{"id":"a617fa50-3641-441e-922f-cd0633552a02","arxiv_id":"2607.00809","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Vanadium resonators show surface-dominated non-TLS losses mitigated by Ta capping, with Nb-buffered films unexpectedly exhibiting higher losses despite better crystallinity.","lead":"This paper grows four vanadium film structures on silicon wafers using Nb buffers and Ta caps, fabricates coplanar waveguide resonators, and measures internal quality factor versus photon number. The goal is to link film structure to microwave losses for potential use in quantum circuits.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Attribution of elevated non-TLS loss in Nb-buffered V films to oxides/hydrides rests on suggestive but unquantified EDX/XPS data without exclusion of confounders","rationale":"The reader's weakest_assumption identifies exactly this inference gap. Because the work is purely experimental and the full text supplies no additional quantitative controls or falsification tests beyond the abstract-level description, the concern remains load-bearing and the UNVERDICTED verdict is unaffected.","tokens_in":1920,"tokens_out":348,"duration_ms":26874,"concrete_test":"Re-measure the four resonator types with XPS depth profiling and quantitative EDX on the same devices used for microwave testing; fit non-TLS loss tangent vs integrated oxide signal across samples. If the correlation coefficient is low or if interface roughness (AFM) or Nb interdiffusion (TEM) correlates more strongly, the oxide/hydride attribution does not hold.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the observed higher non-TLS loss (and lower Q_int) in Nb/V vs direct V/Si structures, despite superior lattice uniformity, is caused by V oxides or hydrides at surfaces/grain boundaries. The paper supports this only by noting that EDX and XPS detect such species; it provides no quantitative correlation between oxide/hydride areal density or depth profile and the extracted non-TLS loss rate, nor does it report controls that isolate this from Nb/V interface scattering, post-etch contamination, or differences in resonator geometry/film stress. The Ta-capping mitigation claim is less affected because it rests on direct structure comparisons, but the Nb-buffer contrast is the load-bearing inference.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript reports fabrication and microwave characterization of coplanar waveguide resonators from four vanadium film structures (Nb/V/Ta, Nb/V, V/Ta, V) sputtered on Si(001). XRD and AFM show Nb-buffered films have superior lattice-orientation uniformity and smaller grains. Q_int versus <n_ph> measurements indicate that V-surface losses are dominated by <n_ph>-independent non-TLS mechanisms that are reduced by Ta capping; Nb-buffered films nevertheless show lower Q_int (10^0 to 10^6 photons) and higher non-TLS loss than direct V/Si films, with EDX/XPS suggesting V oxides or hydrides at surfaces and grain boundaries as a possible origin.","tokens_in":2098,"tokens_out":569,"duration_ms":19462,"significance":"If the non-TLS dominance and Ta-mitigation claims are confirmed with quantitative controls, the work would supply concrete guidance on surface-chemistry engineering for low-loss vanadium resonators in quantum circuits, separating structural uniformity from chemical-loss contributions.","major_comments":[{"comment":"Abstract and results on Q_int(<n_ph>) analysis: the claim that non-TLS losses dominate and are higher in Nb-buffered films rests on post-measurement fitting whose details (functional form, error bars, number of resonators per structure, exclusion criteria for outlier devices) are not supplied, preventing independent evaluation of the extracted non-TLS loss coefficients.","section":"Abstract / Q_int vs <n_ph> analysis"},{"comment":"Discussion of Nb-buffer versus direct-growth contrast: the inference that elevated non-TLS loss originates from V oxides/hydrides (detected by EDX/XPS) lacks any quantitative correlation between oxide/hydride areal density or depth profile and the reported non-TLS loss rates; no controls isolating this from Nb/V interface scattering, film stress, or post-etch contamination are described.","section":"Discussion of origins of trends"},{"comment":"Table or figure presenting Q_int and non-TLS loss values: without tabulated values, uncertainties, or statistical tests for the four structures, the statement that Nb/V films exhibit “higher non-TLS loss” cannot be assessed for magnitude or significance.","section":"Results / loss-parameter table"}],"minor_comments":[{"comment":"Notation for average photon number is inconsistent between <n_ph> and ⟨n_ph⟩; standardize throughout.","section":null},{"comment":"EDX and XPS spectra or depth profiles should be shown with quantitative elemental