Feedback Cooling of a Room Temperature Mechanical Oscillator close to its Motional Ground State

Preparing mechanical systems in their lowest possible entropy state, the quantum ground state, starting from a room temperature environment is a key challenge in quantum optomechanics. This would not only enable creating quantum states of truly macroscopic systems, but at the same time also lay the groundwork for a new generation of quantum-limited mechanical sensors in ambient environments. Laser cooling of optomechanical devices using the radiation pressure force combined with cryogenic precooling has been successful at demonstrating ground state preparation of various devices, while a similar demonstration starting from a room temperature environment remains an outstanding goal. Here, we combine integrated nanophotonics with phononic band gap engineering to simultaneously overcome prior limitations in the isolation from the surrounding environment and the achievable mechanical frequencies, as well as limited optomechanical coupling strength, demonstrating a single-photon cooperativity of 200. This new microchip technology allows us to feedback cool a mechanical resonator to around 1 mK, near its motional ground state, from room temperature. Our experiment marks a major step toward accessible, widespread quantum technologies with mechanical resonators.

Figure 1

(a) Shown is a stitched microscope image of the fabricated structure (top) and the corresponding mechanical simulation of the long beam (bottom). The enlargement shows a fishbone structure designed to reduce bending losses. (b) Mechanical (top) and photonic (bottom) simulation of the center part containing the photonic crystal. A short second beam forming the other half of the photonic crystal cavity is fixed close to the mechanical beam. The short structure does not feature any mechanical motion around the defect mode. The two structures form an optical cavity with the light strongly confined in the gap between the two beams. The mechanical motion changes the gap size, shifting the optical resonance frequency and hence giving rise to the optomechanical interaction. (c) Sketch of the central photonic crystal and coupling waveguide. Clearly visible is the short second beam in the center. A cross section is shown in the box on the bottom. (d) Band diagram for the in-plane mode of the phononic fishbone structure. We apply a periodic boundary condition in the $x$ direction, with $k_x$ being the wave vector. The blocks in the long beam form a band gap between 0.6 and 1.1 MHz. (e) Bending (${\partial }^{2}v/\partial x^2$) normalized to the displacement $v$ within the photonic fishbone structure. In the center, the magnitude of the bending alternates between the thin (large bending, white) and thick (small bending, gray) parts of the structure. As the mechanical losses are proportional to the cube of the width, the fishbone devices exhibit significantly higher mechanical $Q_m$ than a uniform beam of equal width (orange), typically used for photonic crystal zipper cavities [25].

Figure 2

(a) Mechanical spectrum. The red dashed line marks the high-$Q_m$ defect mode. Gray lines indicate additional mechanical modes that strongly couple to the optical cavity, while most other spurious peaks arise from the mixing of these modes in the detection itself. (b) Phase response of the feedback control. The circuit has a phase of $-0.5\pi$ at the resonance of the defect mode, with the gray area indicating the unstable region due to heating. (c) Gain of the feedback control. We implement several filter functions with large bandwidth, allowing us to suppress and partly cool other modes that are excited in order to stabilize the system.

Figure 3

(a) Sketch of the feedback cooling setup. A laser is first tuned on cavity resonance and phase modulated to generate a calibration tone. It is then split into two arms, with both intensities being controlled through variable optical attenuators (VOAs). The bottom path is the local oscillator for the homodyne detection scheme, where the phase can be controlled using a fiber stretcher ($\varphi$). The light in the upper (signal) arm is intensity modulated in an electro-optical modulator (EOM) and sent to the waveguide [34], where it is then evanescently coupled into the optical cavity. At the end of the waveguide, we pattern a photonic crystal mirror, which allows the light from the cavity to be reflected back into the fiber with a collection efficiency of 91%. This light is then mixed with the local oscillator on a beam splitter and measured in a home-built low-noise balanced photodetector with a quantum efficiency of 70% in order to perform the phase-sensitive measurement. The detected signal is electronically processed in a FPGA-based controller, which directly modulates the light in real time through the EOM, and hence allows us to cool the mechanical resonator. (b) Cooled mechanical spectra $S_{yy}$, with increased feedback gain from orange to blue and constant intracavity photon number $n_c$. The dashed lines are fits to the spectra, while the gray dotted line indicates the quantum-limited noise floor. (c) Average phonon number extracted from the spectra in (b), with corresponding color coding. The gray dashed line represents the theoretically predicted quantum-noise-limited phonon number, and the dark blue dotted line is the expected phonon number when taking the noise into account.