{"id":"85aa0545-9bbc-436f-bd1c-b73f0cc65964","arxiv_id":"1908.07146","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"ALD-deposited niobium nitride films produce microwave resonators with internal quality factors above 10^6 for a substantial fraction of devices, with critical temperatures up to 10.9 K.","lead":"The paper tests niobium nitride films made with atomic layer deposition, a precise coating method, for use in superconducting microwave resonators. The films reach quality factors above one million for many resonators, suggesting ALD niobium nitride could be a useful alternative to sputtered films for detector arrays.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"High-Q yield claim depends on unstated sample sizes, a Qc selection window, and 10x same-wafer Qi variation; need full per-resonator data to verify.","rationale":"The reader's CONDITIONAL verdict is reasonable: direct S21 measurements provide independent support for the existence of high-Q ALD NbN resonators, but the interpretation as a reproducible yield is undercut by unreported sample sizes, a factor-of-10 same-wafer variation, and the Qc selection window. The reader's formal weakest assumption (process-induced film damage) is less directly load-bearing: Qi is measured on fully processed resonators, so if the lithography/etch/clean degraded the film, the reported Qi would be lower than intrinsic; that would weaken material comparisons but not the observation of high-Q devices. The more serious risk is that the high-Q fraction is an artifact of which resonators were fit and which devices were selected. Testing this requires the full per-resonator data table and a robustness check over the Qc acceptance criterion and per-chip variation. Agreement is partial because the reader's rationale mentions the factor-of-10 variation and small statistics, although the stated weakest assumption focuses on process damage.","tokens_in":5424,"tokens_out":8307,"duration_ms":92309,"concrete_test":"Obtain the complete resonator log from the authors: for every measured resonator, report f0, Qi, Qc, readout power, stage temperature, and wafer/chip position, including all resonators rejected by the Qc ∈ [10^4, 10^6] window. Recompute the fraction with Qi > 10^6 under three treatments: (i) all resonators, (ii) only Qc-in-window resonators, and (iii) a bootstrap over chips and wafer positions. If the fraction is not stable at roughly ≥40% in all treatments, the \"≥43%\" yield claim is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing step is the yield statistic in the abstract: \"≥43% of resonators above Qi = 10^6.\" Section 3.3 and Fig. 5 support this with no per-wafer or per-condition resonator counts, no confidence intervals, and a selection criterion described only in the caption: \"These Qi values were obtained with a formal fit for resonators with Qc values between 10^4 and 10^6.\" If resonators with Qc outside this range were excluded, the reported fraction is a property of the coupling design and fitting acceptance window, not of ALD NbN. The same paragraph states \"we measured different devices from the same 4” wafer and saw average quality factors differ by a factor of 10.\" A factor-of-10 wafer-level spread means the fraction above 10^6 can change dramatically depending on which chip or wafer position is measured. With no N reported, the headline fraction cannot be distinguished from positional or fitting selection. The claim that ALD NbN is a viable sputtered-NbN alternative requires the yield to be reproducible, so this missing statistical backbone is the weakest point.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports atomic layer deposition of niobium nitride (ALD NbN) films and their DC and microwave characterization for use in high-Q superconducting resonators. Films of roughly 4-30 nm thickness were grown at 250 and 300 °C; the authors report growth rates, XRD texture, AFM roughness, DC critical temperatures (up to 10.87 K), and resistivity. Single-layer lumped-element resonators were fabricated and measured at dilution temperatures, yielding internal quality factors Qi above 10^5 for nearly all resonators, with 75% above 10^6 for a 250 °C film and 43% above 10^6 for a 300 °C film. Additional fits of Mattis-Bardeen relations yield kinetic inductance fractions and RF critical temperatures, and comparison with aluminum reference devices gives Lk = 1.7-5.4 pH/square. The paper concludes that ALD NbN is a viable alternative to sputtered NbN for microwave resonators.","tokens_in":5668,"tokens_out":6142,"duration_ms":62844,"significance":"If