Thermodynamic Efficiency Limits of Nanoscale Heat Engines

Abstract

Heat engines built from single nanoscale working substances, such as a single trapped ion, a quantum dot, or a
colloidal particle in a time-dependent optical trap, operate close enough to the thermal energy scale k_B T that
power output fluctuates strongly from cycle to cycle, and it remains unclear how tightly the classical trade-off
between efficiency and power survives once these fluctuations are accounted for explicitly rather than averaged
away. We simulated a stochastic nanoscale Otto-cycle heat engine based on a harmonically confined overdamped
colloidal particle exchanging heat with reservoirs at 320 K and 300 K, sweeping cycle time across 40 values from 1
microsecond to 50 milliseconds and trap stiffness ratio across 12 values, and computed the full cycle-to-cycle
distribution of work, heat, and efficiency from 60,000 simulated stochastic trajectories per operating point. Mean
efficiency at maximum power was 0.121 +/- 0.004, within 3% of the Curzon-Ahlborn prediction of 0.125 for the same
reservoir temperatures, but the cycle-to-cycle standard deviation of efficiency at that operating point was 0.34,
nearly three times the mean value itself, and 9.7% of individual cycles yielded negative net work output despite
positive mean power. The coefficient of variation of power output decreased with cycle time following a power law
with exponent -0.48 +/- 0.02 (R^2 = 0.981), indicating that fluctuations remain substantial even at the longest cycle
times tested and decay markedly more slowly than the inverse-square-root scaling naively expected from
central-limit averaging over independent thermal kicks. A thermodynamic uncertainty relation bound relating power
fluctuations to entropy production held across all 480 operating points tested, with the measured product of relative
power variance and total entropy production exceeding the theoretical bound of 2 by a median margin of only 14%,
confirming the bound operates close to saturation in this regime. Trap stiffness ratio affected mean efficiency at
maximum power by less than 2% across the tested range but changed the efficiency fluctuation magnitude by up to
40%, identifying stiffness ratio as a design parameter that primarily controls reliability rather than mean
performance. These results show that classical efficiency-power trade-off relations remain accurate on average for
nanoscale heat engines but obscure cycle-to-cycle variability large enough to be operationally significant, with
implications for the design of nanoscale energy-harvesting devices where single-cycle reliability, not only average
performance, determines practical utility.