{"id":"92152cb6-71fc-44c9-8ccc-f51cdf8d2bf9","arxiv_id":"2412.08006","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A superconducting qubit strongly coupled to a 25-millisecond-lifetime silicon nanomechanical oscillator generates non-classical Fock states and extends mechanical coherence to about 1 millisecond via dynamical decoupling.","lead":"This paper reports a tiny silicon drum that stores quantum information for about 25 milliseconds, far longer than earlier vibration-based quantum memories. The team coupled it to a superconducting qubit to create and measure single quantum packets of vibration, then used echo techniques to extend the storage time to about one millisecond.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 25 ms intrinsic T1 is an inverse-Purcell-corrected value, not a directly measured single-phonon decay; raw Γ(Δ) and κ data are not released, so the headline number is not independently checkable.","rationale":"Good-faith read: the experimental core—strong coupling with resolved vacuum Rabi oscillations, Wigner-negative Fock states, ~20 ms direct single-phonon decays, and DD-enhanced coherence—is well supported and internally consistent. The multiple lifetime channels (single-phonon, intermediate phonon number, high-phonon waveguide devices) and the TLS spectroscopy provide convergent evidence. The least secure link in the central quantitative claim is exactly the conversion of the measured ~19–21 ms decays into the headline 25 ms intrinsic lifetime via the model subtraction in Appendix H1. This is not an internal inconsistency, but it is a load-bearing model correction whose inputs cannot be verified from the published text because the raw fit data are withheld. A 20–30% correction means that even modest κ errors move T1 by several ms. However, the scientific conclusion—an ultra-long-lived mechanical mode in the quantum regime—does not hinge on 25 versus 20 ms, and the strong-coupling, Wigner-negativity, and dynamical-decoupling results are independent of this subtraction. I therefore agree with the reader's identification of the weakest assumption and recommend keeping the ACCEPT verdict, with a note requesting the raw Γ(Δ) data for independent verification.","tokens_in":33132,"tokens_out":14245,"duration_ms":159174,"concrete_test":"Request the raw Γ(Δ) data behind Fig. S10, including the independently measured κ and g_em at the same flux/voltage point. Re-fit Eq. H1 with Γi and κ both free, and recompute Γi using only the largest-detuning points (Δ ≥ 40 MHz). If the resulting intrinsic T1 differs from 25 ms by more than the quoted ±2 ms, the headline value should be revised to the model-free directly measured value. Cross-check by acquiring a single-phonon decay at Δ ≥ 60 MHz (or with reduced g_em) so the inverse-Purcell correction is below about 4 s^-1 and comparing the asymptotic decay rate.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is an intrinsic single-phonon lifetime T1 = 25 ± 2 ms. This value is not what Fig. 4 directly shows: the measured single-phonon decay times for mechanics A/B are 19.3 ± 1.2 / 21.1 ± 1.1 ms (Table SI), and 25 ms is obtained only after subtracting the inverse-Purcell contribution Γκ = (g_em/Δ)^2 κ (Eq. H1, Appendix H1). At the 30–40 MHz detunings used during the long waits, with g_em/2π ≈ 160 kHz at 40 V and κ ≈ 5.9 × 10^5 s^-1 from T1,q ≈ 1.7 µs, Γκ is roughly 9–12 s^-1 against an intrinsic rate of about 40 s^-1, i.e., a 20–30% correction. The fitted correction therefore materially controls the headline number. If κ is mischaracterized, if the ac-Stark detuning changes κ or the qubit thermal population, or if the (g/Δ)^2 scaling has a prefactor error, the inferred intrinsic rate moves by several s^-1 and T1 shifts by several ms. The paper states that the model fits Fig. S10, but the raw Γ(Δ) points, fit residuals, and the independently measured κ are not shown or released; the data-availability statement is 'upon reasonable request'. This concern does not undermine strong coupling, Wigner negativity, or the directly measured ~20 ms decays, so it is a reproducibility and quantitative-transparency issue rather than a fatal flaw; nevertheless, the exact 25 ms milestone is not independently checkable from the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a quantum electromechanical platform in which a transmon qubit is electrostatically coupled to a single-crystal silicon phononic-crystal nanomechanical resonator. The authors demonstrate strong coupling (gem/2π ≈ 200–230 kHz), ground-state cooling and thermometry (mechanics at 72 ± 9 mK), deterministic preparation of a single-phonon Fock state, and Wigner tomography with negativity W(0) = −0.10 ± 0.02. They measure single-phonon energy relaxation times of 19.3 ± 1.2 ms (mechanics A) and 21.1 ± 1.1 ms (mechanics B), and infer an intrinsic T1 = 25 ± 2 ms after subtracting an inverse-Purcell contribution. Ramsey coherence T2* ≈ 64 µs is extended via Hahn echo to 562 ± 42 µs and via Carr-Purcell-2 to 1.02 ± 0.15 ms. Spectroscopy as a function of bias voltage reveals a sparse bath of TLS defects with avoided crossings; the observed frequency noise, Lorentzian spectra, and echo efficiency are modeled with a phenomenological ensemble of telegrapher fluctuators. The paper argues that TLS-induced losses account for the residual dissipation, with an estimated total TLS loss rate ≈2π × 2.5 Hz close to the observed ≈2π × 6 Hz.","tokens_in":33488,"tokens_out":5078,"duration_ms":50144,"significance":"If the results hold, the system sets a new benchmark for quantum control of ultra-high-Q mechanical oscillators at microwave frequencies: it combines deterministic single-phonon state preparation, Wigner negativity, and millisecond-scale dynamical-decoupling-protected coherence in a compact silicon device, exceeding previous circuit quantum acoustodynamics lifetimes by two orders of magnitude. The central measurements—avoided crossings, linear gem(V), vacuum Rabi oscillations, and directly measured ~20 ms decays—are internally consistent and presented with stated uncertainties. The main quantitative milestone, the 25 ms intrinsic lifetime, depends on a model-based correction whose raw inputs are not fully disclosed; this is a transparency issue rather than evidence of error. The manuscript also provides a detailed and self-critical account of loss mechanisms, including explicit acknowledgment that the fluctuator model is phenomenological.","major_comments":[{"comment":"The quoted intrinsic T1 = 25 ± 2 ms is not directly measured but obtained by subtracting the inverse-Purcell rate Γκ = (gem/Δ)² κ from the measured total rate. At the 30–40 MHz detunings used, with gem/2π ≈ 160 kHz at 40 V and κ ≈ 5.9 × 10⁵ s⁻¹, Γκ is ≈9–12 s⁻¹ against an intrinsic rate ≈40 s⁻¹, i.e., a 20–30% correction. The paper states that the model fits Fig. S10 but does not show the raw Γ(Δ) points, fit residuals, or the independently measured κ; the data-availability statement ('upon reasonable request') does not allow independent verification. Please provide the raw data, the measured κ with uncertainty, fit residuals, and a sensitivity analysis of T1 to plausible variations in κ and to deviations from the (g/Δ)² scaling. Without this, the 25 ms milestone is not checkable from the manuscript.","section":"Appendix H1, Eq. (H1), Fig. S10"},{"comment":"The text states that relaxation is 'well explained by a single exponential' with 'no evidence of commonly observed degradation in the single-phonon regime.' However, the only data below ~1 phonon are the same decay traces used for the inverse-Purcell subtraction, and the correction is applied only at these low phonon numbers. A phonon-number-dependent intrinsic rate (e.g., weak saturable TLS loss) could be partially absorbed into the fitted intrinsic rate via the correction. The manuscript should show the uncorrected and corrected decay rates together with the phonon-number range of each measurement, and state explicitly how a saturable contribution would appear in the presented data.","section":"Mechanical decoherence, Fig. 4"}],"minor_comments":[{"comment":"The caption for the decay-rate parameters contains a typo: 'Parameters κe,q and κe,q denote...' should read 'κe,q and κe,r'.","section":"Table SI"},{"comment":"The abstract states 'T2 ≈ 1 ms', while the main text reports T2^CP2 = 1.02 ± 0.15 ms; consider specifying that this is the Carr-Purcell-2 coherence time in the abstract.","section":"Abstract"},{"comment":"The symbol 'kbT' in Eq. (I3) and surrounding text should be typeset as k_B T to avoid ambiguity with the TLS decay rate Γ1,TLS.","section":"Appendix I.2, Eq. (I3)"},{"comment":"In the caption and text, 'γmax = 105' appears without units; specify γmax = 10⁵ s⁻¹ as done in the main text of Appendix J.1.","section":"Appendix J.1, Fig. S13b"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a strong experimental contribution that fits the journal well. My main reservation is the lack of public raw data for the inverse-Purcell subtraction, which affects the headline 25 ms intrinsic lifetime; the journal's data-availability policy may warrant requiring a repository deposit before acceptance. The reference list is appropriate, and prior work by the same group is properly cited."