly into on-site cooling would satisfy the ERF requirement while eliminating the infrastructure cost of connecting to an external heat network.
What Still Needs to Happen
Dr. Young Kim described the path forward in a statement: “We are currently conducting demonstration tests with the 10 kW-class prototype. Building on these tests, we plan to pursue further research to advance the technology for field demonstration.”
The gap from 10 kW (10,000 W) prototype to a commercial data center cooling plant — which typically operates at hundreds of kilowatts to tens of megawatts — is substantial. Three engineering challenges stand between the current prototype and a commercially deployable system.
The first is scale-up validation. Membrane electrode assemblies function differently at large area than at cell scale; heat exchanger design must be re-engineered for higher mass flow rates; and the adsorption bed’s thermal cycling behavior under continuous commercial operation has not been demonstrated. The critical review published in 2025 specifically flagged the lack of prolonged closed-loop system testing in electrochemical heat pump research as a field-wide gap — the KIMM system has not yet generated the longevity data that commercial customers would require.
The second is independent performance verification. The SCP claim of 346.5 W/kg and the 40°C operational threshold are from KIMM’s own press release. These figures need to be validated by an independent laboratory before they can be cited in procurement decisions.
The third is ammonia safety engineering. Ammonia is a natural refrigerant with zero global warming potential, making it the right choice from a regulatory standpoint as the Kigali Amendment phases down high-GWP synthetic refrigerants globally. It is also toxic at high concentrations and flammable in air, which means commercial deployments — particularly in enclosed data center environments — require leak detection systems, ventilation design, and safety protocols that add cost and complexity compared to conventional HFC systems.
The team’s stated next step is field demonstration beyond the current prototype testing, with a potential commercialization vehicle.
How Does Waste Heat Cooling Work?
How does a data center benefit from this technology?
A data center using this system would route some of its warm exhaust streams — typically 35–45°C (95–113°F) water or air — into the KIMM heat pump rather than rejecting that heat to atmosphere. The heat pump uses that thermal input to drive a cooling cycle, producing chilled water or cooled air that reduces the demand on the facility’s conventional chillers. The more cooling load the facility generates, the more driving heat it has available — creating a partial feedback loop that reduces purchased electricity for cooling.
What is specific cooling power, and why does it matter?
Specific cooling power (SCP) measures how much cooling output an adsorption system can produce per kilogram of adsorbent material. Higher SCP means a smaller, lighter system can deliver the same cooling capacity — which matters because data center retrofits and industrial installations have limited floor space. KIMM claims 346.5 W/kg, which it says is more than double current international benchmarks. That figure needs independent verification.
Why hasn’t this been possible before?
Adsorption cooling has required driving heat above 70°C (158°F) because the sorbent materials in conventional systems — typically silica gel or zeolite paired with water — do not desorb refrigerant effectively at lower temperatures. KIMM’s claimed breakthrough is a redesigned adsorption bed that achieves stable desorption at 40°C (104°F), combined with an electrochemical compressor that can handle the resulting lower-pressure refrigerant cycle without mechanical parts.
What is the “noiseless” aspect, and who benefits from it?
The electrochemical compressor has no mechanical moving parts — no pistons, rotors, or scroll elements — and therefore generates essentially no noise or vibration. This matters most for deployment locations where mechanical compressors are currently impractical: hospitals, university buildings, office towers, and residential facilities. For data centers, the noise benefit is secondary; the energy efficiency advantage is the primary value proposition.
Frequently Asked QuestionsWhat temperature does the KIMM system need to produce cooling?
The KIMM prototype demonstrated cooling using waste heat at approximately 40°C (104°F) — roughly 30°C (54°F) below the threshold at which conventional adsorption cooling systems can function. Most factory exhaust streams and data center cooling returns fall within this temperature range, which is precisely why it had been an untapped resource until now.
How does the electrochemical compressor work without moving parts?
The electrochemical compressor uses a direct-current voltage applied across an ion-exchange membrane. At the anode side, ammonia molecules in the working fluid are ionized into ammonium ions, which travel through the membrane to the cathode side under the electrical potential. At the cathode, the ions are reduced back to ammonia gas at elevated pressure. The pressure difference created by this ionic transport is what drives the refrigeration cycle — no pistons, no rotating shaft, and no lubrication oil required. The technology achieves compression efficiency above 90% in controlled laboratory conditions, compared to roughly 65% for typical mechanical compressors.
Has the world-record performance claim been independently verified?
Not yet. The specific cooling power figure of 346.5 W/kg is taken from KIMM’s own press release and has not been independently confirmed by a third-party laboratory as of this writing. The claim is consistent with the direction of recent research into advanced composite adsorbent materials, and KIMM’s broader research portfolio — 74 SCI papers and 29 registered patents through the KETEP Alchemist Project — lends institutional credibility to the team. Independent verification will be required before the figure can be treated as a certified benchmark.
What are the main hurdles before this reaches commercial data centers?
Three: scale-up from a 10 kW prototype to megawatt-class commercial systems; prolonged closed-loop system testing to establish membrane and electrode durability; and ammonia safety engineering for enclosed data center environments. The team’s stated next step is a field demonstration program. Commercial deployment is not imminent, but the prototype result establishes the technological feasibility of the approach.