{"id":"56067841-2f94-4699-a314-d5543aaf7516","arxiv_id":"2508.10889","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Niobium hydride precipitates were detected in the Nb films of Rigetti superconducting qubit chips, implicating them as a new decoherence source.","lead":"This paper reports that tiny niobium-hydride particles form inside the niobium films of commercial superconducting qubit chips when they are cooled, and that these particles can degrade qubit performance. The finding points to a new, previously unaccounted source of decoherence and cooldown-to-cooldown variability in quantum computers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The entire hydride phase identification rests on a single weak XRD peak at d = 2.45 Å, measured on a co-fabricated film, not on the Rigetti chip; if this assignment is wrong, the AFM features are uncharacterized 'unknown structures' and the central discovery claim lacks direct structural support.","rationale":"The most load-bearing step in the argument is the identification of the NbH0.89 phase. Everything downstream—'hydride precipitates form in qubit chips' and 'are a new decoherence source'—depends on that crystallographic assignment. The authors themselves narrow the evidence to a single XRD reflection, note the absence of other hydride lines, and use a sample that was co-fabricated rather than the device itself. This is a textbook situation where the identification is underdetermined. I therefore agree with the reader that this is the weakest assumption. The concern is not that the authors are wrong; it is that the current data do not uniquely force the conclusion. Cryo-AFM shows temperature-dependent topography, but the caption calls the features 'unknown structures' and no compositional analysis is reported. ToF-SIMS shows hydrogen, but lattice hydrogen and hydride precipitates are different physical states. The bulk-cavity RF degradation is compelling motivation, but it is a bulk measurement with a deliberately large H load; translating 'factor of 2.5 H concentration gives 10x Q0 drop' into a microscopic rate for sparse, nanoscale precipitates in a 150 nm film is not quantified. A single additional experiment—searching for multiple hydride reflections or doing cryo-TEM—would resolve the ambiguity. Because the authors have provided a coherent narrative and some converging observations, this is not grounds for rejection, but it is grounds for the conditional verdict already given. I would keep the verdict unchanged; the onus is now on the authors to provide the missing structural confirmation. The TLS claim in the abstract is internally inconsistent with the Discussion's caveat, but correcting that would not alter the existence question; it is a separate overreach.","tokens_in":8944,"tokens_out":7093,"duration_ms":78422,"concrete_test":"Re-measure the Nb film (ideally from the same deposition run as the actual Rigetti chip, or a micro-XRD spot on the chip pad) by cryo-GIXRD at ~120 K using a high-efficiency area detector and long integration, aiming to detect at least two additional reflections of orthorhombic NbH0.89 besides the 2.45 Å line. Analyze the full pattern by Rietveld refinement including the NbH0.89 phase and report detection limits for the hydride phase fraction. Independently, prepare a FIB lamella from the same film and perform cryogenic TEM/electron diffraction to image hydride precipitates and obtain selected-area diffraction. If no additional hydride reflections and no hydride particles are found, the single XRD peak is insufficient to establish the phase.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central existence claim reduces to the assignment of the weak GIXRD reflection at d = 2.45 Å (Fig. 3, Results) to the (111) plane of orthorhombic NbH0.89. The text explicitly states that no other hydride reflections were observed, so this one line carries the entire crystallographic identification. It was measured on a film co-fabricated with, not cut from, the Rigetti chip, and the direct morphological evidence on the actual chip—the cryo-AFM features—is labeled 'unknown structures' in the Fig. 1 caption and was not chemically identified. ToF-SIMS shows H enrichment near the Nb surface (Fig. 2) but cannot distinguish interstitial hydrogen from a hydride phase. The bulk-cavity RF data (Fig. 4) demonstrate that externally loaded H degrades Q0 in bulk Nb, but this is a different geometry and material system; the quantitative extrapolation to 100–200 nm qubit films and to T1/T2 is not derived. If the 2.45 Å peak is a misidentification—e.g., a surface oxide, a different hydride stoichiometry, or a strain-related Nb reflection—the discovery claim loses its only direct structural evidence, and the remaining observations are compatible with, but do not prove, hydride precipitates.