Observing and evading quantum back-action on a kilogram-scale oscillator

AI summary

Overview

• Research area: Quantum optomechanics and gravitational-wave detector instrumentation — specifically quantum measurement noise (imprecision and radiation-pressure back-action) in the LIGO interferometer, and its manipulation with squeezed light.
• Technical level: Advanced. The paper assumes familiarity with quantum measurement theory, input-output formalism for optical cavities, squeezed states of light, and feedback control of interferometers.
• Scope (one sentence): The paper reports the direct observation and subsequent cancellation (evasion) of quantum radiation-pressure back-action on a feedback-trapped, kilogram-scale mechanical mode in the LIGO Livingston detector, using frequency-dependent squeezing.

What This Paper Is About

Any continuous quantum measurement of a mirror's position must pay for its precision with a random force — radiation-pressure back-action — that shakes the mirror. In LIGO, the 40 kg mirrors are suspended at about 0.5 Hz, far below the band where quantum noise dominates, so the back-action had previously been seen only as a broadband rise in noise, never separately resolved. The authors reshape the detector's control loop to create an artificial mechanical resonance at 151 Hz where imprecision and back-action are comparable, then show that frequency-dependent squeezing produces correlations between the two that partially cancel the back-action, reducing the observed motion by (47 ± 5)% relative to vacuum and (31 ± 7)% relative to frequency-independent squeezing.

Key Contributions

1. A feedback-trapped kilogram-scale optomechanical mode. By reshaping the differential-arm control loop, the authors confine the differential motion of the four 40 kg mirrors into a sub-Hz-linewidth mode at 151 Hz with an effective cold-damped quality factor of ∼100 and an effective mass of ≈10 kg, placing the mode where radiation-pressure back-action dominates the motion.
2. Separate observation of imprecision and back-action. Comparing quantum vacuum, frequency-independent squeezing (FIS), and frequency-dependent squeezing (FDS) injected at the antisymmetric port, they show that FIS lowers the high-frequency imprecision floor while raising low-frequency motion (enhanced back-action), whereas FDS preserves the imprecision benefit and suppresses motion near and below the trapped-mode resonance.
3. A general decomposition of quantum noise into imprecision, back-action, and correlation. Using the known power scaling of the two noise terms, the displacement-referred quantum noise is written as S_xx^QN(Ω, P) = A(Ω)/P + B(Ω)·P + C(Ω), where P is optical power and the three terms correspond to imprecision, back-action, and correlation. The authors state that this form relies only on the power scaling of the two noises and holds for any linear multiport interferometer.
4. A back-action-evasion metric and a reservoir interpretation. They define a frequency-resolved evasion metric Ξ_BAE(Ω) = −S_xx^{imp,ba}(Ω) / [S_xx^imp(Ω) + S_xx^ba(Ω)], which estimates the fraction of positive quantum noise cancelled by correlations at each frequency, and they describe the squeezed field as an engineered reservoir characterized by an effective frequency-dependent occupation number in a modified fluctuation-dissipation theorem.

Main Findings

• Back-action is directly observed and is quantum in origin. Around the trapped resonance, injecting FDS reduces the motion below the level measured with ordinary vacuum. The authors state this is consistent with the expectation that, even in the vacuum case, a fraction of the trapped oscillator's motion is driven by back-action.
• Quantified evasion: correlations from FDS reduce the total displacement-noise power near resonance by (47 ± 5)% relative to the vacuum case and (31 ± 7)% relative to the FIS case.
• FIS and FDS agree at high frequency. The close agreement of the FIS and FDS spectra at high frequency is presented as evidence that the resonance-region suppression in the FDS case does not come from reduced classical noise or a drift in injected squeezing, but from imprecision–back-action correlations.
• The correlation term is negative for FDS below about 170 Hz. The decomposition shows significant back-action for FIS below 170 Hz, substantial suppression in the same band for FDS, and a negative correlation term for FDS in that band (Fig. 3B).
• FIS does not evade back-action. Near the trapped resonance, FIS produces a small positive correlation that adds to the observed noise rather than cancelling it; the authors note that in the FIS case correlations augment the motion in addition to the increased radiation-pressure back-action.
• Unsqueezed vacuum shows a small effect attributed to ponderomotive correlations, visible in the Ξ_BAE comparison among unsqueezed, FIS, and FDS configurations.
• Residual classical noise is consistent across configurations. After normalizing by the suppression factor SF = χ_eff/χ_0, the inferred classical residuals agree across the three input states in magnitude and spectral shape, and are consistent with independent estimates that coating thermal noise dominates this band.
• Detector parameters used: 40 kg mirrors, 4 km arm cavities, 280 kW circulating power, displacement noise S_xx(2π·150 Hz) ≈ (2 × 10⁻²⁰ m/√Hz)², effective mass ≈10 kg, natural suspension frequency ≈0.5 Hz, mechanical quality factor Q ≳ 10⁸.
• Reservoir framework: a modified fluctuation-dissipation theorem, S_FF(Ω) = 2ℏ(n_eff(Ω) + 1/2) Im χ⁻¹(Ω), with n_eff(Ω) = N(Ω) + Re[M(Ω)e^{−2iθ(Ω)}], is used to describe the squeezed thermal reservoir. The figure presenting the inferred effective phonon occupancy is described in the excerpt, but its numerical values are not reported in the provided text.

