Reaching the quantum ground state of an optically levitated nanomechanical oscillator – QLeviO

Three work streams were pursued in parallel. ICMCB developed an original synthesis route via decomposition of hydrogen silsesquioxane in supercritical fluid to produce spherical particles with a silicon core and a silica shell of controllable thickness. Despite sustained effort, the silicon core remained polycrystalline throughout the project, as confirmed by cross-section TEM, XRD and Raman spectroscopy. LP2N designed and delivered two laser sources unavailable commercially: a 7 W trapping source at 1550 nm (MOPA architecture, Er-Yb co-doped double-clad fibre, RIN ≤ −160 dBc/Hz from 100 Hz to 10 MHz) and a 4 W directly-modulable cooling source at 1560 nm capable of MHz-bandwidth intensity modulation. LOMA converted its optical tweezer from 1064 nm to 1550 nm, reached pressures below 10⁻⁸ mbar with a NEG-ion pump combination, and implemented axial homodyne position detection coupled to a real-time LQG controller on an FPGA for cold damping feedback.

The project produced two major results. First, the complete experimental infrastructure was validated on silica nanoparticles, with demonstration of centre-of-mass cooling and publication of two high-impact articles (Physical Review Letters, Nature Communications) related to optical forces. Second, a fundamental and unexpected discovery: polycrystalline Si@SiO₂ nanoparticles evaporate systematically below ~0.1 mbar. Two independent diagnostics — 90° scattered intensity and translational damping rates — provide convergent evidence of progressive mass loss during pump-down. The root cause is the semiconductor nature of silicon: at the internal temperatures reached in vacuum, thermally excited free carriers generate an additional absorption channel triggering thermal runaway. A self-consistent numerical model coupling the heat equation with semiconductor corrections to the Hertz-Knudsen evaporation law quantitatively reproduces the observations. A key highlight: the evaporation kinetics enable the first measurement of the vaporisation enthalpy on a single isolated nano-object, with no contact, no substrate, and no ensemble average.

The most striking result of QLeviO is the demonstration that semiconductor nanoparticles cannot be trapped in ultra-high vacuum using standard dielectric thermal models — a fundamental warning to the levitated optomechanics community. This discovery simultaneously opens a new experimental field: single-particle semiconductor calorimetry. The immediate next steps are twofold. On the quantum cooling side, the complete infrastructure is in place for ground-state cooling of silica nanoparticles along the axial z-axis, with extension to the transverse x and y axes via acousto-optic deflectors — a more flexible alternative to the original multi-lens assembly. On the semiconductor thermodynamics side, the vaporisation enthalpy measurement will be extended to other semiconductors (Ge, GaAs, InP, GaN), for which the same thermal runaway is predicted numerically. In the longer term, achieving a fully crystalline silicon core would reopen the original QLeviO goal of silicon-based quantum optomechanics, with its intrinsic advantages in scattered photon yield and quantum backaction.

Levitated nanomechanical oscillator presents some advantages with respect to its clamped optomechanical counterparts, as e.g. its ability during free-fall and coherent evolution to expand its wavefunction up to overlap the particle’s size. Wave-matter interference are expected to test the fundations of quantum mechanics through collapse models and to verify the role of gravity as a fundamental limit in the coherence of a macroscopic quantum superposition. With its extreme isolation from environment and the highest recorded mechanical quality factor, levitated particles are also ultra-sensitive force sensors with the ability to test the existence or not a fifth force, arising from beyond standard models.
A prerequisite for carrying out theses studies is to ground state cool the centre-of-mass motion of the optically trapped particle. There are several types of cooling schemes, such as parametric and linear feedback, as well as, cavity-based side-band and coherent scattering cooling. The latter enabled to achieve ground state cooling for only one vibrational degree of freedom.
LOMA is a pioneer in France in the study of optical vacuum trapping of dielectric nanoparticles. To go further in this field, it is desirable to cool the particle in the three directions of space in a completely optical manner. Our all-optical cooling protocol will enable to directly measure in real-time weak vectorial forces.
The QleviO project proposes to ground state cool all vibrational degrees of freedom of a nanoparticle by radiation pressure without requiring an optical resonator or a cryostat. Up to now, optical trapping in high vacuum restricts nanoparticle materials to silica because of its efficient heat dissipation by emitted black body photons. To extend to other materials, the quality of the material and the laser sources are of the utmost importance. To control the particle motional state with light, the refractive index should be high and the light absorption should be weak, while mitigating the heat excess in high vacuum.
It turns out that the only laboratory that best meets all of these conditions is the ICMCB that produces core@shell Si@SiO2 particles. The core has a high index and the silica shell efficiently dissipate the heat excess due to its higher emissivity. This now requires to trap and cool such particles to the most favorable wavelength, 1550 nm. Futhermore, the intensity noise of the laser source has to be specifically minimized at the oscillator frequencies. The experts in high-power ultra-low intensity noise laser sources are in LP2N and will develop such sources : one for trapping (and detecting the particle’s motion) and one source dedicated to cool the center-of-mass motion. The interactions between partners will be strong and the consortium strategy will be to work in parallel in the material side, the optomechanical setup and the laser sources.
Remarkably, almost all position information is carried by the backscattered photons, facilitating a homodyne measurement operating close to the Heisenberg limit. Thus, to reach the ground state in the three dimensional space of an optically tweezed particle, we will use measurement-based quantum feedback. This is based on the fact that we can resolve the zero point motion in the decoherence scale time and implement a measurement-based feedback control to cool the oscillator towards its ground state. In other words, the real-time feedback anticipates and cancels the disturbance due to backaction-induced motion.