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports evidence for niobium hydride precipitates in HiPIMS-deposited Nb films on Si substrates in superconducting qubit chips fabricated at Rigetti Computing. The evidence is drawn from room-temperature and cryogenic AFM showing surface features that appear on cooling, ToF-SIMS depth profiles showing hydrogen enrichment beneath the native oxide, synchrotron GIXRD on a co-fabricated film showing a weak reflection at d = 2.45 Å assigned to NbH0.89 (111), and RF measurements on a 3D bulk Nb cavity after deliberate hydrogen loading showing degradation of the quality factor. The central claim is that these hydrides are a previously unidentified source of decoherence in superconducting qubits, contributing to quasiparticle and possibly TLS losses and explaining cooldown-to-cooldown variability.","tokens_in":9262,"tokens_out":3786,"duration_ms":48183,"significance":"If the central claim holds, the paper identifies a practically important decoherence mechanism for planar superconducting qubits, connecting established bulk SRF cavity hydride physics to the 2D qubit geometry. It proposes a concrete mitigation pathway (oxygen doping) and provides a plausible explanation for qubit performance variability across cooldowns. The study has notable strengths: it uses actual Rigetti qubit chips for AFM and SIMS, acquires temperature-dependent cryo-AFM and synchrotron XRD data, and leverages a controlled bulk-cavity hydrogen-loading experiment to quantify RF loss. The authors also explicitly acknowledge several limitations, including that their bulk study does not address TLS losses and that further film-based resonator studies are needed. These candid statements are to their credit, but they also delimit the strength of the conclusions as currently stated.","major_comments":[{"comment":"The entire crystallographic identification rests on a single weak reflection at d = 2.45 Å, assigned to the (111) plane of orthorhombic NbH0.89. The text explicitly states that no other hydride reflections were observed. The figure shows no error bars, peak intensity, signal-to-noise ratio, or detection limits, and the Rietveld refinement is mentioned without parameters or residuals. Alternative assignments (e.g., NbO, Nb2O5, NbC, strain-modified Nb reflections) must be quantitatively excluded. In addition, the XRD was performed on a co-fabricated film, not on the actual Rigetti qubit chip; the representativeness of this film for the qubit Nb pads needs justification. As written, this one peak carries the structural identification, which is a load-bearing point for the paper's central claim.","section":"§II, Fig. 3"},{"comment":"The AFM features on the actual qubit chip are labeled 'unknown structures' in the caption and are not chemically identified. Their interpretation as niobium hydrides relies solely on morphological similarity to features previously observed in bulk Nb SRF cavities. Since the XRD was not performed on this chip, the connection between the AFM features and the hydride phase is circumstantial. Direct compositional or structural evidence on the same features (e.g., TEM/EELS, EDX, or nano-SIMS) is needed to substantiate the claim that these are hydrides. If such evidence is not available, the text should explicitly state that the AFM features are consistent with, but not proof of, hydride precipitates.","section":"§II, Fig. 1 caption"},{"comment":"The ToF-SIMS depth profile shows hydrogen enrichment beneath the Nb2O5 layer, but hydrogen in interstitial solid solution cannot be distinguished from hydrogen in a hydride phase by this measurement. The statement that this hydrogen 'potentially drives Nb hydride formation' is a reasonable hypothesis but is not tested. No spatial correlation is shown between the H-enriched regions and the AFM features, and no estimate is given for whether the measured H concentration exceeds the solubility limit at the temperatures where the features appear. This limits the directness of the chemical evidence.","section":"§II, ToF-SIMS (Fig. 2)"},{"comment":"The bulk 3D Nb cavity data demonstrate that externally introduced hydrogen degrades Q0, but the quantitative extrapolation to qubit films is not derived. The abstract states that the authors 'quantify RF dissipation on a superconducting qubit' from these bulk-cavity measurements, yet the geometry, field regime, and dimensionality differ (bulk cavity vs. 100–200 nm film; <1 MV/m vs. single-photon powers). No model connects the measured H concentration and Q0 reduction to a hydride volume fraction or to qubit T1/T2. The paper itself concedes that the bulk study does not address TLS losses and that film-based resonator tests are needed. The wording of the abstract and conclusion should be tempered to reflect that the RF quantification is on a bulk model system, not on a qubit.","section":"§II, Fig. 4 and §III Discussion"}],"minor_comments":[{"comment":"The phrase 'quantify RF dissipation on a superconducting qubit' overstates what was