the yield claim is reproducible, this is a useful advance for the detector community: ALD provides wafer-scale thickness control in the ultrathin regime where sputtered NbN uniformity is a known concern, and the reported quality factors are competitive with some sputtered NbN devices. The paper's strengths include the multi-technique characterization (ellipsometry, XRD, AFM, DC transport, resonator S21 fits), the explicit reporting of wafer-to-wafer and device-to-device variation, and the use of an external aluminum reference for the kinetic inductance extraction. However, the headline statistic that 43-75% of resonators exceed Qi = 10^6 is not yet supported by the data as presented, because no resonator counts, confidence intervals, or acceptance criteria are given. The paper also does not establish that the full lithography and etching process preserves the as-deposited film properties. These issues must be addressed before the quantitative claims can be accepted.","major_comments":[{"comment":"The central yield claim in the abstract ('with ≥43% of resonators above Qi = 10^6') is not supported by the reported data. Section 3.3 reports 75% above 10^6 for a 250C film and 43% for a 300C film, but gives no number of resonators measured, no per-chip or per-wafer breakdown, and no confidence intervals. The Fig. 5 caption states that the Qi values were 'obtained with a formal fit for resonators with Qc values between 10^4 and 10^6,' which means resonators outside this coupling window were excluded; this selection must be justified and its effect on the yield quantified. The same paragraph notes that average quality factors from the same 4-inch wafer differ by a factor of 10, so without the full per-resonator data the reported fraction cannot be distinguished from position-dependent or fitting-selection effects. Please provide per-resonator data (or a table/histogram with N) for each deposition condition and a sensitivity analysis of the yield to the Qc acceptance window.","section":"§3.3, Fig. 5"},{"comment":"The paper does not reconcile the RF-derived critical temperatures with the DC measurements. Section 3.3 reports T_c^(RF) ~ 11K for 250C films and ~12K for 300C films, while Section 3.2 reports a maximum DC Tc of 10.87K, with 12% sample-to-sample variation. Because T_c^(RF) is obtained by fitting Mattis-Bardeen expressions for the gap Δ0 and then converting with the BCS relation Δ0 = 1.76 k Tc, fit degeneracies or the local-limit assumption could bias the converted temperature. Please report the uncertainties in the fitted Δ0 and αk, show the quality of the fits, and quantify whether the apparent 1-2 K excess over DC Tc is statistically significant. If the discrepancy is real, it should be discussed; if it is a fitting artifact, the T_c^(RF) values should be presented only with appropriate caveats.","section":"§3.2 and §3.3"},{"comment":"The DC Tc and resistivity are measured on unpatterned as-deposited films, while Qi is measured on patterned resonators that have undergone photoresist coating, fluorine-based ICP etching, oxygen ash, and heated NMP cleaning. The manuscript does not test whether these steps degrade the superconducting or microwave properties of the film. If the fabrication process damages the film (e.g., by etching or oxidizing the surface), the reported Qi values would not represent the intrinsic quality of ALD NbN. Please add a control measurement, such as DC Tc and sheet resistance measured before and after the full patterning process, or a comparison of patterned versus unpatterned films from the same deposition.","section":"§2.2 vs §3.2"},{"comment":"Equation (3) is typeset as a single relation but is meant to contain two distinct equations for δf0/f0 and δ(1/Qi). As printed, the expression is ungrammatical and the second relation appears to have σ2 in the numerator, whereas the standard Mattis-Bardeen result for δ(1/Qi) has the conductivity change divided by σ2 (or otherwise normalized). Because this equation is the basis for the αk and Δ0 fits that produce T_c^(RF), the fitting procedure cannot be independently reproduced as written. Please correct the equation, define every symbol (including the meaning of 'd' versus 'δ'), and provide the explicit functional forms used in the fit.","section":"Eq. (3)"}],"minor_comments":[{"comment":"The abstract states only '≥43% of resonators above Qi = 10^6' while Section 3.3 reports 75% for the 250C film and 43% for the 300C film; please state both numbers or explicitly say that 43% is the lower bound obtained from the less