Inside an ultracold chamber, two atomic clouds collide: a few dozen cesium atoms acting as the working fluid, and thousands of rubidium atoms serving as the heat reservoir. Each collision is a microscopic heat exchange, akin to tuning a musical instrument: the reservoir's temperature sets the tempo at which atoms exchange energy. This tempo, not just the magnetic fields, determines how quickly the quantum engine completes its cycle.

In the experiment, heat exchange occurs through inelastic collisions that shift cesium atoms among seven energy levels defined by the magnetic field. During heating, atoms gain energy; during cooling, they release it. The speed of these processes depends on the velocity of the rubidium atoms, i.e., their kinetic temperature. By raising the reservoir temperature from 679 to 1140 nanokelvins, the physicists changed the law of heat exchange: heating slowed down, while cooling sped up. This allowed them to shorten the total cycle time without altering the magnetic fields.

The key result is an optimum at 1410 nanokelvins, where power peaked at 28.0 nK/ms (in units of the Boltzmann constant), while efficiency remained at 0.489. Previously, it was thought that speeding up required changing the system itself, but here it sufficed to "tune" the reservoir. Simulations based on realistic scattering cross-sections reproduced the experiment exactly and showed that simplified models with a single relaxation rate overestimate power by a factor of 1.5 to 2.

This discovery links atomic collision physics with finite-time thermodynamics. Controlling the coupling to the reservoir proved more effective than controlling the system itself, and it paves the way for designing nanoscale heat engines where the spectrum of relaxation times becomes a resource. The tuning continues: ahead lie reservoirs with memory and optimized control protocols.