percentages rather than qualitative “presence suggested.”","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the careful reading and constructive comments. We address each major point below. Where details were omitted, we will add them in revision; where inferences are suggestive rather than quantitative, we will clarify the limitations of the data.","responses":[{"response":"We agree that the fitting procedure and supporting statistics were not described in sufficient detail. In the revised manuscript we will specify the functional form used to separate TLS and non-TLS contributions, include error bars on all Q_int data points, state the number of resonators measured for each film structure, and document the criteria applied to exclude outlier devices. These additions will enable independent assessment of the extracted non-TLS coefficients.","revision_made":"yes","referee_comment":"[Abstract / Q_int vs <n_ph> analysis] Abstract and results on Q_int(<n_ph>) analysis: the claim that non-TLS losses dominate and are higher in Nb-buffered films rests on post-measurement fitting whose details (functional form, error bars, number of resonators per structure, exclusion criteria for outlier devices) are not supplied, preventing independent evaluation of the extracted non-TLS loss coefficients."},{"response":"The manuscript already qualifies the link as suggestive (“might be relevant … suggested by”). We will revise the discussion to state explicitly that no quantitative correlation between oxide/hydride density and loss rates has been established, and that other possible contributions (interface scattering, stress, contamination) have not been isolated by dedicated controls. This change will prevent over-interpretation while retaining the EDX/XPS observations as one plausible origin.","revision_made":"yes","referee_comment":"[Discussion of origins of trends] Discussion of Nb-buffer versus direct-growth contrast: the inference that elevated non-TLS loss originates from V oxides/hydrides (detected by EDX/XPS) lacks any quantitative correlation between oxide/hydride areal density or depth profile and the reported non-TLS loss rates; no controls isolating this from Nb/V interface scattering, film stress, or post-etch contamination are described."},{"response":"We will add a table (or expanded figure caption) that reports representative Q_int values at selected photon numbers, the fitted non-TLS loss coefficients with uncertainties, and basic statistical information (mean and standard deviation across devices) for each of the four film structures. This will allow direct evaluation of the magnitude and significance of the reported differences.","revision_made":"yes","referee_comment":"[Results / loss-parameter table] Table or figure presenting Q_int and non-TLS loss values: without tabulated values, uncertainties, or statistical tests for the four structures, the statement that Nb/V films exhibit “higher non-TLS loss” cannot be assessed for magnitude or significance."}],"tokens_in":1589,"tokens_out":637,"duration_ms":24849,"standing_objections":["We do not possess additional EDX/XPS data that would allow quantitative areal-density or depth-profile correlations with the measured non-TLS loss rates; the existing spectra were acquired for qualitative identification only."]},"desk_editor":{"model":"grok-4.3","letter":"The paper reports that vanadium films on silicon can be turned into microwave resonators whose internal Q improves with a tantalum cap but drops when a niobium buffer is added, even though the buffer improves lattice orientation and reduces grain size. The authors conclude that non-TLS surface losses dominate and that the buffer penalty likely comes from vanadium oxides or hydrides seen in EDX and XPS.\n\nThey grew four stack combinations, characterized them with XRD and AFM, fabricated CPW resonators, and measured Q_int across six orders of photon number. The separation into TLS and non-TLS components and the clear improvement from the Ta cap are the usable results. Those measurements give a first data point on vanadium as a possible circuit material.\n\nThe soft spot is the causal claim for the Nb buffer. The structural benefit is shown directly, but the higher non-TLS loss is linked to oxides only by their detection in spectra; no areal densities, depth profiles, or quantitative correlation with the extracted loss rates are given. Interface scattering, film stress, or etch differences are not excluded. The abstract itself uses “might be,” which matches the evidence level.