{"id":"065d22ea-a261-4ec6-bee9-c486929a8f1c","arxiv_id":"1908.08689","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An Ar-H plasma applied before aluminum-oxide growth raises peak mobility in shallow InAs quantum wells to 45,300 cm2/Vs and lowers sample-to-sample variance compared with untreated or TMA-treated samples.","lead":"This paper shows that treating the surface of shallow InAs quantum wells with an argon-hydrogen plasma before adding an insulating oxide raises electron mobility, up to 45,300 cm2/Vs, and reduces variability across samples. It matters because Majorana qubit devices need shallow, high-mobility InAs layers, and removing the superconductor normally damages the surface.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Observed mobility gain is credible, but the claim that Ar-H plasma 'repairs' the surface is underdetermined by transport data alone because the density shift in Fig. 5 can explain the mobility gain without invoking defect passivation.","rationale":"The reader correctly identified the weakest assumption as the inferred interface-passivation mechanism without direct interface characterization. My stress-test pass agrees that this is the most load-bearing concern, and sharpens it by noting that the Fig. 5 density-mobility correlation provides a concrete alternative explanation: the plasma-treated samples have lower zero-gate densities, and if peak mobility is reached at lower densities (where screening and scattering balance differently), part of the gain could be a density effect rather than a defect-density effect. This is not an objection to the measured mobility values, which appear to be competent Hall measurements with good quantum Hall quantization for sample B, but it is an objection to the mechanism-level claim that motivates the title. The variance-reduction claim also rests on only two chips per treatment, with no error bars, no inter-chip statistics, and a wafer position ('near' vs 'far') that alone explains a large mobility spread. These are exactly the conditions on which the reader's conditional verdict is based, so I agree with the verdict and the weakest-assumption identification.","tokens_in":10626,"tokens_out":1576,"duration_ms":15627,"concrete_test":"Re-plot each near-center mobility-versus-density trace as mobility versus n in the same figure, and compute the peak mobility of each trace at the same density, say n = 1.0 x 10^12 cm^-2, rather than at each sample's own peak density. If the Ar-H plasma sample still exceeds the untreated control by more than roughly 20% at matched density, the passivation interpretation survives this specific check; if the advantage shrinks or reverses at matched density, the headline claim should be weakened to 'plasma treatment shifts the density at which peak mobility occurs.' Additionally, running one XPS or TEM cross-section comparison of the InAs/oxide interface for an Ar-H plasma-treated sample versus an untreated sample would directly test the proposed interface repair mechanism.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central mechanistic claim in Section III and IV is that Ar-H plasma passivates charged interface states, thereby increasing mobility. But the paper's own Fig. 5 shows an inverse correlation between zero-gate density and peak mobility, spanning roughly 0.75 to 1.6 x 10^12 cm^-2. Because the plasma samples sit at the low end of that density range, their higher peak mobility could be primarily a consequence of lower carrier density (hence weaker remote-charge scattering at the measured peak location), rather than of a reduced areal defect density at the InAs/Al2O3 interface. The paper does not characterize the interface spectroscopically (no XPS, AES, TEM) and does not extract a quantitative scattering model, so the two hypotheses are not separated. A second, more subtle confound is that the mobility is extracted at fixed B = 0.05 T, and all samples have second-subband occupation at V_TG = 0; unless the mobility extraction is confirmed to be in the low-field Drude regime for every density trace, apparent peak shifts could partly reflect magnetoresistance or parallel-conduction artifacts. These concerns do not undercut the empirical result that the treated samples reached 45,300 cm^2/(V s), but they do undercut the 'repair/passivation' interpretation and the claim of reduced variance, which rests on just two chips per treatment with no error bars or statistical test.