{"id":"d0a008c2-b370-40e0-92f6-8e0fee0b8bc2","arxiv_id":"2506.14660","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Hall bar transistors on Si/SiGe grown in a BiCMOS pilot line show reproducible peak mobility near 4.25x10^5 cm2/Vs and percolation density near 5.9x10^10 cm-2 across six wafers.","lead":"The authors made electron transport test chips from silicon-germanium materials produced in a 200 mm chip factory pilot line and measured them at 1.5 kelvin. They find consistently high electron mobility and low disorder across multiple wafers, which supports using industrial manufacturing for many spin qubit devices.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Wafer-to-wafer reproducibility claim rests on one center device per wafer; edge outlier on wafer A shows sampling can miss degraded regions.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: one device per wafer cannot establish wafer-to-wafer reproducibility, and 12 devices from one wafer cannot fully map wafer-scale uniformity. The manuscript's own data support this concern, as the near-edge outlier in wafer A shows that sampling location matters. If the center-only sampling missed defective regions in wafers B-F, the narrow statistical distributions would not reflect the full process variability. This does not invalidate the measured transport values for the probed devices, but it prevents the reproducibility and homogeneity claims from being fully established. I find no additional internally inconsistent or formally flawed step: the magnetotransport analysis, percolation extraction, and structural characterization are plausibly executed, and the two-laboratory cross-check provides independent support. The title's 'high yield' is an unsupported assertion, but it is secondary to the sampling issue. The conditional verdict remains appropriate: accept the specific device results, but require additional spatial sampling across wafers before accepting the wafer-scale reproducibility claim.","tokens_in":9781,"tokens_out":3507,"duration_ms":39236,"concrete_test":"Measure peak mobility and percolation density for at least 12 devices from two additional wafers (e.g., B and C) using the same spatial map as wafer A, including edge positions. If near-edge devices on those wafers show similar reductions (mobility more than one standard deviation below the center average, high-density exponent near -0.6), the center-only wafer-to-wafer comparison is biased and the uniformity claim must be restricted to central regions or the reported error bars expanded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of wafer-to-wafer reproducibility and low disorder is supported primarily by six devices, one selected from the center of each wafer (Fig. 2a). The wafer-scale study of wafer A with 12 devices reveals a near-edge device with peak mobility far outside the standard deviation and a high-density mobility exponent of -0.6, indicating locally enhanced disorder. This demonstrates that device performance can degrade near the wafer edge. Because only center devices were selected for wafers B-F, the reported wafer-to-wafer average of (4.25 +/- 0.17) x 10^5 cm^2/Vs and percolation density of (5.9 +/- 0.18) x 10^10 cm^-2 may systematically exclude the same type of edge degradation, thereby overstating process uniformity. The authors acknowledge the edge outlier yet still conclude homogeneous performance across the wafer; without equivalent edge sampling on multiple wafers, the reproducibility claim is not fully established. The title's 'high yield' claim is also unsupported by yield statistics, but the sampling bias is the more load-bearing concern because it directly undermines the quantitative uniformity result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports magnetotransport characterization of Hall-bar field-effect transistors fabricated on Si/SiGe heterostructures grown in a 200 mm BiCMOS pilot line. The authors measure carrier density, Hall mobility, and percolation density at 1.5 K for one center device from each of six nominally identical wafers and for 12 additional devices across one wafer (wafer A). They report an average peak mobility of (4.25±0.17)×10^5 cm^2/Vs and a percolation density of (5.9±0.18)×10^10 cm^-2 across the six wafers, with similar wafer-scale values, and conclude that the process yields low-disorder, homogeneous heterostructures suitable for scalable spin-qubit devices. Structural characterization (STEM/EDX, XRD) is used to support layer thickness, interface sharpness, and strain claims.","tokens_in":9948,"tokens_out":5504,"duration_ms":54271,"significance":"If the reported uniformity and reproducibility hold, the paper demonstrates that an industry-standard BiCMOS pilot line can produce Si/SiGe field-effect stacks with transport properties comparable to or better than current state-of-the-art qubit-compatible