\n\nGroups screening new superconductors for quantum circuits will want the Q versus photon-number curves and the stack comparisons. The work is straightforward enough to deserve referee time, though any review should ask for tighter surface analysis before the mechanism story is treated as settled.","headline":"Vanadium resonators with Ta cap cut non-TLS loss while Nb buffer raises it despite better crystallinity, but the oxide explanation stays qualitative.","tokens_in":2609,"tokens_out":350,"would_cite":false,"duration_ms":23036,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Losses in vanadium superconducting resonators are dominated by non-TLS mechanisms at the surface that a tantalum capping layer can reduce.","keywords":["vanadium","superconducting resonators","microwave losses","non-TLS losses","tantalum capping","niobium buffer","silicon substrates","quantum devices"],"falsifier":"Measuring resonators from V films where EDX and XPS show no oxides or hydrides but still finding the same differences in Q_int and non-TLS loss between Nb-buffered and direct-growth samples would falsify the proposed origin.","tokens_in":2834,"feed_emoji":"","tokens_out":683,"duration_ms":27207,"temperature":0.7,"pith_summary":"The authors fabricate coplanar waveguide resonators from vanadium films grown on silicon in four configurations using niobium buffers and tantalum caps. They measure internal quality factors as a function of average photon number and separate the contributions from two-level systems and other losses. The results show that non-TLS losses at the vanadium surface limit performance and are reduced by the tantalum cap. Films grown on niobium buffers show worse performance than those grown directly on silicon despite better crystal orientation, which the authors link to oxides or hydrides detected by spectroscopy.","feed_headline":"Tantalum cap reduces non-TLS losses in vanadium resonators","feed_subtitle":"Niobium buffers increase losses despite better crystal uniformity, implicating surface oxides or hydrides.","key_machinery":"The photon-number dependence of internal quality factor Q_int in resonators made from four different V-film stacks (Nb/V/Ta, Nb/V, V/Ta, V), used to separate TLS and non-TLS loss contributions.","core_discovery":"Loss at the V surface is dominated by ⟨n_ph⟩-independent non-two-level-system (non-TLS) losses, which can be mitigated by introducing the Ta capping layer. The V films on the Nb buffer layers exhibit lower Q_int and higher non-TLS loss than that directly grown on Si wafers, even though the former has higher lattice-orientation uniformity. Origins of these trends might be relevant to V oxides and/or V hydrides at surfaces and grain boundaries.","pith_inferences":["Vanadium could serve as an alternative material for quantum circuits if surface losses are controlled.","Similar capping strategies might apply to other superconducting films prone to oxide formation.","Direct growth without buffers may be preferable for V on Si to minimize additional loss channels.","Further studies could test hydride formation by varying processing conditions."],"forward_implications":["Introducing a Ta capping layer mitigates non-TLS losses in V resonators.","Nb buffer layers increase non-TLS loss in V films compared to direct growth on Si.","Better lattice orientation uniformity does not guarantee lower losses in these V films.","V oxides or hydrides at surfaces and grain boundaries are likely sources of the non-TLS losses.","Mitigation strategies should target surface chemistry rather than just improving crystallinity."],"fun_headline_variants":["Ta cap cuts non-TLS losses in V resonators","Nb buffer raises non-TLS loss in uniform V films","V oxides hydrides drive non-TLS resonator losses","Tantalum reduces surface losses in vanadium films"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The performance differences between film stacks are caused by V oxides or hydrides at surfaces and grain boundaries rather than by differences in interface quality or measurement artifacts.","fun_headline_variants_meta":{"raw":{"variants":["Ta cap cuts non-TLS losses in V resonators","Nb buffer raises non-TLS loss in uniform V films","V oxides hydrides drive non-TLS resonator losses","Tantalum reduces surface losses in vanadium films"]},"model":"grok-4.3","cost_usd":0.003648,"raw_usage":{"total_tokens":1888,"prompt_tokens":804,"num_sources_used":0,"completion_tokens":61,"cost_in_usd_ticks":36478000,"prompt_tokens_details":{"text_tokens":804,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1023,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":804,"tokens_out":61,"duration_ms":11747,"temperature":1.0,"reasoning_tokens":1023,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-02T10:02:16.249048+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Measuring resonators from V films where EDX and XPS show no oxides or hydrides but still finding the same differences in Q_int and non-TLS loss between Nb-buffered and direct-growth samples would falsify the proposed origin.","supporting_citations":[],"review_version":1}