{"id":"09af18ec-e5f4-48c7-85e6-c2ea00e33a6b","arxiv_id":"2501.11627","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":10,"one_line_summary":"A shadow-evaporated InAs Andreev spin qubit shows a spin relaxation time around 30 microseconds, comparable to etched devices, implying that surface disorder and quasiparticle poisoning are not the dominant relaxation mechanisms.","lead":"Andreev spin qubits are a new kind of superconducting quantum bit; this paper measures how long the spin state survives in a carefully made indium arsenide nanowire weak link. The measured relaxation time reaches about 30 microseconds, no better than older, more disordered devices, which suggests surface damage and stray quasiparticles are not what limits these qubits.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The null result cannot establish that disorder and quasiparticles do not limit T1, because the measured weak link's lower disorder is assumed from ref [16] and the comparison baseline is a different setup.","rationale":"The reader's CONDITIONAL verdict is well matched to the evidence. The central T1 measurement (~30 µs) and direct-readout design are credible, and the paper is unusually transparent about degeneracies and caveats, including the ancillary-level parameter ambiguity and the double-exponential long-pulse behavior in Appendix F. Data and code are deposited. The weak point is interpretive: the headline conclusion that surface disorder and QP poisoning do not limit spin relaxation requires that this specific wire be in the lower-disorder class and that the Hays data be a valid same-setup baseline. Neither condition is established. The paper's own word 'suggestive' is appropriate, and a CONDITIONAL verdict that asks for a matched control or disorder verification is the right outcome. I do not see an internal inconsistency that would justify REJECT; the concern is about the strength of the causal claim, not the validity of the measured T1. A secondary observation is that the abstract's mention of comparable dephasing rates is not supported by an explicit T2 measurement, but the main disorder/QP claim does not depend on that phrase. Therefore no verdict change is recommended.","tokens_in":18162,"tokens_out":7164,"duration_ms":84055,"concrete_test":"Co-fabricate on the same chip, from the same nanowire batch, two weak links with identical resonator and gate geometry, one shadow-evaporated and one etched; measure T1 vs flux and gate in the same cooldown, and compare at matched spin-flip frequency (e.g., tune Vg so both have Eσ/h near 500 MHz) and matched detuning from the Kramers point. If the shadow device still shows no >2σ improvement after matching, the Sec. IV conclusion survives; if it shows longer T1, the paper's negative result was confounded by device-to-device differences rather than isolating disorder.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The Sec. IV inference is a causal negative: because this shadow-evaporated weak link should have lower atomic-scale disorder (per ref [16]) and a gap-engineered QP filter, observing T1 ~30 µs comparable to Hays et al. is taken to mean that 'atomic-scale disorder induced by the etch process ... does not induce new spin relaxation mechanisms.' The load-bearing premise is that the measured device actually realizes the lower-disorder regime and that the only relevant differences from the literature baseline are disorder and QP poisoning. Neither is verified for this device. Lower disorder is imported from a separate shadow-junction study; no disorder-sensitive characterization (STM, normal-state conductance fluctuations, mobility) is reported for this wire or a co-processed sister device. The baseline is literature data [4] from a different setup, with different Eσ/h (0.23-1.5 GHz across the paper's own bias points), charging energy, microwave environment, and cooldown history; T1 is frequency-dependent (Appendix H, row iv), so matching a single point near 470-580 MHz does not control these confounds. With n=1 and no same-setup etched control, the null result is also consistent with this particular weak link being no less disordered or having a different dominant relaxation channel. The paper's 'suggestive' hedge is honest, but the Sec. IV sentence overstates what one device can establish.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports measurements on an InAs nanowire weak link fabricated with in situ shadow-defined junctions, using a microwave resonator designed to be directly sensitive to the spin-dependent inductance of the weak link, together with superconducting gap engineering intended to trap quasiparticles. The authors identify stable odd-parity bias points, map the spin-flip transition by two-tone EDSR, and measure the spin relaxation time T1 over a wide gate-voltage and flux range, finding a maximum near 30 microseconds, comparable to the best previously reported values for etched weak links. They also observe a higher-frequency transition that they attribute to parity flipping via an ancillary sub-gap state. From the lack of T1 improvement relative to literature values, they suggest that atomic-scale