stacks, which is an important step for scalable spin qubit manufacturing. Strengths of the work include the use of standard transport formulas with an externally sourced percolation exponent (p = 1.31), cross-laboratory validation of two devices, and quantitative structural analysis (interface widths from STEM, strain from XRD). The main caveat is that the central reproducibility claim rests on a small, center-selected sample, with an acknowledged edge outlier on wafer A that is not sampled on the other wafers.","major_comments":[{"comment":"The wafer-to-wafer reproducibility claim is built on one center-selected device per wafer (wafers A–F, six devices total), while the wafer-scale study of wafer A includes a near-edge device whose peak mobility and high-density exponent (alpha = -0.6) lie far outside the reported standard deviations. Because no edge or near-edge devices were measured on wafers B–F, the quoted wafer-to-wafer averages (mu = (4.25±0.17)×10^5 cm^2/Vs, n_p = (5.9±0.18)×10^10 cm^-2) may systematically exclude the kind of degradation that the edge outlier reveals. The conclusion of 'homogeneous' wafer-scale performance is therefore stronger than the data support; the authors should either present edge-location measurements from multiple wafers or explicitly restrict the uniformity claim to the central wafer area.","section":"Magnetotransport characterization and Fig. 2"},{"comment":"The title and concluding section claim 'high yield,' but the manuscript contains no yield statistics: no definition of a working device, no pass/fail criterion, and no fraction of devices meeting a performance specification. The reported averages are computed from a small, selected set of devices (one center device per wafer for six wafers, plus 12 devices on one wafer). To support the 'high yield' claim, the authors need to provide yield data or at least a clear statement of the selection procedure; otherwise the claim should be removed or qualified.","section":"Title, abstract, and conclusion"}],"minor_comments":[{"comment":"The text contains a duplicated word: 'If if we consider local-field corrections' should read 'If we consider local-field corrections.'","section":"Fig. 2c paragraph"},{"comment":"The abbreviation 'HADDF' in the Fig. 1 caption should be 'HAADF' to match the high-angle annular dark-field terminology used in the main text.","section":"Fig. 1 caption"},{"comment":"The top interface width is reported as 4*tau_top = 0.79 nm while the bottom is reported as tau_bot = 1.04 nm; please clarify whether the 4*tau parameter was applied consistently and state the definition for each interface.","section":"Quantum well interface width, Fig. 1d"},{"comment":"Please specify the magnetic-field range and the B->0 extrapolation procedure used to extract mobility from rho_xx and rho_xy, and define n_min in the percolation fitting range (n_min <= n_range <= 1.2×10^11 cm^-2).","section":"Transport extraction methods"},{"comment":"The caption states that the red dashed line is a linear fit to the mean trend; please clarify whether the reported capacitance statistics come from fits to individual devices or from fits to the mean curve.","section":"Figure 2b and capacitance statistics"},{"comment":"There are minor typographical issues: 'F orschungszentrum' should be 'Forschungszentrum', the license should be identified accurately as CC BY-NC-SA 4.0, and reference groupings such as '192' and '23171926' should be formatted as separate citations.","section":"Affiliation and references"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid process-characterization report and the transport analysis itself is internally consistent. The main risk is overclaiming process uniformity and yield from a very small, center-selected device sample. I recommend major revision rather than rejection because the sampling issue can be addressed by adding edge data or by carefully qualifying the claims in the title and conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a solid, incremental process-development paper: Si/SiGe heterostructures and gate stacks in an industrial 200 mm BiCMOS pilot line, characterized with standard Hall-bar magnetotransport at 1.5 K. If you work on Si/SiGe spin qubits, the multi-wafer numbers are worth knowing: average peak mobility (4.25 +/- 0.17) x 10^5 cm2/Vs and percolation density (5.9 +/- 0.18) x 10^10 cm-2 across six wafers, plus a twelve-device map of one wafer. Those values are plausible and compare favorably with state-of-the-art stacks from research fabs. The structural characterization (STEM/EDX, XRD strain, interface widths) and the capacitance cross-check are careful. The analysis is standard; the percolation exponent is taken from external literature, and two devices were cross-checked in independent labs. This is a new dataset and an honest extension of the authors' ECS Trans. process-development paper.