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main news here is real: a transmon electrostatically coupled to a single-crystal silicon nanomechanical oscillator, with vacuum Rabi oscillations, Wigner-negative single-phonon states, direct TLS spectroscopy, and dynamical decoupling out to ~1 ms. The platform is genuinely new—no piezoelectric, just electrostatic forces on a phononic-bandgap silicon beam—and the measured Q ≈ 0.8×10^9 at 5 GHz is two orders of magnitude beyond previous cQAD systems. The appendices are dense and honest: thermometry, voltage-bias checks, leakage current, fabrication disorder simulations, and a detailed TLS loss budget. That is real work, and it shows. The one soft spot is exactly what the stress-test note says: the quoted intrinsic T1 = 25 ± 2 ms is not the directly measured single-phonon decay. What Fig. 4 shows is 19–21 ms decays at ~40 V; the 25 ms comes from subtracting an inverse-Purcell contribution using Eq. H1, with g_em, Δ, and κ as inputs. The subtraction is a 20–30% correction, so the headline number depends on modeling. They do show a detuning-dependent decay-rate plot (Fig. S10) and say the model fits, but raw points, residuals, and an independent κ are not in the paper, and data is only available on request. That is a transparency problem, not a fatal one: the directly measured decays are already long, and the strong coupling, Wigner negativity, and coherence results do not rest on the 25 ms number. Still, the abstract's clean \"T1 ≈ 25 ms\" overstates what was directly observed, and a referee should push for the raw subtraction data or a clearer qualifier. The TLS noise model in Appendix J is phenomenological—they call it that themselves—and the order-of-magnitude loss estimates in Appendix I carry large uncertainty. That is fine; the qualitative conclusions (TLS dominate dephasing, not relaxation) are supported by the avoided crossings and the telegrapher noise. Nothing in the paper looks internally inconsistent or circular. Bottom line: this deserves a serious referee and will likely be a milestone in quantum acoustics. I would send it to review, with a request that the inverse-Purcell data and the qubit κ measurement be released or at least shown in full. The paper is worth citing for the electrostatic interface and the millisecond-scale mechanical coherence, even if the exact 25 ms number is taken with a grain of salt.","headline":"Strong experimental paper: electrostatic coupling to an ultra-high-Q silicon mechanical oscillator with demonstrated quantum control; the 25 ms headline T1 is model-corrected but the directly measured ~20 ms decays and all qualitative claims are solid.","tokens_in":694,"tokens_out":861,"would_cite":true,"duration_ms":26985,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A transmon qubit couples strongly to a single-crystal silicon nanomechanical oscillator with an intrinsic $25$ ms single-phonon lifetime, enabling deterministic preparation of non-classical states and millisecond-scale coherence under…","keywords":["quantum acoustodynamics","electromechanical coupling","phononic crystal resonator","two-level-system defects","mechanical quantum memory","dynamical decoupling","Wigner tomography","phonon Fock state"],"falsifier":"Measure the single-phonon mechanical decay rate as a function of qubit\\textendash mechanics detuning $\\Delta$ and test the fit against $\\Gamma = \\Gamma_i + (g_{\\rm em}/\\Delta)^2\\kappa$ in the regime where the subtraction term is large; a deviation from the $(g_{\\rm em}/\\Delta)^2$ scaling, or an extrapolated $\\Delta \\to \\infty$ intercept that disagrees with a direct waveguide-based single-phonon measurement, would undercut the intrinsic $25$ ms claim. A device whose qubit can be detuned so far that the subtraction term is negligible should show a single-phonon lifetime near $25$ ms on its own.","tokens_in":32934,"feed_emoji":"💾","tokens_out":13261,"duration_ms":115537,"temperature":0.7,"pith_summary":"This paper claims that a microwave-frequency quantum state can be stored in a purely mechanical resonator with an energy lifetime of roughly 25 ms, while a superconducting qubit supplies the deterministic handle needed to write, protect, and read the state. The authors couple a transmon qubit to a single-crystal silicon nanobeam through a voltage-biased vacuum-gap capacitor, reaching strong coupling at $g_{\\rm em}/2\\pi \\approx 200$\\textendash$230$ kHz with large cooperativities ($C_{T_1}\\approx 1.5\\times10^5$, $C_{T_2}\\approx 150$). Using the qubit as actuator and probe, they prepare a single-phonon Fock state with a Wigner function that goes negative, measure a $25 \\pm 2$ ms intrinsic mechanical lifetime at the single-phonon level, and trace the dominant dephasing to individual two-level-system defects. Echo sequences run through the qubit lift the coherence time to $T_2^{\\rm CP2} = 1.02 \\pm 0.15$ ms. If these numbers hold, mechanical oscillators become compact bosonic memories that outlive superconducting qubits and planar microwave resonators by more than an order of magnitude.","feed_headline":"25 ms phonon memory stores quantum