measured; consider saying 'on a bulk Nb resonator as a model system for qubit-relevant RF dissipation.'","section":"Abstract and §III"},{"comment":"The text says 'X-ray penetration depth (λ = 0.68 Å)'; λ is the wavelength, not the penetration depth. Please correct and report the estimated penetration depth separately.","section":"§II, GIXRD setup"},{"comment":"The shift of the hydride peak by ~0.02 Å between cooldowns is reported without an uncertainty estimate. Provide the fit uncertainty and the temperature at which each measurement was taken.","section":"§II, Fig. 3"},{"comment":"The AFM features are called 'unknown structures' in the caption but 'hydride-related' in the text. This inconsistency should be resolved once the features are identified or explicitly labeled as presumed hydrides.","section":"§II, Fig. 1"},{"comment":"There is a typo: 'Regetti qubit device' should be 'Rigetti qubit device.'","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely and relevant topic for superconducting qubit materials research, and the authors have assembled complementary measurements on real devices. The main concern is the strength of the evidence for the hydride assignment, which currently rests on a single XRD peak on a co-fabricated film and on unlabeled AFM features. The bulk-cavity RF data are useful but their quantitative connection to qubit performance is asserted rather than derived. I believe the central claim is defensible, but it needs additional quantitative support and more cautious wording. No concerns about citation integrity or scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I've read the paper. The punchline: the basic claim is probably true—niobium hydrides can form in as-fabricated HiPIMS films on qubit chips—but the evidence as presented is a bit less direct than the abstract lets on.\n\nWhat's genuinely new: previous work (Ref 25) showed hydrides in Nb thin films after fluoride processing. This paper shows the phenomenon in as-deposited HiPIMS films on actual Rigetti chips: cryo-AFM sees temperature-dependent surface features, ToF-SIMS shows hydrogen accumulating just under the oxide, and temperature-dependent XRD shows a peak that appears only below ~150 K. That's a reasonable convergence. The bulk-cavity RF data are also solid evidence that hydrogen loading degrades Nb at low fields, and the practical suggestion about oxygen doping is sensible.\n\nWhere I'd push back. The XRD identification is the load-bearing piece, and it's one weak peak at d = 2.45 Å, on a co-fabricated film, not the chip itself, with no error bars or Rietveld detail shown. The AFM features are honestly labeled 'unknown structures.' SIMS shows H but doesn't distinguish interstitial H from hydride phase. The abstract's TLS claim is explicitly walked back in the Discussion. And the bulk-cavity results are a different geometry; mapping that to a 100 nm film's T1 is an extrapolation, though a reasonable one.\n\nNone of this kills the central claim—hydrides in Nb films are well documented in the SRF world, and the temperature-dependent peak is at least consistent. But the paper overstates certainty. The good news is the weaknesses are addressable: more XRD peaks, on the actual chip, with error bars, and a clearer separation of what is directly measured vs inferred.\n\nWho it's for: people working on Nb-based qubit materials and SRF cavities. It's a useful data point even if the headline is softened. I'd send it to peer review—it's substantive and the core question matters. I'd also bring it to a reading group to discuss how much evidence is enough to claim a 'discovery.'","headline":"Plausible case for niobium hydrides in qubit films, but the structural evidence rests on a single XRD peak on a surrogate film—worth refereeing, not a slam dunk.","tokens_in":9777,"tokens_out":2044,"would_cite":true,"duration_ms":23785,"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":"Niobium hydride precipitates form in superconducting qubit films and act as a previously unaccounted source of decoherence.","keywords":["niobium hydride","superconducting qubits","decoherence","quasiparticle loss","two-level system","cooldown variability","hydrogen contamination","cryogenic microscopy"],"falsifier":"Cryogenic transmission electron microscopy or scanning nano-X-ray diffraction on an actual qubit chip cooled below 150 K should resolve multiple NbH0.89 lattice reflections and precipitates whose number density scales with locally measured hydrogen content and inversely with measured T1. If the only hydride fingerprint remains the single 2.45 Å reflection, or if no precipitates are found inside the device film, the structural identification is not supported.","tokens_in":8788,"feed_emoji":"❄️","tokens_out":10596,"duration_ms":109996,"temperature":0.7,"pith_summary":"This paper sets out to show that niobium hydride precipitates form inside the niobium films of real superconducting qubit