A Stirling-type thermoacoustic engine has reached 49% of the Carnot limit, a record for a machine whose only working part is a gas column vibrating with sound (Tijani and Spoelstra, Journal of Applied Physics 2011). That figure caps a climb that began when Los Alamos showed in 1999 that a no-moving-parts engine could convert 30% of its heat into sound power, beating the 0.10 to 0.25 range of earlier no-moving-parts engines (Backhaus and Swift, Nature 1999). Generators built on the same physics now deliver 15.0 kW of electricity (Chen et al., Cell Reports Physical Science 2024), and the pulse tube cryocooler, a cousin of the technology, has held detectors at 58 K on a NASA satellite for over 17 years (NASA JPL AIRS Thermal System page). We aggregated data from Los Alamos National Laboratory papers in Nature and JASA, the Energy Research Centre of the Netherlands, the Chinese Academy of Sciences, Zhejiang University, NIST, NASA and JPL, and peer-reviewed work in Applied Energy, Energy, Cryogenics and Scientific Reports. The same high-amplitude sound fields that move heat here also lift droplets in mid-air, a topic we track in our acoustic levitation statistics.
Two terms come up throughout. Percentage of Carnot efficiency compares a real machine with the theoretical ceiling: the Carnot efficiency is the most work any engine can extract from heat flowing between a given hot and cold temperature, so an engine at 41% of Carnot captures 41% of what physics allows at those temperatures. Standing-wave vs traveling-wave describes how the sound moves: in a standing-wave device the wave bounces back and forth in a straight tube and the gas exchanges heat with the walls with a delay, which wastes energy; in a traveling-wave device the wave circulates through a loop and a fine metal mesh (the regenerator), so the gas follows a near-ideal Stirling cycle and loses far less.
TL;DR
- A Stirling-type thermoacoustic engine reached 49% of Carnot efficiency, a record for the technology (Tijani and Spoelstra, Journal of Applied Physics 2011).
- The Los Alamos traveling-wave engine converted 30% of heat into acoustic power (41% of Carnot) at 710 W, more than 50% better than earlier standing-wave designs (Backhaus and Swift, JASA 2000).
- A hybrid thermoacoustic electric generator delivered 15.0 kW, with 39.2% peak efficiency at 11.1 kW (Chen et al., Cell Reports Physical Science 2024).
- A three-stage traveling-wave generator produced 4.69 kW at 15.6% thermal-to-electric efficiency (Bi et al., Applied Energy 2017).
- A multi-stage looped engine started oscillating with a temperature difference of just 17 °C (Jin et al., Scientific Reports 2017).
- A heat-driven thermoacoustic refrigerator hit a COP of 1.12 with 2.53 kW of cooling, 2.7 times the previous best (Xiao et al., Cell Reports Physical Science 2024).
- A coaxial thermoacoustic-Stirling cooler reached 25% of Carnot and -54 °C (Tijani and Spoelstra, Cryogenics 2008).
- Pulse tube refrigerators have reached 24% of Carnot at 80 K and temperatures as low as 2 K (Radebaugh, NIST 2003).
- Webb’s MIRI detectors run nominally around 6.2 K, cooled by a three-stage pulse tube precooler and a Joule-Thomson loop (NASA Webb Cryocooler page 2024).
- Conventional MRI magnets hold about 1,500 to 2,000 liters of liquid helium, while sealed designs need as little as 7 or 0.7 liters (Manso Jimeno et al., MRI magnet review 2022).
1. Heat-to-Acoustic Efficiency: How Close the Engines Get to Carnot
The engine is the heart of every thermoacoustic system: heat goes in at one end of a porous core, and sound power comes out. The 1999 Los Alamos result, 30% thermal efficiency or 41% of Carnot, put an engine with no moving parts inside the range of mainstream machinery: the same Nature paper put internal combustion engines at 0.25 to 0.40 and piston-driven Stirling engines at 0.20 to 0.38 (Backhaus and Swift, Nature 1999). The detailed study that followed showed the trade-off every designer still faces: push for maximum power (890 W) and efficiency drops to 0.22 (Backhaus and Swift, JASA 2000).