disorder induced by etching and quasiparticle poisoning are not limiting spin relaxation in InAs nanowires. The appendices provide resonator design, fabrication, additional bias points, temperature dependence, and an estimate of the quasiparticle-poisoning bound.","tokens_in":18480,"tokens_out":9303,"duration_ms":97581,"significance":"The experimental contributions are substantial: the direct inductance-based readout is validated by the approximately 10% agreement between the spin-orbit energy E_sigma extracted from the readout shift and from two-tone EDSR; T1 is measured in the time domain with reported uncertainties; the gap-engineering approach yields an estimated poisoning bound of 0.7 ms; and the data and code are publicly available. If the null result is accepted, it provides an important constraint on relaxation mechanisms in Andreev spin qubits and supports the transferability of the design strategies to other materials. The interpretation, however, rests on a single-device comparison to a literature baseline from a different setup and on an assumed lower-disorder property imported from Ref. [16]; the causal claim in Sec. IV is therefore stronger than the evidence supports. The paper's hedged language ('suggestive') is appropriate, but the abstract and Sec. IV should be brought in line with the demonstrated scope. No circularity issue is present: T1 is a direct time-domain measurement and E_sigma is independently cross-checked.","major_comments":[{"comment":"Section IV concludes that the similar T1 'suggest[s] that the atomic-scale disorder induced by the etch process in previous experiments does not induce new spin relaxation mechanisms.' This is a causal negative that requires that (i) the measured shadow-evaporated weak link actually realizes the lower-disorder regime and (ii) the only relevant difference from the etched-device baseline is disorder and quasiparticle poisoning. Neither is demonstrated for this device: the lower-disorder property is imported from Ref. [16] without disorder-sensitive characterization of this wire or a co-processed sister device, and the comparison baseline (Hays et al., Ref. [4]) comes from a different setup with different E_sigma/h (1.4-1.5 GHz in Appendix H columns a and b versus 230 MHz in column c), charging energy, microwave environment, and cooldown. Appendix H row (iv) itself shows a dependence of T1 on spin-transition frequency, so matching a single point near 470-580 MHz does not control these confounds. With n=1 and no same-setup etched control, the null result is also consistent with this particular weak link being no less disordered or with a different dominant relaxation channel. I recommend either softening the causal claim to an explicit hypothesis or adding supporting evidence, such as normal-state conductance fluctuations, a same-setup etched control, or a quantitative treatment of the frequency dependence.","section":"Sec. IV; Appendix H"},{"comment":"The main-text T1 values are obtained from single-exponential fits to the first 250 microseconds (Sec. III B and Fig. 3(c)), but Appendix F shows that long saturation pulses produce double-exponential decays with a slow component of 138-212 microseconds (Fig. S5(e)-(f)). The authors state that the double-exponential appears less frequently with the short Gaussian pulse, but they do not show representative short-pulse decay curves over the full time range or test the sensitivity of the extracted T1 to the choice of fit window. Without such a check, the reported approximately 30 microsecond time constant could be a truncation artifact of the 250 microsecond window. Please provide the full short-pulse decay data and quantify how T1 changes with the fit range.","section":"Appendix F; Sec. III B"},{"comment":"The stated maximum T1 of approximately 30 microseconds in Sec. III B and Fig. 3(c) is inconsistent with the value T1 = 40.3 +/- 1.8 microseconds reported in Appendix F, Fig. S5(d) at Vg = -1.786 V and Phi = -0.27 Phi0. The main text notes that Fig. 3(c) is averaged over repeated scans, but the discrepancy is not explained. Please clarify whether the appendix value is from a single scan, whether a different E_sigma was used for the drive frequency, and why the averaged map shows a lower maximum. The abstract's headline value ('about 30 microseconds') should be reconciled with the appendix.","section":"Sec. III B; Appendix F"}],"minor_comments":[{"comment":"The sentence 'about 2% of the total 'active' gate voltage range ... exhibits a stable odd parity state for for all flux' contains a duplicated 'for', and the Fig. 2 caption contains 'mayb' instead of 'maybe'.","section":"Sec. II B"},{"comment":"There are typographical errors: 'Ultrasonicaton' in Appendix B and 'attenuaors' in Appendix C should be corrected to 'Ultrasonication' and 'attenuators', respectively.","section":"Appendices B and C"},{"comment":"The value of E_sigma/h is quoted as 255 MHz from single-tone readout, approximately 240 MHz from two-tone spectroscopy, and 237 MHz for the relaxation-time measurements; please state explicitly which value is used for each figure and why the value used for the relaxation