chips, and that they are a previously unrecognized source of decoherence. Using cryogenic atomic force microscopy, synchrotron X-ray diffraction, and time-of-flight secondary-ion mass spectrometry, the authors find hydrogen beneath the niobium surface oxide and observe a weak diffraction peak assigned to a NbH0.89 hydride phase appearing only between 150 K and 100 K, together with surface features that evolve during cooldown. They connect this to qubit behavior by showing, in a bulk niobium cavity used as a controlled hydrogen-loading testbed, that a modest rise in hydrogen concentration cuts the low-field quality factor by about an order of magnitude. If right, the work matters because hydride precipitates are non-superconducting, change size and location between cooldowns, and would explain why qubit performance varies across thermal cycles and ages with repeated cooldowns.","feed_headline":"Niobium hydrides are a new qubit decoherence source","feed_subtitle":"Non-superconducting hydride specks form in niobium films, shift between cooldowns, and drain qubit coherence.","key_machinery":"The load-bearing object is the niobium hydride precipitate, indexed as orthorhombic NbH0.89 with a (111) spacing d = 2.45 Å in space group Pnnn. It is a normal or weak-gap inclusion embedded in the superconducting niobium film and does its damage through proximity coupling: microwave fields experience a locally suppressed superconducting order parameter, and the inclusions act as quasiparticle sinks that increase dissipation and can trigger the high-field Q slope. The paper also uses a 1.3 GHz bulk niobium cavity with controlled hydrogen loading as a clean single-interface proxy to quantify how much RF loss a given hydrogen concentration produces.","core_discovery":"On the basis of cryogenic AFM, synchrotron grazing-incidence X-ray diffraction, and ToF-SIMS, the paper asserts that niobium films on silicon in superconducting qubits can precipitate nanocrystalline niobium hydride, indexed as orthorhombic NbH0.89, during cooldown beginning below about 150 K. The precipitates are non-superconducting inclusions; proximity coupling between them and the surrounding superconducting niobium degrades microwave performance. The same physics is demonstrated semiquantitatively in a 3D bulk niobium cavity whose hydrogen content was increased by mechanical grinding and electrochemical loading: roughly a 2.5-fold rise in hydrogen concentration produced an order-of-magn","pith_inferences":["If the identification holds, hydrogen should be treated as a first-class materials variable in qubit fabrication: controlled hydrogen-concentration measurements on actual devices would help separate cooldown-dependent hydride loss from stationary two-level-system loss.","A natural test would be to fabricate matched qubit chips with deliberately different hydrogen loads and track T1 and T2 across repeated cooldowns; the bulk-resonator data predict a monotone, cooldown-history-dependent degradation.","The mechanism suggests that 'aging' effects and sporadic coherence jumps may have a reversible microscopic cause—re-cooling into a different hydride configuration—rather than permanent damage, which cycling experiments could distinguish.","Because the authors note their bulk-cavity experiment cannot assess two-level-system losses from hydrides, extending the same hydrogen-loading protocol to film-based resonators would determine how much of qubit TLS loss overlaps with the hydride channel."],"forward_implications":["Cooldown-to-cooldown variation in qubit coherence can result from hydride precipitates that change size, morphology, and location on each thermal cycle.","Hydride regions have a suppressed superconducting gap and act as quasiparticle sinks, so even low-power operation suffers dissipation through the proximity effect.","Because hydrides form throughout the film thickness, changing film thickness alone will not eliminate this decoherence channel.","Hydrogen introduced during deposition or during wet and mechanical processing steps can be retained under the surface oxide and later precipitate, making fabrication controls a direct lever on qubit performance.","Introducing higher oxygen concentration into niobium films may trap hydrogen and suppress hydride formation, a practical mitigation route proposed by the paper."],"supporting_citations":[{"why":"Showed hydride precipitation in bulk niobium after different treatments, establishing the morphology and cooldown-cycle evolution that the film data are compared with.","marker":"[20]"},{"why":"Earlier direct observation of hydride formation in cavity-grade niobium, the methodological precedent for the diffraction and microscopy identification.","marker":"[21]"},{"why":"Proximity-breakdown model attributing RF loss and high-field Q slope to hydride inclusions, used to explain qubit dissipation.","marker":"[22]"},{"why":"Reference for Nb-H phase structures