The ceiling has moved slowly since. The Dutch engine at 49% of Carnot (Tijani and Spoelstra, Journal of Applied Physics 2011) remains the most cited record, and newer work focuses on scale: a 2025 four-stage multi-looped engine produced 4 x 2,100 W of acoustic power, and its designers predict 60 kW at full design pressure and heating temperature (Yu et al., Energy 2025). The question is no longer whether sound can carry heat efficiently, it is whether that efficiency survives scale-up.
| Metric | Value | Source |
|---|---|---|
| Thermal efficiency of the Los Alamos thermoacoustic-Stirling engine at its most efficient point | 0.30 (30%), equal to 41% of Carnot | Backhaus and Swift, JASA 2000 |
| Acoustic power at that most efficient point | 710 W | Backhaus and Swift, JASA 2000 |
| Maximum acoustic power of the same engine, and its efficiency at that point | 890 W at 0.22 efficiency | Backhaus and Swift, JASA 2000 |
| Performance gain of this traveling-wave engine over earlier standing-wave engines | More than 50% | Backhaus and Swift, JASA 2000 |
| Efficiency range of earlier heat engines with no moving parts | 0.10 to 0.25 | Backhaus and Swift, Nature 1999 |
| Record efficiency of a Stirling-type thermoacoustic engine, as a share of Carnot | 49% | Tijani and Spoelstra, Journal of Applied Physics 2011 |
| Cascade engine (1 standing-wave stage plus 2 traveling-wave stages): acoustic power and efficiency | Up to 2 kW at 20% | Gardner and Swift, JASA 2003 |
| Four-stage multi-looped engine (2025): acoustic power output efficiency in experiments | 16.15%, with 4 x 2,100 W of acoustic power | Yu et al., Energy 2025 |
Context note: the cascade engine’s authors expected that extending their design should yield 35% to 40% of Carnot (Gardner and Swift, JASA 2003). Most recent available data for the single-engine record: Tijani and Spoelstra, 2011. Resonator and impedance design are also the core tools of acoustic metamaterials, the other field where engineers shape sound fields rather than suppress them.
2. Thermal-to-Electric Generators: From Watts to 15 Kilowatts
Converting sound power into electricity adds a second loss, the alternator, so thermal-to-electric figures always run below heat-to-acoustic ones. The 2004 Los Alamos and Northrop Grumman generator converted 18% of its heat into electricity, against 7% for the radioisotope thermoelectric generators then used on spacecraft (ScienceDaily, reporting Los Alamos National Laboratory, 2004). The engine inside that generator ran at 35.7% of Carnot (Yu, Jaworski and Backhaus, Proc. IMechE Part A 2010, describing Backhaus et al.).
Chinese groups then pushed the power level into the kilowatts. A 1 kW-class traveling-wave generator reached 19.8% peak efficiency (Wu et al., Applied Energy 2014), a three-stage machine delivered 4.69 kW (Bi et al., Applied Energy 2017), and a hybrid design with a post-positioned gas spring reached 15.0 kW (Chen et al., Cell Reports Physical Science 2024). The hybrid’s 39.2% peak efficiency is the standout number, but it uses a piston, so it is not a pure no-moving-parts machine.