favorable deposition condition.","section":"Abstract and §3.3"},{"comment":"The XRD peak labels in Fig. 2 are '111' and '200', but the text refers to the ⟨100⟩ orientation; please reconcile the crystallographic notation (e.g., by explaining that the 200 reflection corresponds to the ⟨100⟩ texture).","section":"§3.1, Fig. 2"},{"comment":"Growth rates are written as '.51 ± .05 Å/cycle' and '.62 ± .05 Å/cycle' without leading zeros; please use '0.51' and '0.62' for consistency with the rest of the text.","section":"§3.1"},{"comment":"The right panel of Fig. 5 is labeled 'Resonator d f/f vs T behavior' and is said to correspond to a 75-cycle film, but the caption does not identify which resonator or measurement power is shown; please add a legend or the missing axis details.","section":"Fig. 5"},{"comment":"The comparison with sputtered NbN resonators (refs. 12 and 13) is qualitative and does not state the film thickness, measurement temperature, or readout power for the compared devices; a short table with these conditions would make the comparison meaningful.","section":"§4"}],"recommendation":"major_revision","confidential_remarks":"The paper is an interesting initial demonstration that ALD NbN can support high-Q microwave resonators, but the headline yield statistic needs to be backed by full per-resonator data and a clear statement of the selection criteria. If those data already exist, a revision can address this; otherwise the abstract and conclusions should be weakened to a claim about individual high-Q devices. I would not reject the paper, because the underlying measurements appear genuine and the material is of community interest, but the current form cannot support the quantitative yield claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a first report on ALD-grown NbN films in microwave resonator geometries, and that alone makes it worth a look if you work on superconducting detectors. The fabrication is described in enough detail to reproduce, the RF fitting uses standard tools, and the authors are refreshingly open about the variations they saw, including the factor-of-10 device-to-device Q spread and the 12% Tc nonuniformity.\n\nThe genuine new result is that PEALD NbN can produce resonators with internal quality factors above 10^6 in a non-trivial fraction of devices. That is useful evidence for the MKID community, where sputtered NbN is the usual default and uniformity at small thickness is a known problem. The paper also reports kinetic inductance values around 1.7-5.4 pH per square, which is in the right range for detector applications.\n\nWhere I would push back is on the headline yield claim: “≥43% of resonators above Qi = 10^6.” In the body this comes with a Qc selection window (10^4 to 10^6) that only appears in the figure caption, and there are no reported per-wafer or per-condition resonator counts. The same paragraph says devices from the same 4-inch wafer can differ in average Q by a factor of ten. With that spread and no N, the 43% figure could shift a lot depending on which chip you happen to mount. That is the weakest link in the paper, and it needs a table of per-resonator values or at least a per-chip breakdown.\n\nTwo other soft spots are minor but worth noting. The RF-derived Tc from Mattis-Bardeen fits comes out around 11-12 K, higher than the maximum DC Tc of 10.87 K they measured; such fits use the BCS gap relation, so this is a model assumption, not a direct measurement, and should be presented with that caveat more prominently. Also, they measure DC Tc on unpatterned films but never check whether the full lithography, plasma etch, and resist strip degrade the superconducting properties of the patterned film. That control would strengthen the claim that the measured Q represents intrinsic ALD NbN quality.\n\nNone of these issues kill the paper. The central observation, that ALD NbN can support high-Q resonators, looks credible on the data shown. The paper is a solid initial characterization, not a finished technology demo. I would send it to a serious referee with instructions to ask for the per-resonator statistics, the sample sizes, and ideally a patterned-film Tc check. It deserves referee time.","headline":"First RF characterization of ALD NbN resonators, with credible high-Q values but a headline yield statistic that needs more statistical backing before it can carry the paper's weight.","tokens_in":6189,"tokens_out":2278,"would_cite":true,"duration_ms":26457,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Atomic-layer-deposited niobium nitride can make microwave