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a process improvement for shallow InAs quantum wells intended for topological superconductor heterostructures. After selective wet etching of an epitaxial Al layer, the authors compare three surface preparations (no pretreatment, TMA reduction, and ArH plasma) before ALD growth of Al2O3, and characterize Hall mobility versus top-gate density at 7 mK using a cryo-CMOS multiplexer for multi-sample, multi-cooldown measurement. The central empirical claim is that ArH plasma treatment followed by TMA/H2O ALD yields the highest peak mobility, up to 45,300 cm^2/(V s), and reduces chip-to-chip variance relative to untreated and TMA-treated samples. The proposed mechanism is that the plasma removes the native oxide and passivates charged impurity states at the InAs/Al2O3 interface, thereby reducing remote charged-impurity scattering. The paper also reports that all samples occupy the second subband at zero gate voltage, which is relevant to Majorana device design.","tokens_in":11012,"tokens_out":6464,"duration_ms":72428,"significance":"If the reported mobility improvement is robust, the ArH plasma step is a simple and valuable addition to the fabrication of shallow InAs 2DEGs for Majorana devices, where a clean surface and high mobility are both required. The use of a cryo-CMOS multiplexer to collect many transport traces in a single cooldown is a useful methodological strength. However, the mechanistic claim of surface 'repair' by defect passivation is inferred from transport data and a correlation plot rather than established by direct interface characterization or a quantitative scattering model, and the variance-reduction claim rests on only two chips per treatment condition. These limitations do not undermine the empirical peak-mobility observation but do require tempering or additional support before the central claim can be accepted as written.","major_comments":[{"comment":"The claim of a 'significant reduction in variance' is not supported by the data as presented. Each treatment/oxidizer pair is represented by only two chips, and despite multiple cooldowns and measurement points the figure shows no error bars, no distribution of individual measurements, and no statistical test. With n=2 per condition, an apparent variance difference could easily be dominated by wafer-position effects or by one outlier chip. Because the abstract explicitly claims a reduction in variance, this is a load-bearing point. Please report the full set of peak-mobility measurements as points or box plots, provide a statistical measure appropriate to the sample size, or explicitly state the sample size and avoid statistically loaded language such as 'significant reduction in variance' unless a test is performed.","section":"Section III, Fig. 4"},{"comment":"The inference that the higher mobility of ArH-treated samples is caused by passivation of charged interface impurities is underdetermined by the data shown. The inverse correlation between the density at VTG=0 and the peak mobility is consistent with fewer surface donors, but it is also consistent with other sample-to-sample differences such as residual etch damage, bulk disorder, or a different density of oxide charges unrelated to hydrogen passivation. The manuscript does not present a quantitative scattering model that decomposes the mobility-versus-density curves into remote-charge, background-impurity, and surface-roughness contributions, nor does it provide interface-sensitive characterization such as XPS or TEM. The correlation in Fig. 5 contains five near-center points with no error bars and no stated uncertainty. Please either add a model-based extraction of the charged-impurity density from the measured mu(n) traces, add direct interface data, or explicitly reframe the 'repair/passivation' claim as a hypothesis supported indirectly by transport evidence rather than as an established mechanism.","section":"Section IV, Fig. 5"}],"minor_comments":[{"comment":"The sentence 'Increasing top gate voltage causes the the distribution of electrons in the quantum well to shift towards the surface' contains a duplicated article ('the the'); please correct.","section":"Section II, paragraph after Fig. 1(c)"},{"comment":"The sentence about In precipitation is unclear: 'above 300°C In begins to precipitate out of the substrate due to the desorption of As' suggests the InP substrate, but the precipitation should presumably occur in the InAs layer; please rephrase to specify the material and mechanism.","section":"Section