| Metric | Value | Source |
|---|---|---|
| Traveling-wave thermoacoustic generator for spacecraft: share of heat converted to electricity vs radioisotope thermoelectric generators | 18% vs 7% | ScienceDaily / Los Alamos National Laboratory, 2004 |
| 1 kW-class traveling-wave generator: maximum electric power and efficiency at that power | 1,043 W at 17.7% | Wu et al., Applied Energy 2014 |
| Same generator: maximum thermal-to-electric efficiency and power at that point (4.0 MPa, 650 °C heating) | 19.8% at 970 W | Wu et al., Applied Energy 2014 |
| Acoustically matched traveling-wave generator (helium at 3.16 MPa): maximum power and best efficiency | 750.4 W; 0.163 | Wang et al., Energy 2016 |
| Three-stage traveling-wave generator (6 MPa helium, 650 °C heating): maximum power and efficiency | 4.69 kW at 15.6% | Bi et al., Applied Energy 2017 |
| Same three-stage generator: maximum efficiency and power at that point | 18.4% at 3.46 kW | Bi et al., Applied Energy 2017 |
| Hybrid generator with post-positioned gas spring: maximum electric output | 15.0 kW | Chen et al., Cell Reports Physical Science 2024 |
| Same hybrid generator: peak efficiency and power at that point | 39.2% at 11.1 kW | Chen et al., Cell Reports Physical Science 2024 |
Outlier note: new conversion paths are still early. A 2026 system that places liquid metal in the resonator of a Backhaus-Swift engine reached 11.0% thermal efficiency but only 13.5 W of average electric power at an 885 K heating temperature (Chen et al., The Innovation Energy 2026).
3. Onset Temperature and Low-Grade Waste Heat
The onset temperature difference is the smallest temperature gap across the engine core at which the gas starts to oscillate on its own; below it, the engine produces nothing. It decides whether a machine can run on factory exhaust or only on a flame. Zhejiang University started a multi-stage looped engine with a gap of only 17 °C, heating at 29 °C against a 12 °C cold end (Jin et al., Scientific Reports 2017). The catch is output: the same paper’s three-stage generator reached a maximum thermal-to-electric efficiency of 1.03% at 94 °C, about 5% of Carnot.
The low-cost end shows the same trade-off. A rural generator prototype built with a loudspeaker as alternator started at an onset difference of about 210 °C and delivered 5.17 W from 800 W of heat, an overall 0.65% (Yu, Jaworski and Backhaus, Proc. IMechE Part A 2010). Low-temperature thermoacoustics can start almost anywhere; making it pay is the open problem.
| Metric | Value | Source |
|---|---|---|
| Lowest onset temperature difference of an asymmetric multi-stage looped engine (CO2, 1 or 1.5 MPa) | 17 °C (heating 29 °C, cold end 12 °C) | Jin et al., Scientific Reports 2017 |
| Maximum thermal-to-electric efficiency of the three-stage low-temperature generator, and heating temperature | 1.03% at 94 °C (about 5% of Carnot) | Jin et al., Scientific Reports 2017 |
| Heating temperature typically needed by Stirling-type thermoacoustic engines for a comparable pressure ratio | Above 600 °C | Jin et al., Scientific Reports 2017 |
| de Blok multi-stage looped engine: efficiency relative to Carnot with a 99 °C heat source | 38% | Jin et al., Scientific Reports 2017 (citing de Blok) |
| de Blok traveling-wave engine: onset temperature difference | 65 K | Yu, Jaworski and Backhaus, Proc. IMechE Part A 2010 |
| Low-cost rural generator: onset temperature difference, and electricity from 800 W of heat | About 210 °C; 5.17 W (0.65% overall) | Yu, Jaworski and Backhaus, Proc. IMechE Part A 2010 |
| Air-filled looped-tube generator with a commercial audio loudspeaker as alternator: electric power | 11.6 W | Yu, Jaworski and Backhaus, Applied Energy 2012 |
Context note: one 2025 waste-heat system reports 170 W of electricity plus 74.73 W of cooling at -10 °C from a 500 °C heat source, with a thermal-to-electrical efficiency of 29.22% and an overall exergy efficiency of 54.01% as defined by its authors (Umar et al., Sustainable Energy Technologies and Assessments 2025). The authors also claim it runs 100 to 200 °C cooler than conventional thermoacoustic systems. See the Data watch note for why that efficiency figure is not directly comparable with section 2. Most recent available data for the 17 °C onset record: Jin et al., 2017. The 11.6 W loudspeaker prototype is documented in Applied Energy 2012.
