Microwave-to-optical transduction has received a great deal of interest from the cavity optomechanics community as a landmark application for electro-optomechanical systems. In this Letter, we demonstrate a novel transducer that combines high-frequency mechanical motion and a microwave cavity for the first time. The system consists of a 3D microwave cavity and a gallium arsenide optomechanical crystal, which has been placed in the microwave electric field maximum. This allows the microwave cavity to actuate the gigahertz-frequency mechanical breathing mode in the optomechanical crystal through the piezoelectric effect, which is then read out using a telecom optical mode. The gallium arsenide optomechanical crystal is a good candidate for low-noise microwave-to-telecom transduction, as it has been previously cooled to the mechanical ground state in a dilution refrigerator. Moreover, the 3D microwave cavity architecture can naturally be extended to couple to superconducting qubits and to create hybrid quantum systems.
Piezoelectric optomechanical platforms have provided a reliable route towards efficient microwave-to-optical signal transduction. In these approaches, the wavelength mismatch between the microwave (≈ cm) and the optical
(≈μm)signals is circumvented by converting the microwave signal to an acoustic wave (wavelength ≈ few
μmat GHz frequencies), using a piezoelectric resonator, thus providing a large overlap (coupling) between the interacting fields [1], [2]. The transduction efficiency of these devices can be significantly enhanced by hybridising the modes of the piezoelectric resonator and the optomechanical cavity to form a mechanical supermode [2], [3]. However, such hybridisation requires both the frequency and impedance matching between the piezoelectric resonator and the optomechanical cavity, which is restricted by the fabrication lithography. In this study, we propose to achieve the in situ impedance and frequency matching using an LC network containing a MEMS varactor. A piezoelectric resonator can be represented using Butterworth-Van Dyke (BVD) model consisting of four element RLC network, which can be tuned when connected to a varactor in the network, and a MEMS capacitor can be integrated within the same device and is highly power efficient.