resonators with internal quality factors above one million.","keywords":["atomic layer deposition","niobium nitride","microwave resonators","internal quality factor","kinetic inductance","superconducting detectors","plasma-enhanced ALD","thin films"],"falsifier":"Take a single ALD NbN wafer, measure its DC $T_c$ and resistivity on an unpatterned piece, pattern identical resonators on the rest, and compare the RF-derived $T_c$ and $Q_i$ after the full etch and clean process: a processing-induced drop in $T_c$, a broadened transition, or a $Q_i$ below $10^6$ would indicate that the reported quality factors do not measure the intrinsic ALD NbN film.","tokens_in":5212,"feed_emoji":"📡","tokens_out":10793,"duration_ms":95143,"temperature":0.7,"pith_summary":"This paper sets out to show that atomic layer deposition (ALD) can produce niobium nitride films good enough for high-quality-factor superconducting microwave resonators, a role currently dominated by sputtered NbN. The authors fabricate single-layer lumped-element resonators from plasma-enhanced ALD NbN films between 4 and 30 nm thick and find internal quality factors $Q_i$ above $10^6$ for 75% of resonators from a 250 °C deposition and for 43% from a 300 °C deposition, with nearly all devices above $10^5$. They also report critical temperatures from 7.5 K to 10.9 K that grow with film thickness, and kinetic inductance values around 1.7–5.4 pH per square. If the high $Q_i$ is not an artifact of processing, ALD NbN becomes a practical route to uniform, wafer-scale, thin-film resonators for detector arrays and kinetic-inductance devices.","feed_headline":"Atomic layer deposition makes million-Q niobium nitride resonators","feed_subtitle":"Uniform 4-30 nm films with million-Q resonators offer a sputtered-NbN alternative for detector arrays.","key_machinery":"The mechanism that carries the argument is plasma-enhanced atomic layer deposition using the niobium precursor TBTDEN and an argon plasma, in which one ALD cycle grows roughly one atomic monolayer. This gives measured growth rates of $0.51 ± 0.05$ Å/cycle at 250 °C and $0.62 ± 0.05$ Å/cycle at 300 °C, and it is this atomic-scale thickness control that provides the wafer-scale uniformity claimed for 4–30 nm films. The deposited film is patterned into a planar lumped-element resonator with an interdigitated capacitor and inductor, and its response is fitted to a complex transmission model; Mattis-Bardeen relations convert the temperature dependence of frequency shift and internal loss into the kinetic inductance fraction, gap energy, and RF critical temperature.","core_discovery":"The central claim is that plasma-enhanced atomic layer deposition of NbN, despite not matching the highest $T_c$ of sputtered films, yields microwave resonators whose internal quality factors are comparable to or better than sputtered NbN. At a stage temperature of 300 mK, 75% of resonators made from films deposited at 250 °C and 43% of those deposited at 300 °C have $Q_i > 10^6$, and almost all measured resonators exceed $10^5$; the highest DC critical temperature observed is 10.87 K, with RF-derived $T_c$ around 11–12 K. The paper also extracts kinetic inductance $L_k = 5.4 ± 2$ pH/square for a 300-cycle 250 °C film and $1.7 ± 0.5$ pH/square for a 300-cycle 300 °C film. These numbers establish ALD NbN as a candidate material for high-$Q$ resonator circuits, with the caveat that films thinner than about 75 cycles were not observed to superconduct.","pith_inferences":["The paper does not isolate where the factor-of-10 $Q_i$ spread comes from; a natural inference is that local thickness or interface variation, not intrinsic film loss, sets the ceiling, which could be tested by measuring the same resonator after successive processing steps.","The transition-width data connect thinner films to broader transitions and lower $T_c$; the implied granularity limit near 4 nm suggests that grain structure, visible by microscopy, may be the same quantity that caps $Q_i$ in the ultrathin regime.","The authors' planned variation of substrate and surface preparation implies a testable hypothesis: the same ALD films on sapphire or (111) Si should show higher $T_c$ and possibly higher $Q_i$ if lattice mismatch and interface contamination are the limiting losses.","The paper lists kinetic-inductance devices as motivation but builds only resonators; a logical next step is an ALD NbN parametric amplifier or nanowire detector, whose noise and saturation behavior would extend the quality-factor result into a full device demonstration."],"forward_implications":["Detector