4. Thermoacoustic Refrigerators: COP and Percent of Carnot
Run the engine backwards and sound pumps heat: that is a thermoacoustic refrigerator. Its efficiency is reported as COP (coefficient of performance), the cooling delivered per unit of energy put in, or as COP relative to Carnot. The first generation of chillers stayed between 8.5% and 19% of Carnot, from the US Navy’s shipboard SETAC unit to the Penn State device completed in 2005 (Kajurek, Rusowicz and Grzebielec, SN Applied Sciences 2019). The traveling-wave coaxial cooler at 25% of Carnot nearly doubled the best standing-wave result (Tijani and Spoelstra, Cryogenics 2008).
The newest numbers come from heat-driven machines, where a thermoacoustic engine drives the cooler directly with no electricity at all. A bypass design reached a COP of 1.12 at a 450 °C heat source, which its authors say beats adsorption chillers and single-effect absorption chillers (Xiao et al., Cell Reports Physical Science 2024). That framing matters: the competitor is not the electric vapor-compression unit in a home (see our heat pump condenser noise statistics), it is the heat-fired chiller that already runs on waste heat.
| Metric | Value | Source |
|---|---|---|
| Hofler refrigerator (1986): efficiency relative to Carnot with a 3 W heat load | 12% of Carnot | Kajurek et al., SN Applied Sciences 2019 (review) |
| STAR thermoacoustic refrigerator: cooling capacity at a 50 °C temperature drop, and relative efficiency | 3 W; 16% | Kajurek et al., SN Applied Sciences 2019 (review) |
| US Navy SETAC shipboard chiller: cooling power and efficiency relative to Carnot | Nearly 420 W; 8.5% | Kajurek et al., SN Applied Sciences 2019 (review) |
| Penn State device (helium-argon at 30 bar): efficiency relative to Carnot | 19% | Kajurek et al., SN Applied Sciences 2019 (review) |
| Coaxial thermoacoustic-Stirling cooler: COP relative to Carnot and lowest temperature | 25%; -54 °C | Tijani and Spoelstra, Cryogenics 2008 |
| Heat-driven refrigerator with bypass configuration: COP and cooling power (450 °C heat, 35 °C ambient, 7 °C cooling) | COP 1.12; 2.53 kW | Xiao et al., Cell Reports Physical Science 2024 |
| Same machine: COP vs previous best heat-driven thermoacoustic refrigerator | 2.7 times higher | Xiao et al., Cell Reports Physical Science 2024 |
| Follow-up heat-driven cooler: COP and cooling power at 550 °C heating | COP 1.34; 2.37 kW | Xiao et al., The Innovation Energy 2024 |
Outlier note: the 1.34 figure is reported as a 240% COP increase over the best prior heat-driven result without the new configuration (Xiao et al., The Innovation Energy 2024). The research team also projects a potential COP of 2 above 800 °C, a simulation, not a measurement (TechXplore, reporting the Chinese Academy of Sciences, 2024).
5. Pulse Tube Cryocoolers: Efficiency, Lowest Temperatures and Heat-Driven Versions
A pulse tube cryocooler is a refrigerator for cryogenic temperatures (the deep-cold range used to liquefy gases) whose cold end has no moving parts: an oscillating gas column and a small orifice do the work a piston would otherwise do. It is the commercial success story of oscillating-flow thermodynamics. Pulse tubes have reached 24% of Carnot at 80 K and temperatures as low as 2 K (Radebaugh, NIST, Pulse Tube Cryocoolers 2003).