Methodology in Plain English

The measurement was performed at the LIGO Livingston Observatory. The interferometer senses the differential motion of four 40 kg mirrors by beating the antisymmetric-port output field against a local oscillator set by a small intentional offset of the interferometer.

To make back-action visible, the team did not build a new oscillator — they used the existing differential-arm feedback control loop. By digitally filtering the error signal and feeding it back to the mirror suspensions, they effectively trapped the differential mode at 151 Hz, giving it a narrow linewidth (quality factor ∼100) in a frequency band where radiation pressure, not thermal noise, dominates. An equivalent linear control diagram in the paper shows how total force (thermal, radiation-pressure back-action, technical) and measurement imprecision combine through the suspension response and the feedback to produce the observed displacement.

They then repeated the measurement with three different quantum states injected into the dark port: ordinary vacuum, frequency-independent squeezing, and frequency-dependent squeezing — the last engineered to be amplitude-squeezed at low frequency and phase-squeezed at high frequency. Holding the trap configuration and mean optical power fixed, they compared the spectra.

Finally, rather than relying on a single-mode model, they exploited the fact that imprecision scales as 1/√P and back-action as √P in the optical field amplitude. Sweeping the interpretation over power lets them fit the observed quantum noise to A/P + B·P + C, cleanly separating imprecision, back-action, and their correlation without needing to know the interferometer's internal details. A detailed numerical input-output model of Advanced LIGO supplies the physically interpretable coefficients.

Why This Matters

Impact on research. This is a direct, quantitative demonstration that quantum back-action can be observed and then evaded on a macroscopic object — a kilogram-scale mirror — rather than a micro- or nanomechanical oscillator. It validates in a full, lossy, multiport interferometer the long-standing proposal that injected quantum correlations can cancel radiation-pressure noise, and it supplies a general, model-independent bookkeeping scheme (A/P + B·P + C) plus a metric (Ξ_BAE) for attributing noise to imprecision, back-action, or correlation. It also frames squeezed light as an engineered reservoir that changes only the fluctuations of the oscillator, not its dissipation, which is a distinct statement from the qubit case.

Real-world applications

• Gravitational-wave astronomy: future detector upgrades and designs that need to reduce quantum radiation-pressure noise in the audio band to increase astrophysical reach.
• Quantum-limited precision metrology: interferometers and optomechanical sensors operating near the standard quantum limit, where back-action evasion is the route to sub-SQL sensitivity.
• Squeezed-light technology: validation of frequency-dependent squeezing as an operational tool in a kilometer-scale, high-power interferometer.
• Macroscopic quantum mechanics: a platform for probing fluctuation-dissipation relations and effective occupation numbers of engineered reservoirs with kilogram-scale test masses.

Industry relevance. The results bear on vendors and programs building squeezed-light sources, optical loss and detection hardware for interferometric sensors, and on any industry using laser interferometry at or near quantum limits (semiconductor metrology, inertial sensing, and future gravitational-wave observatories). The correlation-based evasion technique is a signal-processing/optical-design principle that transfers to other linear multiport measurement systems.

Future Directions

• Reduce classical noise around the trapped mode. The residual classical noise (attributed primarily to coating thermal noise) currently limits how cleanly the quantum decomposition can be performed; lower-loss coatings would sharpen the measurement.
• Extend the technique over a broader band. The demonstration is localized to the trapped-mode band (with back-action evident below 170 Hz); whether similar correlation-driven cancellation can be pushed across the full observation band of a detector is an open question.
• Push on optical power and loss. Since the decomposition hinges on the 1/√P and √P scalings, higher circulating power and reduced loss ports would both strengthen the back-action signal and test the multiport framework further.
• Use the trapped oscillator as a reservoir thermometer. The modified fluctuation-dissipation theorem predicts an effective occupation number n_eff(Ω) that can be probed through the oscillator's response; a systematic mapping of n_eff versus squeezing angle and magnitude would test the reservoir picture directly.

Target Audience

Advanced graduate students and researchers in quantum optomechanics, quantum measurement theory, and gravitational-wave instrumentation; LIGO/Virgo/KAGRA collaboration members and detector-design teams; and engineers working on squeezed-light sources or quantum-limited interferometric sensors who want a concrete, full-scale demonstration that back-action evasion works in a lossy multiport system. Beginners will find the core idea accessible from the abstract, but the methods and decomposition formalism require a background in quantum optics.

Authors’ abstract

Continuous quantum displacement measurements are fundamentally limited by a trade-off between readout imprecision and measurement back-action, constrained by the Heisenberg uncertainty principle. In the Laser Interferometric Gravitational-Wave Observatory (LIGO), these two quantum noise components dominate much of the observation band, making it an excellent testbed. We induce a sub-Hz-linewidth optomechanical mode by trapping the differential motion of the 40-kg mirrors in a band where radiation-pressure back-action dominates the motion. Engineering the quantum state entering the dark port creates correlations between imprecision and back-action that partially cancel their contributions, reducing observed motion near resonance by ~47%. A framework resolving the imprecision, back-action, and correlation terms identifies the origin of this suppression. These results demonstrate quantum back-action evasion and quantum reservoir engineering in a macroscopic optomechanical system.