readout can be multiplexed more aggressively, since a larger fraction of resonators in a given band will sit above $Q_i = 10^6$.","Wafer-scale thickness uniformity down to a few nanometers makes ALD NbN attractive for on-chip mm-wave band-defining features and nanowire detectors where sputtered thickness control is a concern.","Deposition temperature becomes a design lever: 250 °C films show higher quality factors, while 300 °C films grow faster and show lower kinetic inductance.","Even though $T_c$ maxes near 10.9 K, the measured $Q_i$ already rivals sputtered NbN CPW resonators, so the lower transition temperature does not by itself block detector use."],"supporting_citations":[{"why":"Supplies the plasma-enhanced ALD growth-rate baseline for TBTDEN niobium nitride that the paper's measured growth rates are consistent with.","marker":"[7]"},{"why":"Provides the earlier ALD NbN study at higher deposition temperature used as a comparison for growth rate and transition-width trends.","marker":"[8]"},{"why":"Provides the Mattis-Bardeen fitting relations used to extract kinetic inductance fraction and RF critical temperature from resonator data.","marker":"[10]"},{"why":"Supplies the aluminum-resonator kinetic inductance fraction estimate used to solve for geometric inductance and capacitance and extract $L_k$.","marker":"[11]"},{"why":"Gives the sputtered NbN CPW resonator quality factors that the ALD results are compared against.","marker":"[12]"},{"why":"Gives the spiral-design NbN MKID quality factors that the ALD results are reported to exceed.","marker":"[13]"}],"fun_headline_variants":["ALD NbN hits million-Q resonator quality","Atomic-scale NbN films achieve high-Q resonators","Niobium nitride via ALD matches sputtered resonator Q","Million-Q resonators from atomic layer deposition","Precision-deposited NbN yields high-Q microwave resonators"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the lithography and etch sequence—photoresist, fluorine plasma etch, oxygen ash, and hot solvent cleaning—does not degrade the film, because $T_c$ is measured on unpatterned films while $Q_i$ is measured on patterned resonators.","fun_headline_variants_meta":{"raw":{"variants":["ALD NbN hits million-Q resonator quality","Atomic-scale NbN films achieve high-Q resonators","Niobium nitride via ALD matches sputtered resonator Q","Million-Q resonators from atomic layer deposition","Precision-deposited NbN yields high-Q microwave resonators"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000847,"raw_usage":{"total_tokens":3696,"prompt_tokens":965,"completion_tokens":2731,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":2654}},"tokens_in":581,"tokens_out":2731,"duration_ms":18145,"temperature":1.0,"reasoning_tokens":2654,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:23:56.155224+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a single ALD NbN wafer, measure its DC $T_c$ and resistivity on an unpatterned piece, pattern identical resonators on the rest, and compare the RF-derived $T_c$ and $Q_i$ after the full etch and clean process: a processing-induced drop in $T_c$, a broadened transition, or a $Q_i$ below $10^6$ would indicate that the reported quality factors do not measure the intrinsic ALD NbN film.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the plasma-enhanced ALD growth-rate baseline for TBTDEN niobium nitride that the paper's measured growth rates are consistent with."},{"cited_title":"Linzen et al., Superconductor Science and Technology 30, 035010 (2017)","cited_arxiv_id":null,"evidence_quote":"Provides the earlier ALD NbN study at higher deposition temperature used as a comparison for growth rate and transition-width trends."},{"cited_title":"Gao et al., Journal of Low Temperature Physics 151, 557 (2008)","cited_arxiv_id":null,"evidence_quote":"Provides the Mattis-Bardeen fitting relations used to extract kinetic inductance fraction and RF critical temperature from resonator data."},{"cited_title":"Gao, Ph.D","cited_arxiv_id":null,"evidence_quote":"Supplies the aluminum-resonator kinetic inductance fraction estimate used to solve for geometric inductance and capacitance and extract $L_k$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the sputtered NbN CPW resonator quality factors that the ALD results are compared against."},{"cited_title":"Ariyoshi et al., Applied Physics Express 6, 064103 (2013)","cited_arxiv_id":null,"evidence_quote":"Gives the spiral-design NbN MKID quality factors that the ALD results are reported to exceed."}],"review_version":1}