The cost of going colder is brutal. A commercial two-stage unit produced 0.5 W at 4.2 K for about 5.5 kW of compressor input; the math (5,500 W / 0.5 W) is roughly 11,000 W of electricity per watt of cooling. Heat-driven versions skip the electric compressor entirely: a 2025 looped system with four thermoacoustic engines reached 46.6 K, a record for a 100 Hz system, and recovered waste heat in four cascaded stages (Zhang et al., Energy 2025).
| Metric | Value | Source |
|---|---|---|
| Highest pulse tube efficiency at 80 K, as a share of Carnot | 24% | Radebaugh, NIST 2003 |
| Lowest temperature reached by pulse tube refrigerators | About 2 K | Radebaugh, NIST 2003 |
| Commercial two-stage 4 K pulse tube: cooling at 4.2 K, and compressor input power | 0.5 W (or 4.8 L/day of liquid helium); about 5.5 kW | Radebaugh, NIST 2003 |
| Vibration of 4 K pulse tubes vs Gifford-McMahon refrigerators | Over an order of magnitude less | Radebaugh, NIST 2003 |
| Thermoacoustically driven pulse tube for natural gas: refrigeration power, and share of the gas burned to liquefy the rest | About 2 kW at 120 K; about one-third | Radebaugh, NIST 2003 |
| NIST 120 Hz pulse tube: efficiency (referred to input PV power) and net cooling at 80 K | About 22.6% of Carnot; 3.35 W | NIST, 120Hz Pulse Tube Cryocooler 2007 |
| Heat-driven pulse tube cooler with four thermoacoustic engines: lowest temperature, and cooling at 74.031 K | 46.6 K; 3 W | Zhang et al., Energy 2025 |
| Same system, four-stage cascade waste heat recovery: recovery efficiency and cooling power at 75.5 K | 0.66; 2.35 W | Zhang et al., Energy 2025 |
Context note: the NIST 120 Hz cooler cooled from ambient to 80 K in 5.5 minutes (NIST, 2007). The natural gas liquefier figures (about 600 L/day) describe a demonstration unit first shown by NIST and Los Alamos in 1990; most recent available data: Radebaugh, 2003.
6. Pulse Tubes in Space and in MRI Scanners
Space is where the no-moving-parts cold end pays for itself, because nobody can service a cooler in orbit. The AIRS instrument on NASA’s Aqua satellite flew the first pulse tube coolers used in space, holding detectors at 58 K with a stated lifetime of over 17 years (NASA JPL, AIRS Thermal System). Webb’s MIRI pushed the idea to three pulse tube stages, precooling helium to about 14 K before a Joule-Thomson stage takes detectors to around 6.2 K (NASA, Webb Cryocooler page 2024).
On the ground, the same coolers changed MRI economics. Bath-cooled magnets hold 1,500 to 2,000 liters of liquid helium, and even with a recondensing cryocooler they still need 500 liters or more (Manso Jimeno, Vaughan and Geethanath, MRI magnet review 2022). Sealed, conduction-cooled designs cut that to single-digit liters, though a 2025 review notes the inventory figures are largely manufacturer-reported (Alenezi et al., Cureus 2025).
| Metric | Value | Source |
|---|---|---|
| AIRS pulse tube cryocooler: net cooling capacity, and input power per watt of cooling at 55 K | 1.5 W; about 62 W/W | NASA JPL, AIRS Thermal System page |
| AIRS cooler system mass (both redundant assemblies plus structure), and stated lifetime | 37 kg; over 17 years in space | NASA JPL, AIRS Thermal System page |
| Webb MIRI detector temperature: required maximum and nominal | Below 7 K (no more than 6.7 K); about 6.2 K | NASA, Webb Cryocooler page 2024 |
| Webb MIRI precooler: pulse tube stages vs heritage systems, and cooling power | 3 vs 2 stages; about 1/4 W | NASA, Webb Cryocooler page 2024 |
| MIRI cryocooler electronics: design life, mass, and continuous power delivered | Over 10 years; 9 kg; 360 W | Harvey et al., NASA NTRS 2013 |
| Pulse tube cryocoolers built by Northrop Grumman: combined years on orbit (vendor’s own claim) | More than 300 years | Northrop Grumman, Cryocoolers page |
| Liquid helium in a conventional bath-cooled MRI magnet vs a zero boil-off magnet with recondensing cryocooler | 1,500 to 2,000 L vs 500 L or more | Manso Jimeno et al., MRI magnet review 2022 |
| Average MRI cold head lifespan: new vs refurbished system | 4 to 5 years vs 3 to 4 years | Manso Jimeno et al., MRI magnet review 2022 |
Context note: the same MRI review reports sealed NbTi systems at 7 L (1.5 T) and 0.7 L (0.55 T) of liquid helium. Northrop Grumman also states more than 90% reliability at 87,600 hours (10 years) for its space coolers (Northrop Grumman, Cryocoolers); these are vendor claims, not independent audits. MIRI electronics details: NASA NTRS 2013.