measurements differs from the two-tone value.","section":"Sec. III A; Appendix E"},{"comment":"Equation (2) uses both phi and varphi for the superconducting phase; please use a single symbol and define the sign convention for the |E_sigma sin(phi)| term.","section":"Sec. III C; Eq. (2)"},{"comment":"The statement that the spin is 'thermalized down to near 15 mK' is stronger than the evidence (visibility of the transition down to 250 MHz); consider rephrasing to 'consistent with an effective temperature near 15 mK'.","section":"Sec. III A"},{"comment":"The abstract uses 'in situ shadow evaporation' to describe the weak-link formation, while Appendix B explains that the weak link is defined by shadows from other nanowires during Al deposition; consider clarifying the terminology to avoid confusion with metal shadow evaporation.","section":"Abstract; Appendix B"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the paper is a solid experimental advance, but its headline conclusion is a single-device null result framed as a causal statement. The authors hedge in the text with 'suggestive', which helps, but the abstract and Sec. IV still go beyond what one device can establish. I see no circularity problem: T1 is measured directly in the time domain and E_sigma is cross-checked by two independent methods. The main revision should focus on reframing the disorder/quasiparticle conclusion and on the fit-window and consistency issues in Appendix F. Whether a single-device null result is sufficient for the journal is an editorial judgment, but the experimental methodology is strong enough that the paper would be suitable after the requested revisions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one. It is a careful experimental paper with real new capability: a resonator readout directly sensitive to spin-dependent inductance, so readout no longer needs higher orbital states; a full flux-gate map of T1; and gap-engineered quasiparticle filtering that appears to work. The central T1 numbers are credible—two-tone EDSR and the readout shift agree within 10%, and the data and code are on Zenodo. The observation of an ancillary sub-gap transition with spin-orbit effects is a nice bonus. If you work on Andreev spin qubits, this is worth a serious look.\n\nThe soft spots are mostly about interpretation, not measurement. The Sec. IV claim that atomic-scale disorder and QP poisoning do not limit T1 is a causal negative built on two premises that are not verified: that this particular shadow-evaporated weak link is actually lower-disorder than etched wires (imported from ref [16], not demonstrated on this device), and that matching one literature T1 value near 500 MHz controls for differences in spin-orbit energy, charging energy, and microwave environment. It does not. The paper's 'suggestive' hedge is honest, but the sentence right after it—'does not induce new spin relaxation mechanisms'—outruns the data. One device, no same-setup etched control.\n\nMinor points: the main T1 fit is single-exponential over the first 250 microseconds, and the long-pulse data in Appendix F show double-exponential decay. They disclose this and speculate about heating, but it does mean the reported T1 is a short-time effective value. The ancillary-level model has degenerate parameters and they say so; that is fine for a fit, just do not lean on it for the central conclusion. The poisoning bound (0.7 ms) is coarse, but they label it as such.\n\nThe comparison with Hays et al. is legitimate if treated as suggestive, not conclusive. The flux-dependence trend—T1 rising away from the Kramers point, then falling—is consistent with hyperfine-mediated relaxation, and their discussion of that is reasonable.\n\nWho benefits: anyone in the Andreev-spin-qubit or superconducting-nanowire community, especially people thinking about materials choices for the next generation. The design transfer to Ge and C is a sensible message. I would send it to review—the experimental contribution stands even if the disorder conclusion gets softened in revision.","headline":"Solid time-domain T1 study with real new readout capability, but the disorder/QP conclusion outruns what one device can establish.","tokens_in":19060,"tokens_out":1675,"would_cite":true,"duration_ms":17773,"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 shadow-evaporated InAs weak link shows Andreev spin relaxation near 30 microseconds, matching etched devices and suggesting atomic-scale disorder does not limit the spin lifetime.","keywords":["Andreev spin qubits","spin relaxation time","InAs nanowire weak link","shadow evaporation","quasiparticle poisoning","gap engineering","dispersive readout","parity flipping"],"falsifier":"Measure the same shadow-evaporated weak link, then deliberately introduce controlled surface damage (for example a light argon mill) and watch whether $T_1$ drops; if it does, the paper's conclusion is false. Alternatively, run a shadow-evaporated and an etched weak link on the same chip with identical readout and gap engineering, and check whether their $T_1$ values differ.","tokens_in":17954,"feed_emoji":"⚛️","tokens_out":13628,"duration_ms":127361,"temperature":0.7,"pith_summary":"The