states","feed_subtitle":"A silicon nanobeam coupled to a transmon qubit reaches millisecond coherence under echo control.","key_machinery":"The load-bearing element is a phononic-crystal nanobeam resonator machined from single-crystal silicon, whose 'breathing' motion modulates a vacuum-gap capacitor attached to the transmon; the electrostatic interaction appears as a Jaynes\\textendash Cummings coupling $g_{\\rm em} = g_0 V_{\\rm dc}$ that scales linearly with bias voltage, so the qubit both controls and probes the phonon field. Phononic shields with a band gap exceeding $2$ GHz suppress clamping loss and keep mechanical energy away from lossy metal electrodes, while an on-chip notch filter stops the qubit from radiating into the DC bias line. The quantitative instrument behind the headline lifetime is the inverse-Purcell relation $\\Gamma = \\Gamma_i + (g_{\\rm em}/\\Delta)^2\\kappa$ (Appendix H1), which converts raw measured decays of $19$\\textendash$21$ ms into the intrinsic $25 \\pm 2$ ms by subtracting the qubit-mediated decay channel. Around this core sits the dephasing model, an ensemble of thermal fluctuators producing telegrapher frequency noise with $P(\\nu,\\gamma)\\propto 1/(\\gamma\\nu^2)$, which accounts for the exponential Ramsey and echo decays and for the observed echo efficiency of roughly six.","core_discovery":"The authors' claim is that electrostatic coupling, rather than piezoelectric transduction, can join a transmon qubit to an all-silicon phononic-crystal resonator without sacrificing the crystal's ultra-low acoustic loss. The device realizes the Jaynes\\textendash Cummings interaction $\\hat{H}/\\hbar = (\\omega_q/2)\\hat{\\sigma}_z + \\omega_m\\hat{b}^\\dagger\\hat{b} + g_{\\rm em}(\\hat{\\sigma}_+\\hat{b} + \\hat{\\sigma}_-\\hat{b}^\\dagger)$, with $g_{\\rm em}$ linear in the applied bias voltage; at $50$ V the resolved-mode splitting and vacuum Rabi oscillations place the system in the strong-coupling regime. Using the qubit for state transfer and readout, the paper reports an intrinsic single-phonon lifetime of $T_1 = 25 \\pm 2$ ms at $4.9$ GHz ($Q \\approx 0.8\\times10^9$), obtained by subtracting the inverse-Purcell contribution $\\Gamma = \\Gamma_i + (g_{\\rm em}/\\Delta)^2\\kappa$ from raw lifetimes of $19$\\textendash$21$ ms, and it confirms consistent lifetimes across an array of oscillators and across three phonon-number ranges. The same toolbox yields a phase-averaged Wigner function with negativity $W(0) = -0.10 \\pm 0.02$ for the $|1\\rangle_m$ Fock state, and voltage-tuned spectroscopy reveals about sixty individual defects whose coupling and Stark-shift statistics match the standard tunneling model of glassy TLS. Finally, Carr\\textendash Purcell decoupling sequences executed through the qubit extend the mechanical coherence time from $T_2^* = 64 \\pm 4\\,\\mu$s to $T_2^{\\rm CP2} = 1.02 \\pm 0.15$ ms, supporting the paper's framing of the system as a mechanical quantum memory for microwave photons.","pith_inferences":["A consequence the authors leave implicit: with $T_1$ already much longer than the transmon's own coherence, the practical ceiling on memory performance may shift from the mechanics to the qubit multiplexer, so improving transmon lifetimes or swap fidelities could pay off more than further phononic engineering.","Because the electrostatic coupling is material-agnostic, the same recipe could be transplanted to other ultralow-loss crystals, such as diamond or sapphire, where the TLS density might be even lower; fabricating the identical resonator geometry in a different crystal would be a direct test of whether silicon's surface TLS bath, rather than the design, sets the 25 ms number.","The voltage knob that tunes TLS frequencies could be used actively to park the oscillator away from its worst few fluctuators, a TLS-avoidance protocol suggested by the voltage dependence of the noise spectra but not proposed in the paper.","The measured TLS density and deformation potentials on the silicon beam provide a quantitative benchmark for the standard tunneling model in a nanostructure with an engineered phononic density of states, so surface passivation experiments that deplete the TLS bath and watch $T_2^*$ and the echo times move would be a clean, testable extension."],"forward_implications":["Mechanical oscillators of this kind would store microwave-encoded quantum states for tens of milliseconds in footprints far smaller than planar resonators, making them practical bosonic memory elements.","The strong-coupling, high-cooperativity operation makes deterministic single-phonon preparation and measurement routine, and puts the strong dispersive regime\\textendash\\textendash universal control of a large phonon Hilbert space\\textendash\\textendash within reach.","Because the dominant