Summary: Thermoacoustic Engine and Refrigerator Efficiency by the Numbers
| Metric | Value | Source |
|---|---|---|
| Record thermoacoustic engine efficiency relative to Carnot | 49% | Tijani and Spoelstra, Journal of Applied Physics 2011 |
| Los Alamos thermoacoustic-Stirling engine: thermal efficiency and share of Carnot | 30%; 41% | Backhaus and Swift, JASA 2000 |
| Internal combustion engines vs piston Stirling engines, efficiency ranges cited in 1999 | 0.25 to 0.40 vs 0.20 to 0.38 | Backhaus and Swift, Nature 1999 |
| Cascade engine acoustic power at 20% efficiency | Up to 2 kW | Gardner and Swift, JASA 2003 |
| Four-stage multi-looped engine: predicted acoustic power at design conditions | 60 kW | Yu et al., Energy 2025 |
| 2004 thermoacoustic generator vs radioisotope thermoelectric generators | 18% vs 7% of heat to electricity | ScienceDaily / Los Alamos National Laboratory, 2004 |
| Highest efficiency of a 1 kW-class traveling-wave generator | 19.8% at 970 W | Wu et al., Applied Energy 2014 |
| Three-stage generator maximum electric power | 4.69 kW at 15.6% | Bi et al., Applied Energy 2017 |
| Hybrid generator maximum output and peak efficiency | 15.0 kW; 39.2% at 11.1 kW | Chen et al., Cell Reports Physical Science 2024 |
| Liquid-metal MHD thermoacoustic generator thermal efficiency | 11.0% at 13.5 W | Chen et al., The Innovation Energy 2026 |
| Lowest reported onset temperature difference | 17 °C | Jin et al., Scientific Reports 2017 |
| Typical heating needed by Stirling-type thermoacoustic engines | Above 600 °C | Jin et al., Scientific Reports 2017 |
| Early standing-wave chillers, efficiency relative to Carnot | 8.5% (SETAC) to 19% (Penn State) | Kajurek et al., SN Applied Sciences 2019 |
| Coaxial thermoacoustic-Stirling cooler | 25% of Carnot, -54 °C | Tijani and Spoelstra, Cryogenics 2008 |
| Heat-driven thermoacoustic refrigerator COPs | 1.12 at 450 °C; 1.34 at 550 °C | Xiao et al., Cell Reports Physical Science 2024; The Innovation Energy 2024 |
| Pulse tube best efficiency at 80 K | 24% of Carnot | Radebaugh, NIST 2003 |
| Commercial 4 K pulse tube input for 0.5 W at 4.2 K | About 5.5 kW | Radebaugh, NIST 2003 |
| Heat-driven pulse tube record low temperature at 100 Hz | 46.6 K | Zhang et al., Energy 2025 |
| AIRS pulse tube cooler stated lifetime | Over 17 years in space | NASA JPL, AIRS Thermal System page |
| MRI liquid helium: conventional vs sealed 1.5 T design | 1,500 to 2,000 L vs 7 L | Manso Jimeno et al., MRI magnet review 2022 |
Methodology and Sources
- Backhaus, S. and Swift, G. W., “A thermoacoustic Stirling heat engine”, Nature 399, 1999: abstract record; press coverage at ScienceDaily 1999.
