Colloidal heat engine driven by engineered active noise

A recent experiment demonstrated that a colloidal heat engine immersed in an active bath can outperform those subject to equilibrium or passive baths. Although it has motivated several theoretical studies to clarify the possible origins of such high efficiency, the discussed noise effects are often intertwined and hard to distinguish in nature. In this paper, we used the optical feedback trap to mimic a colloidal heat engine in various active baths in a precisely controllable setting to analyze the impacts of individual noise properties in real systems. Our experiment, together with theoretical calculations, show that the efficiency of the active engine can be greatly increased when the magnitude of the active-bath noise with a nonzero correlation time is modulated according to the variation of the confining potential in the quasistatic limit. This result reveals how to tune the noise properties of an active bath to enhance the efficiency of a microscopic machine, which could be used to manipulate the functions of stochastic systems under complex active backgrounds.