paper tackles an open question for Andreev spin qubits — qubits built from the spin of a microscopic bound state in a superconductor–semiconductor weak link, where the spin dictates the supercurrent — namely, what limits their energy relaxation. The authors fabricate an InAs nanowire weak link by shadow evaporation, a route that avoids the atomic-scale surface damage caused by the etching used in all earlier relaxation studies, and add two refinements: a microwave resonator whose frequency shift responds directly to the spin-dependent inductance (no need for higher orbital states), and a gap-engineering layer that blocks quasiparticles (stray excitations) from entering the weak link. With all three ingredients in one device, the measured spin relaxation time reaches about 30 microseconds, matching the best values reported for etched wires. The authors therefore suggest, with explicit caution, that atomic-scale disorder and quasiparticle poisoning are not the dominant causes of Andreev spin relaxation in these nanowires. If right, the practical payoff is a redirection: further polishing the weak-link surface will not stretch spin lifetimes, and progress instead hinges on nuclear-spin-free host materials and on understanding why relaxation depends so weakly on the transition frequency.","feed_headline":"Andreev spin states last 30 microseconds in cleaner InAs weak links","feed_subtitle":"Shadow evaporation does not stretch spin lifetime, hinting surface disorder is not the limit for Andreev qubits.","key_machinery":"The load-bearing piece is the spin-dependent energy-phase relation of the Andreev level, $E_s(\\varphi) = E_\\sigma \\sigma_z \\sin\\varphi$, whose second derivative sets the inverse inductance seen by the microwave resonator. The design chooses a shared inductance of about $200$ pH (sensitivity $S = 803$ MHz/nH) so the dispersive readout responds directly to the spin part of the energy-phase relation, removing the need for higher orbital states. Two mechanisms support the measurement: gap engineering, in which the $25$ nm epitaxial Al on the nanowire has a larger gap than the $100/200$ nm Al resonator and contacts, creating a $\\sim 13$ GHz barrier that traps quasiparticles away from the weak link; and shadow evaporation, which produces the $\\sim 136$ nm weak link without etching the semiconductor. The even-parity branch is fit with a resonant-level model and the odd-parity branch with the minimal $E_0\\cos\\varphi + E_\\sigma\\sigma_z\\sin\\varphi$ model; a four-state model including an ancillary dot level captures the observed parity-flipping transition.","core_discovery":"The central claim is that in a weak link made by in situ shadow evaporation — which prior microscopy and transport work ties to reduced atomic-scale disorder — the Andreev spin relaxation time is not improved relative to etched devices, peaking near $T_1 \\approx 30\\,\\mu\\mathrm{s}$ (with single scans up to $40\\,\\mu\\mathrm{s}$) at the centre of an odd-parity stability window, $V_g \\approx -1.786$ V and $\\Phi \\approx -0.25\\Phi_0$. The readout, tuned to the spin-dependent inductance, gives a spin-orbit energy $E_\\sigma/h \\approx 230$–$255$ MHz from two independent routes, and the paper maps $T_1$ across the full gate-flux range of that bias point, finding the same qualitative flux trend as earlier work: relaxation slows as flux detunes from the Kramers degeneracy. A higher-frequency transition is observed whose gate and flux dependence matches a model of quasiparticle transfer between the Andreev level and an ancillary sub-gap dot — a parity-flipping process rather than a transition within the odd-parity manifold. The authors state the disorder/poisoning conclusion as suggestive, since the comparison rests on literature baselines measured in different setups.","pith_inferences":["If the paper is right, a same-chip comparison between a shadow-evaporated and a deliberately etched weak link with identical readout and gap engineering is the direct test; a difference in $T_1$ there would overturn the conclusion.","The paper's ancillary-level model yields a testable signature: an extra gate that tunes the dot energy should move the parity-flipping transition frequency quadratically in gate voltage, and the transition should disappear when the dot is drained or pushed out of resonance.","A single-shot parity readout, which this experiment lacked, could turn the 0.7 ms poisoning lower bound into a direct poisoning-rate measurement and quantitatively test the gap-engineering mechanism.","The paper's speculation that the nuclear spin bath may limit relaxation suggests a direct experiment: repeat the measurement in a nuclear-spin-free host material (e.g., a carbon-based weak link) and look for a lengthened $T_1$ with the same readout and gap engineering."],"forward_implications":["If the conclusion holds, improving the atomic-scale quality of the weak-link surface will not, by itself, extend Andreev spin relaxation times.","The direct spin-inductance readout removes the need for orbital-state access, making relaxation measurements feasible in short