- Backhaus, S. and Swift, G. W., “A thermoacoustic-Stirling heat engine: Detailed study”, JASA 2000: doi.org/10.1121/1.429343.
- Gardner, D. L. and Swift, G. W., “A cascade thermoacoustic engine”, JASA 2003: doi.org/10.1121/1.1612483.
- Tijani, M. E. H. and Spoelstra, S., “A high performance thermoacoustic engine”, Journal of Applied Physics 2011: doi.org/10.1063/1.3658872.
- Tijani, M. E. H. and Spoelstra, S., “Study of a coaxial thermoacoustic-Stirling cooler”, Cryogenics 2008: IIR FRIDOC record.
- Backhaus, Tward and Petach, traveling-wave thermoacoustic electric generator, Applied Physics Letters 2004, as reported by ScienceDaily 2004.
- Yu, Jaworski and Backhaus, Proc. IMechE Part A 2010: White Rose repository; and Applied Energy 2012: doi.org/10.1016/j.apenergy.2012.04.046.
- Wu et al., Applied Energy 2014: IDEAS/RePEc. Wang et al., Energy 2016: IDEAS/RePEc. Bi et al., Applied Energy 2017: AGRIS record.
- Chen et al., Cell Reports Physical Science 2024: doi.org/10.1016/j.xcrp.2024.101835. Chen et al., The Innovation Energy 2026: journal page.
- Yu et al., Energy 2025: IDEAS/RePEc. Zhang et al., Energy 2025: IDEAS/RePEc. Umar et al., Sustainable Energy Technologies and Assessments 2025: BIT research portal.
- Jin et al., Scientific Reports 2017: PMC5550454.
- Kajurek, Rusowicz and Grzebielec, SN Applied Sciences 2019 (review of refrigerator benchmarks): Springer.
- Xiao et al., Cell Reports Physical Science 2024: doi.org/10.1016/j.xcrp.2024.101815; Xiao et al., The Innovation Energy 2024: KIT repository record; press coverage at TechXplore 2024.
- Radebaugh, R., “Pulse Tube Cryocoolers”, NIST 2003: trc.nist.gov PDF. NIST, “120Hz Pulse Tube Cryocooler for Fast Cooldown to 50K”, 2007: nist.gov.
- NASA JPL, AIRS Thermal System; NASA, Webb Cryocooler; Harvey et al., NASA NTRS 2013; Northrop Grumman, Cryocoolers (vendor claims).
- Manso Jimeno, Vaughan and Geethanath, “Superconducting magnet designs and MRI accessibility: a review”, 2022: arXiv:2205.08918. Alenezi et al., Cureus 2025: PMC12701844.
- Data watch: efficiency definitions differ and are not interchangeable. Heat-to-acoustic efficiency (section 1) excludes alternator losses; thermal-to-electric efficiency (section 2) includes them; “recovery efficiency” in heat-driven coolers (section 5) is defined by each author. The 29.22% thermal-to-electrical efficiency in section 3 comes from a combined power and cooling system and its abstract does not state the exact basis, so it should not be ranked against the section 2 records. Two COP records from the same Chinese Academy of Sciences group appeared in 2024 (1.12 and 1.34) at different heating temperatures (450 °C and 550 °C); both are self-reported comparisons with prior work. Many headline engine records are old: the 49% of Carnot engine (2011), the 17 °C onset (2017) and the pulse tube 80 K efficiency (2003) remain the most recent available data we could verify. A 2026 Cryogenics paper reports 77 K cooling with 45.8% recovery efficiency from a three-stage cascaded cooler, but its full abstract was not accessible, so it is not included. MRI helium inventories for sealed magnets are largely manufacturer-reported, as the 2025 Cureus review itself notes.
Last updated: October 25, 2026. We update this roundup quarterly, and the next refresh is expected when the 2026 International Cryocooler Conference proceedings and the next Applied Energy and Energy issues on thermoacoustic generators are published.