Detectivity and bandwidth limits of cooled and uncooled light detection using nanomechanical resonators
Nanomechanical resonators (NMRs) offer a promising alternative to traditional thermal-based radiation detectors due to their immunity to electrical noise. In recent years, these sensors have reached the previously unattained theoretical detectivity limit set by the fluctuation noise of thermal photons at room temperature. Beyond this point, improvements of NMR resonators do not translate into greater detectivity, but in greater effective bandwidth. There is, however, no simple model predicting the limits of this bandwidth enhancement. Likewise, models predicting the performances of NMR-based radiation sensors under active cooling have not been derived. To address these gaps in knowledge, a key missing ingredient consists of defining the NMR optimal driven amplitude that minimizes additive frequency noise, but without performance degradation from nonlinear phenomena. We find that, in the context of NMR-based radiation sensing, this optimal amplitude ($a_\mathrm{opt}$) is dramatically different than the commonly assumed critical amplitude ($a_\mathrm{c}$) that defines the onset of non-linear phenomena in nanomechanical resonators. Our proposed model for this optimal amplitude allows us to quantify the maximum bandwidth enhancement in NMR-based radiation sensors. We also derive simple equations predicting the maximum detectivity in cryogenically cooled sensors. Finally, combining these two models allows us to define new universal performance limits. This unveils important general conclusions on the ideal geometry of NMR radiation sensors. We find that thermomechanically-limited sensors should be as thin and extended as possible. In contrast, readout-limited sensors should also be thin, but should be just large enough to make radiation heat transfer dominant compared to conduction.