weak links and in material platforms with widely spaced orbitals such as germanium and carbon.","The gap-engineering scheme with thickness-mismatched aluminum gives a quasiparticle-poisoning lower bound of about 0.7 ms, one to two orders of magnitude better than earlier devices without intentional strong charging energy.","The observed parity-flipping transition, fit with $U+\\epsilon_{qd}-\\epsilon_A(\\varphi,V_g)-E_0\\cos\\varphi+|E_\\sigma\\sin\\varphi|$, implies that microwave driving can flip parity via ancillary sub-gap states, a process that must be accounted for in future qubit control.","The flux dependence of $T_1$ around both Kramers points is inconsistent with or opposite to predictions from $1/f$ charge or flux noise and phonon-mediated relaxation, narrowing the set of candidate relaxation mechanisms."],"supporting_citations":[{"why":"Supplies the baseline relaxation times from etched wires (up to ~50 µs) and the flux trend of $T_1$ that this device reproduces.","marker":"[4]"},{"why":"Provides the microscopy/transport evidence that shadow-evaporated weak links have lower atomic-scale disorder than etched ones.","marker":"[16]"},{"why":"Establishes the gap-engineering approach of trapping quasiparticles in lower-gap aluminum films, adapted here to protect the weak link.","marker":"[17]"},{"why":"Demonstrates gap-engineering and quasiparticle-trapping behaviour in aluminum devices, supporting the expected ~13 GHz barrier.","marker":"[18]"},{"why":"Proposes the ancillary sub-gap dot model for parity flips that the paper's higher-frequency transition is matched against.","marker":"[20]"},{"why":"Derives the spin-dependent inductance/dispersive-shift relation that the direct spin readout is designed around.","marker":"[22]"},{"why":"Provides the resonant-level model for the even-parity energy-phase relation used to extract $\\Delta_\\mathrm{eff}$ and $\\tau_\\mathrm{eff}$ from readout.","marker":"[23]"},{"why":"Gives the minimal spin-dependent energy-phase model $E_0\\cos\\varphi + E_\\sigma\\sigma_z\\sin\\varphi$ and predictions for $1/f$ charge/flux noise relaxation.","marker":"[24]"}],"fun_headline_variants":["Shadow evaporation doesn't lengthen Andreev spin T1","Andreev spin relaxation time stays ~30 us despite clean fabrication","No gain in Andreev spin T1 from shadow evaporation","Cleaner InAs weak links don't extend Andreev spin lifetimes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion that surface disorder and quasiparticle poisoning do not limit $T_1$ assumes that this particular shadow-evaporated nanowire really has lower atomic-scale disorder than the etched wires in earlier work, and that the earlier relaxation times are a fair same-setup baseline.","fun_headline_variants_meta":{"raw":{"variants":["Shadow evaporation doesn't lengthen Andreev spin T1","Andreev spin relaxation time stays ~30 us despite clean fabrication","No gain in Andreev spin T1 from shadow evaporation","Cleaner InAs weak links don't extend Andreev spin lifetimes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000588,"raw_usage":{"total_tokens":2800,"prompt_tokens":1025,"completion_tokens":1775,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":1701}},"tokens_in":641,"tokens_out":1775,"duration_ms":16468,"temperature":1.0,"reasoning_tokens":1701,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T18:02:04.902006+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the same shadow-evaporated weak link, then deliberately introduce controlled surface damage (for example a light argon mill) and watch whether $T_1$ drops; if it does, the paper's conclusion is false. Alternatively, run a shadow-evaporated and an etched weak link on the same chip with identical readout and gap engineering, and check whether their $T_1$ values differ.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the baseline relaxation times from etched wires (up to ~50 µs) and the flux trend of $T_1$ that this device reproduces."},{"cited_title":"Scappucci, C","cited_arxiv_id":null,"evidence_quote":"Provides the microscopy/transport evidence that shadow-evaporated weak links have lower atomic-scale disorder than etched ones."},{"cited_title":"Aliferis and J","cited_arxiv_id":null,"evidence_quote":"Establishes the gap-engineering approach of trapping quasiparticles in lower-gap aluminum films, adapted here to protect the weak link."},{"cited_title":"Webster, S","cited_arxiv_id":null,"evidence_quote":"Demonstrates gap-engineering and quasiparticle-trapping behaviour in aluminum devices, supporting the expected ~13 GHz barrier."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proposes the ancillary sub-gap dot model for parity flips that the paper's higher-frequency transition is matched against."},{"cited_title":"Connolly, P","cited_arxiv_id":null,"evidence_quote":"Derives the spin-dependent inductance/dispersive-shift relation that the direct spin readout is designed around."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the minimal spin-dependent energy-phase model $E_0\\cos\\varphi + E_\\sigma\\sigma_z\\sin\\varphi$ and predictions for $1/f$ charge/flux noise relaxation."}],"review_version":1}