dephasing is caused by slow TLS fluctuators, dynamical decoupling through the qubit recovers coherence to about 1 ms, and with higher-fidelity swap gates longer echo trains should push the coherence time toward the 25 ms energy-lifetime limit.","The observed TLS statistics and the absence of saturable loss at the single-phonon level indicate that dissipation is set by a sparse defect bath in the silicon, so phononic and surface engineering could push lifetimes still higher.","Reaching the strong dispersive regime would allow biased-noise bosonic qubits encoded in gighertz-frequency mechanical oscillators, a concrete route toward fault-tolerant quantum computation with mechanical elements."],"supporting_citations":[{"why":"Supplies the phononic-bandgap resonator design with ultralong phonon lifetime that this device's shields and loss analysis build on.","marker":"[14]"},{"why":"Prior electrostatic electromechanical interface from the same group; provides the capacitive coupling mechanism and the equivalent-circuit model used here.","marker":"[28]"},{"why":"The piezoelectric 'mechanical qubit' system whose lifetime this work exceeds by over two orders of magnitude; the key comparison baseline.","marker":"[10]"},{"why":"Provides the qubit-mechanics Rabi oscillation readout and Wigner tomography method used to characterize non-classical phonon states.","marker":"[19]"},{"why":"Gives the Rabi-population thermometry protocol used to measure the residual thermal occupations of qubit and mechanics.","marker":"[39]"},{"why":"Theoretical framework for TLS-induced loss, noise, and nonlinearity under reduced phononic dimensionality; basis for the decoherence and dephasing calculations.","marker":"[44]"},{"why":"Source of the inverse-Purcell formula used in Eq. (H1) to subtract the qubit's contribution and extract the intrinsic 25 ms lifetime.","marker":"[83]"},{"why":"Provides the Gaussian-noise Ramsey/echo theory against which the measured exponential decays and echo efficiency are benchmarked.","marker":"[51]"},{"why":"Establishes the extremely low mechanical loss of single-crystal silicon at low temperature, the material premise for the oscillator design.","marker":"[20]"}],"fun_headline_variants":["25 ms phonon memory stores quantum states","Millisecond phonon memory for quantum states","Strong coupling to a 25 ms phonon memory","Phonon memory reaches millisecond coherence","Qubit couples to 25 ms phonon memory"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline $25$ ms lifetime rests on the assumption that mechanical decay through the qubit is exactly described by the inverse-Purcell formula $\\Gamma = \\Gamma_i + (g_{\\rm em}/\\Delta)^2\\kappa$ with a fully characterized qubit decay rate $\\kappa$; if that model or the measured $\\kappa$ is inaccurate, the quoted intrinsic lifetime shifts.","fun_headline_variants_meta":{"raw":{"variants":["25 ms phonon memory stores quantum states","Millisecond phonon memory for quantum states","Strong coupling to a 25 ms phonon memory","Phonon memory reaches millisecond coherence","Qubit couples to 25 ms phonon memory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000634,"raw_usage":{"total_tokens":3034,"prompt_tokens":1164,"completion_tokens":1870,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":780,"completion_tokens_details":{"reasoning_tokens":1799}},"tokens_in":780,"tokens_out":1870,"duration_ms":14372,"temperature":1.0,"reasoning_tokens":1799,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T18:19:20.899610+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the single-phonon mechanical decay rate as a function of qubit\\textendash mechanics detuning $\\Delta$ and test the fit against $\\Gamma = \\Gamma_i + (g_{\\rm em}/\\Delta)^2\\kappa$ in the regime where the subtraction term is large; a deviation from the $(g_{\\rm em}/\\Delta)^2$ scaling, or an extrapolated $\\Delta \\to \\infty$ intercept that disagrees with a direct waveguide-based single-phonon measurement, would undercut the intrinsic $25$ ms claim. A device whose qubit can be detuned so far that the subtraction term is negligible should show a single-phonon lifetime near $25$ ms on its own.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Theoretical framework for TLS-induced loss, noise, and nonlinearity under reduced phononic dimensionality; basis for the decoherence and dephasing calculations."},{"cited_title":"Minkov, U","cited_arxiv_id":null,"evidence_quote":"Source of the inverse-Purcell formula used in Eq. (H1) to subtract the qubit's contribution and extract the intrinsic 25 ms lifetime."},{"cited_title":"Catto, W","cited_arxiv_id":null,"evidence_quote":"Provides the Gaussian-noise Ramsey/echo theory against which the measured exponential decays and echo efficiency are benchmarked."}],"review_version":1}