and thermodynamics, underpinning the assignment of hydrogen site occupancy and hydride phases.","marker":"[24]"},{"why":"Showed hydride formation and microwave loss in niobium thin films after fluoride chemical processing, the nearest thin-film analogue supporting film relevance.","marker":"[25]"},{"why":"Establishes hydrogen absorption and the low-temperature solubility threshold near 150 K that explains when hydrides appear.","marker":"[28]"},{"why":"Documents hydrogen-related Q degradation in niobium cavities, connecting dissolved hydrogen to macroscopic RF loss.","marker":"[29]"},{"why":"Direct nanometer-scale hydride observation in superconducting niobium, used to interpret the cryo-AFM surface features as hydrides.","marker":"[31]"},{"why":"Describes the fabrication process of the tested qubit device, providing the deposition and processing context for hydrogen incorporation.","marker":"[30]"}],"fun_headline_variants":["New culprit in qubit decoherence: niobium hydride specks","Hydrides in niobium films drain qubit coherence","Niobium hydrides: hidden defect inside superconducting qubits","Qubit performance drop traced to niobium hydride formation","Cryogenic hydrides: fresh source of qubit energy loss"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The single weak X-ray reflection at d = 2.45 Å, seen only between about 100 and 150 K and assigned to the (111) plane of orthorhombic NbH0.89, is genuinely niobium hydride, and it is representative of the actual qubit film rather than only a co-fabricated witness film.","fun_headline_variants_meta":{"raw":{"variants":["New culprit in qubit decoherence: niobium hydride specks","Hydrides in niobium films drain qubit coherence","Niobium hydrides: hidden defect inside superconducting qubits","Qubit performance drop traced to niobium hydride formation","Cryogenic hydrides: fresh source of qubit energy loss"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00063,"raw_usage":{"total_tokens":2771,"prompt_tokens":788,"completion_tokens":1983,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":532,"completion_tokens_details":{"reasoning_tokens":1910}},"tokens_in":532,"tokens_out":1983,"duration_ms":13719,"temperature":1.0,"reasoning_tokens":1910,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T20:13:11.494057+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Cryogenic transmission electron microscopy or scanning nano-X-ray diffraction on an actual qubit chip cooled below 150 K should resolve multiple NbH0.89 lattice reflections and precipitates whose number density scales with locally measured hydrogen content and inversely with measured T1. If the only hydride fingerprint remains the single 2.45 Å reflection, or if no precipitates are found inside the device film, the structural identification is not supported.","supporting_citations":[{"cited_title":"Barkov, A","cited_arxiv_id":null,"evidence_quote":"Showed hydride precipitation in bulk niobium after different treatments, establishing the morphology and cooldown-cycle evolution that the film data are compared with."},{"cited_title":"Barkov, A","cited_arxiv_id":null,"evidence_quote":"Earlier direct observation of hydride formation in cavity-grade niobium, the methodological precedent for the diffraction and microscopy identification."},{"cited_title":"Romanenko, F","cited_arxiv_id":null,"evidence_quote":"Proximity-breakdown model attributing RF loss and high-field Q slope to hydride inclusions, used to explain qubit dissipation."},{"cited_title":"Schober and H","cited_arxiv_id":null,"evidence_quote":"Reference for Nb-H phase structures and thermodynamics, underpinning the assignment of hydrogen site occupancy and hydride phases."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Showed hydride formation and microwave loss in niobium thin films after fluoride chemical processing, the nearest thin-film analogue supporting film relevance."},{"cited_title":"Isagawa, Hydrogen absorption and its effect on low‐temperature electric properties of niobium, J","cited_arxiv_id":null,"evidence_quote":"Establishes hydrogen absorption and the low-temperature solubility threshold near 150 K that explains when hydrides appear."},{"cited_title":"Q Disease","cited_arxiv_id":null,"evidence_quote":"Documents hydrogen-related Q degradation in niobium cavities, connecting dissolved hydrogen to macroscopic RF loss."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Direct nanometer-scale hydride observation in superconducting niobium, used to interpret the cryo-AFM surface features as hydrides."},{"cited_title":"Nersisyan et al., Manufacturing Low Dissipation Superconducting Quantum Processors, in 2019 IEEE International Electron Devices Meeting (IEDM) (2019), p","cited_arxiv_id":null,"evidence_quote":"Describes the fabrication process of the tested qubit device, providing the deposition and processing context for hydrogen incorporation."}],"review_version":1}