Changing the Pairing: A New Quantum Heat Engine Across the BCS–BEC Crossover

A new quantum-engine proposal replaces a familiar question—how far should we compress a gas?—with a stranger one: how should we change the way its particles pair up? A September 29, 2026 preprint, Quantum Statistical Thermal Engine at the BCS–BEC crossover, investigates a strongly interacting ultracold gas whose thermodynamic behaviour can be reshaped by tuning interactions between its atoms.^[1]

The headline result needs careful unpacking. The authors report a theoretical cycle that can reach the Carnot efficiency while retaining nonzero work per cycle, and a separate, fixed-density operating mode with calculated efficiency up to 38%.^[1][2] Neither result demonstrates a practical power generator, finite power at Carnot efficiency, or a violation of thermodynamics. What the work offers is a new way to design the working medium itself.

The material becomes a control knob

In a conventional engine picture, the working substance is something we squeeze, heat, expand and cool. This proposal adds another adjustable property: the interactions that determine whether the relevant particles behave as loosely paired fermions or tightly bound, molecule-like bosons.^[2] The distinction changes the gas’s internal energy and entropy, giving the researchers another route through its thermodynamic landscape.

The study is by Santiago Henríquez Lira, Felipe Isaule, Martín HvE Groves, Francisco J. Peña, Patricio Vargas and Thomás Fogarty, with affiliations in Chile and at the Okinawa Institute of Science and Technology in Japan.^[2] Their central tool is a calculation of the gas’s equation of state: a map connecting temperature, density, interaction strength and thermodynamic properties.

The opportunity is not to escape the energy ledger. It is to gain another controlled way of moving energy through it.

That is why this paper belongs in the quantum-energy conversation. It asks what becomes possible when the internal organisation of a working medium is deliberately varied during operation, rather than treated as fixed background.

BCS and BEC, without the alphabet soup

Fermions and bosons obey different rules for occupying quantum states. In the ultracold-gas setting, two fermionic atoms can form a composite pair. On the BEC side of the crossover, the pairs are tightly bound molecules with bosonic behaviour; on the BCS side, pairing is looser and the gas has predominantly fermionic thermodynamic character.^[2][3]

The atoms do not literally change their fundamental identity. “Bosonization” and “fermionization” are operational labels for moving between regimes with different effective many-body behaviour. The new paper explicitly uses the terms in this sense.^[2] That qualification matters: this is controlled pairing physics, not an alchemical conversion of one elementary particle into another.

Experiments can navigate this crossover by using a magnetic field to tune interactions near a Feshbach resonance. The 2023 demonstration used a trapped lithium-6 gas and moved between a molecular Bose–Einstein condensate and a unitary Fermi gas.^[3] The new proposal builds on this experimentally established control capability, rather than requiring an entirely unknown material transformation.

A six-stroke cycle, with two unusual steps

The proposed Statistical Thermal Cycle combines two density-changing strokes, two fixed-density thermalisation strokes and two interaction-changing strokes at constant temperature.^[2] In plain language, its sequence is:

1. Compress without exchanging heat. Increase the density while preserving entropy and holding the specified dimensionless interaction parameter fixed.
2. Contact the hot reservoir. Bring the gas to the hot-bath temperature at fixed density and interaction parameter.
3. Change the pairing regime while hot. Tune the interaction at constant temperature and density. Both heat and work can be exchanged.
4. Expand without exchanging heat. Return to the lower density along an entropy-preserving path.
5. Contact the cold reservoir. Bring the gas to the cold-bath temperature at fixed density and interaction parameter.
6. Restore the original pairing regime while cold. Reverse the interaction transformation at constant temperature and density, closing the cycle.^[2]

The unfamiliar steps are the hot and cold interaction sweeps. Ordinary expansion is not necessary during these strokes: the density stays fixed, but the energy landscape and entropy change as interactions are adjusted. The authors separate the energy change into heat exchanged with the reservoir and work associated with the interaction control.^[2]

Think of a machine with an additional adjustable internal mechanism, not a piston that somehow produces energy without moving anything. The control field and reservoir contacts remain part of the physical operation.

The 38% result: work without compression

The authors first remove the density change altogether. When the compression ratio is one, the compression and expansion steps disappear, leaving two thermalisation strokes and two isothermal interaction sweeps. They describe this as a statistical Stirling-like cycle: interactions replace the volume changes of the familiar isothermal strokes.^[2]

In that calculation, the strongest performance comes from choosing interaction endpoints that make effective use of the contrasting entropy structures on the two sides of the crossover. The reported maximum occurs at a cold-to-hot temperature ratio of 0.35 and selected interaction endpoints near the relevant superfluid-transition conditions.^[2]

The authors also find an important limit to the strategy: heating more is not indefinitely better. At sufficiently high hot-branch temperatures, pairing correlations weaken and the statistical advantage diminishes.^[2] The resource is not simply “more heat.” It is the combination of a temperature difference with a working medium whose correlations respond usefully to the chosen control path.