II, paragraph on ALD temperature"},{"comment":"The reaction equations assume In2O3 and As2O3 as the native oxides, but InAs native oxides are typically a more complex mixture; please label the equations as a schematic representation rather than a quantitative reaction model.","section":"Section III, Eqs. (1)-(2)"},{"comment":"The x-axis is the total density at VTG=0, which includes electrons in the second subband for all samples; since the peak mobility occurs in the single-subband regime, please clarify why this quantity, rather than the density at peak mobility, is the relevant variable for the correlation.","section":"Section IV, Fig. 5"},{"comment":"Please indicate explicitly in the caption or legend that the traces in Fig. 3 are only from near-center chips, since Fig. 4 shows that far-chip mobilities are systematically lower and the reader might otherwise infer a stronger treatment effect than is demonstrated.","section":"Fig. 3"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the empirical result is likely of interest. The main issues are statistical overreach in the variance claim and an under-supported mechanistic interpretation; both are fixable by adding data/analysis or by softening the wording. I do not see circularity or any problematic citation pattern; the self-citation to the cryo-CMOS multiplexer is appropriate for the measurement method."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper reports a useful process result: after wet-etching epitaxial Al from a shallow InAs 2DEG, an Ar-H plasma pre-treatment before ALD Al2O3 growth raises peak Hall mobility to about 45,000 cm2/Vs and appears to tighten sample-to-sample scatter. The transport data are shown honestly, with multiple measurement points and cooldowns. That empirical effect is credible.\n\nWhat is new is applying a known hydrogen-plasma cleaning step to this specific post-etch InAs surface and quantifying the gain in a 10 nm deep quantum well. For anyone fabricating Majorana devices, that is directly relevant.\n\nThe soft spots are real but not fatal. The 'reduced variance' claim rests on two chips per treatment; no error bars or statistical test, so it is preliminary. The 'repair the surface' mechanism is inferred entirely from transport. The paper sees an inverse correlation between zero-gate density and peak mobility and interprets it as fewer charged surface states, but a lower carrier density would itself move the peak mobility even at fixed defect density. Without XPS/AES/TEM or a quantitative scattering model, passivation is a plausible interpretation, not a demonstrated one. The fixed-field extraction at 0.05 T is probably acceptable since the peak mobility occurs in the single-subband regime, but checking B-dependence would remove a lingering worry about parallel conduction.\n\nThis deserves a serious referee. The process step is likely reproducible and useful, and the mechanism concern is addressable either by adding interface characterization or by softening the language to 'consistent with passivation' while keeping the empirical result.\n\nRecommendation: send to peer review. Require more samples or a caveat on the variance claim, and ideally some direct interface probe before the 'repair' language is used.","headline":"Useful process result with a credible mobility boost, but the passivation mechanism and variance reduction are not yet established.","tokens_in":11486,"tokens_out":3360,"would_cite":false,"duration_ms":34387,"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":"Argon-hydrogen plasma repairs the InAs–dielectric interface and raises shallow-2DEG mobility to 45,300 cm²/(V·s).","keywords":["InAs two-dimensional electron gas","shallow quantum well","Hall mobility","surface passivation","argon-hydrogen plasma","atomic layer deposition","Majorana zero modes","charged surface impurities"],"falsifier":"X-ray photoelectron spectroscopy (XPS) of InAs surfaces that received the ArH plasma and Al$_2$O$_3$ deposition, compared with untreated and TMA-treated surfaces, could settle it: if the plasma-treated surface does not show reduced arsenic-oxide and reduced charged-defect signal while mobility rises, the proposed repair mechanism is wrong, even if the mobility improvement is real.","tokens_in":10427,"feed_emoji":"⚛️","tokens_out":8350,"duration_ms":78587,"temperature":0.7,"pith_summary":"Shallow InAs quantum wells are promising hosts for topological (Majorana) qubits, but the wet-etch step that removes the epitaxial aluminum