Nor is the best-efficiency point automatically the best-work point. The study identifies a trade-off: settings that maximise efficiency need not maximise the useful work delivered in each cycle.^[2] That distinction should stay visible whenever a single percentage becomes the headline.

Reaching Carnot does not mean beating Carnot

When compression and expansion are restored, the proposal has another notable operating point. The authors find a condition where the gas reaches the hot-bath temperature at the end of compression and the cold-bath temperature at the end of expansion. The two intervening thermalisation strokes then have zero heat exchange.^[2]

What remains is a four-stroke cycle with two entropy-preserving density changes and two isothermal interaction changes. In the model, the Otto benchmark and the Carnot efficiency coincide at this point. Crucially, the interaction sweeps still support a finite entropy change, so the calculated work per cycle does not collapse to zero.^[2]

The comparison is specifically with a conventional Otto cycle approaching its Carnot boundary, where its work output can vanish. It should not be read as saying that every ordinary reversible heat-engine cycle must have zero work per cycle.^[2] The new design preserves a useful interaction-driven route through state space; it does not remove the need to analyse cycle duration and irreversible losses.

For a research hub interested in beyond-Carnot thermodynamics, the distinction is essential. This paper treats Carnot as the upper bound for its heat-engine operation and reports fixed-density efficiencies below it.^[2] Its advance is a proposed route to the limit under stated assumptions, not evidence that the second law has failed.

What the earlier experiment actually established

There is a real experimental foundation, but it must not be confused with verification of the new proposal. In Nature in 2023, Jennifer Koch and colleagues reported a “Pauli engine” using an ultracold lithium gas. They demonstrated work production through a cycle that exploited the energy difference between bosonic and fermionic ensembles, reporting efficiency up to 25% under their engine’s definition.^[3]

That experiment replaced the usual heat strokes with changes in quantum statistics. The 2026 proposal instead explicitly includes hot and cold reservoir contacts and isothermal interaction sweeps.^[2][3] Their efficiencies therefore cannot be treated as a simple league table showing a measured improvement from 25% to 38%.

The defensible connection is narrower and more useful: interaction control across the crossover has already been used in a functioning microscopic engine. Whether the new thermal cycle can reproduce its calculated performance remains an experimental question.

Four operating modes, not just an engine

The new study maps regimes where the same working medium acts as an engine, refrigerator, heat accelerator or heater.^[2] Those names describe the directions of energy flow, not different substances:

• Engine: draw net heat from the hot reservoir, reject heat to the cold reservoir and deliver work.
• Refrigerator: consume work to move heat from the cold reservoir toward the hot one.
• Heat accelerator: consume work to increase heat transfer in its natural hot-to-cold direction.
• Heater: consume work and release heat to both reservoirs.^[2]

There is a practical qualification: in the parameter study, interaction tuning alone does not open every mode. Reaching refrigeration also requires a density-dependent change in the Fermi energy.^[2] “Programmable thermal machine” is therefore a reasonable interpretation, but “one interaction knob does everything” would overstate the result.

Where Floquet engineering could connect

This is a quantum-thermodynamics proposal, not a reported Floquet-material experiment. Repeating a thermodynamic cycle does not, by itself, establish that its performance comes from Floquet-engineered quasienergies or sidebands. The paper’s analysis centres on the equilibrium thermodynamics of a tunable interacting gas.^[2]

The connection to Floquet research is a prospective design question: could a periodic control protocol help implement interaction sweeps or manage unwanted excitations during a future version of such a cycle? That is an avenue to investigate, not a result established here. Any proposed benefit would have to be assessed alongside drive energy, additional heating and control losses.

This separation keeps the research agenda honest. The present paper supplies a thermodynamic target and a set of control variables. A future driven-system study would need to show what periodic engineering adds, rather than attaching the Floquet label to an already cyclic machine.

What would make the next result convincing?

The equation of state is obtained using a functional renormalization group calculation, a non-perturbative method with an explicitly truncated modelling ansatz.^[2] That makes it a substantial theoretical study, not an exact, assumption-free description of every experimental gas. Independent modelling and measured state variables would strengthen the performance claims.

For experimental follow-up, our assessment is that four tests would be especially valuable:

• Close the complete energy balance. Measure reservoir heat and interaction-control work across the full cycle, including the paper’s distinction between heat and internal correlation-related contributions.
• Measure duration as well as work. Establish power and losses at realistic sweep and thermalisation times.
• Test repeatability. Show that repeated cycles return the working medium to the intended state rather than consuming an untracked initial resource.
• Separate device and apparatus efficiency. Distinguish the microscopic engine’s performance from the energy cost of cooling, trapping and controlling the atoms.

The bottom line is a design insight, not a power-plant claim: changing how quantum particles pair can supply an additional work-producing stroke. The new preprint makes that idea quantitative across a strongly interacting crossover, while leaving finite-time implementation and full experimental verification open.^[1][2] For quantum energy research, that is a meaningful next target—without needing to promise energy from nowhere.