superconductor damages the semiconductor surface and drags down electron mobility. This paper argues that the dominant damage is charged impurity states at the InAs/dielectric interface, and that an argon-hydrogen plasma applied before atomic-layer deposition of alumina repairs that interface by passivating arsenic dangling bonds and removing the native oxide. The treatment raises peak Hall mobility to 45,300 cm$^2$/(V s) in a 10 nm deep quantum well and reduces sample-to-sample variance compared with untreated or trimethylaluminum-treated samples. If correct, the result gives device makers a simple pre-deposition step that restores mobility in exactly the shallow heterostructures needed for hard-gapped proximity superconductivity.","feed_headline":"Plasma repair lifts InAs quantum-well mobility to 45,300","feed_subtitle":"Argon-hydrogen passivation of charged surface states cuts device-to-device scatter in shallow InAs heterostructures.","key_machinery":"The load-bearing step is the ArH plasma pre-treatment: atomic hydrogen bonds to arsenic atoms at the InAs surface, saturating dangling bonds and passivating charged impurity states, and also dry-etches the native oxide to leave an abrupt semiconductor–dielectric interface before atomic-layer deposition (ALD) of Al$_2$O$_3$ using trimethylaluminum (TMA) and H$_2$O as the oxidizer. The diagnostic that carries the argument is the Hall mobility-versus-density trace under a top gate: mobility rises with density as screening improves, peaks, then falls as the electron wavefunction shifts toward the surface, and the position of that peak plus the zero-gate density identify which scatterers dominate. The inverse trend between zero-gate density and peak mobility is the paper's main evidence that reducing charged surface states is what repairs mobility.","core_discovery":"The paper's central claim is that in processed InAlAs/InAs/InGaAs heterostructures with the quantum well 10 nm below the surface, scattering off charged impurities at the semiconductor–dielectric interface is the main factor limiting mobility after the epitaxial aluminum is wet-etched away. Treating the surface with a remote argon-hydrogen plasma for 120 s before growing 10 nm of Al$_2$O$_3$ by atomic-layer deposition passivates those charged states, yielding peak Hall mobilities up to 45,300 cm$^2$/(V s) and reducing the spread of mobilities observed across cooldowns and measurement positions. The mechanism is inferred from the density dependence of mobility — mobility peaks and then falls as the gate pushes electrons toward the surface — and from an inverse correlation between zero-gate density and peak mobility, consistent with fewer charged surface donors. Ozone-grown oxide is found to hurt mobility, attributed to oxygen-rich alumina adding remote charged scatterers.","pith_inferences":["The authors do not test this, but the same plasma repair should generalize to other shallow III-V 2DEG platforms, such as InSb or InGaAs, where native-oxide charge limits mobility and hard-gap proximity superconductivity is desired.","A testable extension follows from the inverse density–mobility trend: zero-gate density, measured in a single quick Hall measurement, could be used as a process-control screen for surface passivation quality without sweeping the full gate range.","The reduced variance across cooldowns and measurement locations suggests the plasma step may also improve device-to-device reproducibility for scaled topological-qubit fabrication, though the number of samples in this study is too small to establish a yield benefit."],"forward_implications":["A 120-second argon-hydrogen plasma exposure before ALD alumina growth becomes a standard pre-deposition step for shallow InAs 2DEG devices, raising peak mobility to about 45,300 cm$^2$/(V s) in 10 nm deep wells.","Because hard-gapped proximity superconductivity requires shallow quantum wells, the plasma repair directly addresses the damage from aluminum wet-etch that previously cut mobility from about 44,000 to 1,000–2,000 cm$^2$/(V s).","Samples with the highest mobility also have the lowest electron density at zero gate voltage, so zero-gate density can serve as a fast electrical proxy for interface charge in these heterostructures.","Ozone-based ALD oxidation should be avoided for this platform, since oxygen-rich alumina introduces remote charged scatterers that shift peak mobility to higher density and lower its value."],"supporting_citations":[{"why":"Supplies the mechanism: atomic hydrogen bonds to arsenic atoms and saturates dangling bonds, passivating the surface.","marker":"[24]"},{"why":"Establishes that charged surface states in InAs/InAlAs heterostructures pin the Fermi level and set the zero-gate electron density.","marker":"[42]"},{"why":"Shows that clean InAs(100) surfaces support intrinsic electron accumulation layers, tying surface states to the measured density.","marker":"[49]"},{"why":"Provides the growth baseline for the InAs quantum well on InP substrate used in this study.","marker":"[21]"},{"why":"Gives the roughly 44,000 cm²/(V s) mobility of near-surface InAs heterostructures before the wet-etch damage.","marker":"[53]"},{"why":"Documents a hard superconducting gap in a 2D semiconductor–superconductor platform and, with [54], the post-etch mobility of 1,000–2,000 cm²/(V s).","marker":"[15]"},{"why":"Supplies the screening model used to interpret the density dependence of mobility and assign scattering mechanisms.","marker":"[33]"},{"why":"Provides the first-principles reaction model for TMA cleaning of native oxides, the alternative treatment being compared.","marker":"[23]"},{"why":"Shows hydrogen plasma etches GaAs oxide, supporting the claim that the plasma creates an abrupt interface.","marker":"[47]"}],"fun_headline_variants":["Plasma passivation boosts InAs 2DEG mobility to 45,300","Argon-hydrogen plasma heals InAs surface, mobility hits 45,300","Surface repair via plasma lifts InAs well mobility to 45,300","InAs heterostructure repair yields 45,300 mobility","Remote plasma passivation raises InAs mobility to 45,300"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes the mobility-limiting defects are charged impurities sitting at the InAs–alumina interface and that the argon-hydrogen plasma removes or neutralizes them; the paper never directly images or chemically probes that interface, so the repair mechanism is inferred from transport data and prior surface-science studies.","fun_headline_variants_meta":{"raw":{"variants":["Plasma passivation boosts InAs 2DEG mobility to 45,300","Argon-hydrogen plasma heals InAs surface, mobility hits 45,300","Surface repair via plasma lifts InAs well mobility to 45,300","InAs heterostructure repair yields 45,300 mobility","Remote plasma passivation raises InAs mobility to 45,300"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000443,"raw_usage":{"total_tokens":2217,"prompt_tokens":891,"completion_tokens":1326,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":507,"completion_tokens_details":{"reasoning_tokens":1227}},"tokens_in":507,"tokens_out":1326,"duration_ms":12697,"temperature":1.0,"reasoning_tokens":1227,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:31:46.355545+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"X-ray photoelectron spectroscopy (XPS) of InAs surfaces that received the ArH plasma and Al$_2$O$_3$ deposition, compared with untreated and TMA-treated surfaces, could settle it: if the plasma-treated surface does not show reduced arsenic-oxide and reduced charged-defect signal while mobility rises, the proposed repair mechanism is wrong, even if the mobility improvement is real.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the mechanism: atomic hydrogen bonds to arsenic atoms and saturates dangling bonds, passivating the surface."},{"cited_title":"Affentauscheggand H","cited_arxiv_id":null,"evidence_quote":"Establishes that charged surface states in InAs/InAlAs heterostructures pin the Fermi level and set the zero-gate electron density."},{"cited_title":"Noguchi, K","cited_arxiv_id":null,"evidence_quote":"Shows that clean InAs(100) surfaces support intrinsic electron accumulation layers, tying surface states to the measured density."},{"cited_title":"Kjaergaard, F","cited_arxiv_id":null,"evidence_quote":"Documents a hard superconducting gap in a 2D semiconductor–superconductor platform and, with [54], the post-etch mobility of 1,000–2,000 cm²/(V s)."},{"cited_title":"Das Sarmaand E","cited_arxiv_id":null,"evidence_quote":"Supplies the screening model used to interpret the density dependence of mobility and assign scattering mechanisms."},{"cited_title":"Klejnaand S","cited_arxiv_id":null,"evidence_quote":"Provides the first-principles reaction model for TMA cleaning of native oxides, the alternative treatment being compared."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows hydrogen plasma etches GaAs oxide, supporting the claim that the plasma creates an abrupt interface."}],"review_version":1}