\n\nThe soft spots are statistical and presentational, not mathematical. The wafer-to-wafer claim rests on one selected center device per wafer. The wafer-A map includes a near-edge device with peak mobility far outside the standard deviation and a high-density mobility exponent of -0.6; the authors acknowledge it but still conclude homogeneous performance across the wafer. Center-only sampling on wafers B-F may systematically miss the same kind of edge degradation, so the reproducibility claim is broader than the evidence. Also, the title says \"high yield\" but there are no yield statistics anywhere. The supplementary material is referenced but not present in the arXiv posting, and the data are available only on request. None of this undermines the measured numbers; it qualifies the conclusions.\n\nWho is this for? Process engineers in the spin-qubit field and anyone benchmarking Si/SiGe stacks for scaling. It is not a conceptual advance, but it is a useful manufacturability reference. I'd send it to peer review, asking for additional edge sampling on at least a couple more wafers and some yield information, or a title that doesn't claim yield. With those revisions, it would be a solid APL-type paper.\n\nBest.","headline":"Useful BiCMOS Si/SiGe process benchmark with clean transport data, but the multi-wafer statistics rest on one center device per wafer and the 'high yield' title outruns the data.","tokens_in":10574,"tokens_out":3148,"would_cite":true,"duration_ms":30375,"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":"This paper claims that a 200 mm BiCMOS pilot line reproducibly yields low-disorder Si/SiGe heterostructures, with peak mobility $(4.25\\pm0.17)\\times10^5$ cm$^2$/Vs and percolation density $(5.9\\pm0.18)\\times10^{10}$ cm$^{-2}$ across six…","keywords":["Si/SiGe heterostructures","spin qubits","two-dimensional electron gas","magnetotransport","Hall bar field-effect transistor","BiCMOS pilot line","percolation density","quantum well disorder"],"falsifier":"Measure transport on a dense map of devices across whole wafers from several runs, including edge dies and multiple radii. If a significant fraction of devices shows peak mobility below the reported distribution or percolation density above roughly $1\\times10^{11}$ cm$^{-2}$, the claimed wafer-scale homogeneity and run-to-run reproducibility would be refuted.","tokens_in":9558,"feed_emoji":"⚛️","tokens_out":11267,"duration_ms":103284,"temperature":0.7,"pith_summary":"The paper claims that a 200 mm BiCMOS pilot line can reproducibly fabricate Si/SiGe heterostructures and gate stacks whose two-dimensional electron gases (2DEGs) are both high-quality and low-disorder, which is what scalable spin-qubit manufacturing requires. The evidence comes from Hall-bar field-effect transistors measured at 1.5 K: across six separately grown wafers, peak Hall mobility is $(4.25\\pm0.17)\\times10^5$ cm$^2$/Vs and the percolation density is $(5.9\\pm0.18)\\times10^{10}$ cm$^{-2}$, with narrow spreads of a few percent. Twelve additional devices across one wafer return similar values, supporting wafer-scale homogeneity. The authors conclude that this material and process provide a suitable platform for the development of scalable Si/SiGe-based quantum devices.","feed_headline":"Peak mobility 4.25 × 10^5 repeats across six Si/SiGe wafers","feed_subtitle":"Uniform low-disorder electron layers mean spin-qubit chips could be mass-produced on 200 mm wafers.","key_machinery":"The load-bearing measurement platform is the Hall-bar field-effect transistor (HB-FET): an optically patterned Hall bar with a TiN top gate and a plasma-deposited SiO$_2$ dielectric whose stack mirrors a qubit gate stack while giving direct magnetotransport access to the two-dimensional electron gas in the 8 nm Si quantum well. Three quantitative probes carry the argument: the carrier density versus gate voltage slope yields the stack capacitance and therefore dielectric and barrier thickness; the density-dependent Hall mobility ($\\sigma_{xx} = ne\\mu$) separates scattering regimes through the power-law exponent $\\alpha$; and a fit of low-density conductivity to the percolation law $\\sigma_{xx} \\propto (n - n_p)^p$ with $p = 1.31$ extracts the percolation density $n_p$, a direct low-density disorder metric. These probes convert wafer-scale uniformity and run-to-run reproducibility into numbers.","core_discovery":"The central discovery is that an industry-standard 200 mm BiCMOS pilot line produces shallow undoped Si/SiGe quantum wells whose transport properties are both high and tightly distributed. Across wafers A to F the maximum Hall mobility averages $(4.25\\pm0.17)\\times10^5$ cm$^2$/Vs, and across one wafer it averages $(4.20\\pm0.28)\\times10^5$ cm$^2$/Vs; the percolation densities are $(5.9\\pm0.18)\\times10^{10}$ cm$^{-2}$ and $(6.24\\pm0.33)\\times10^{10}$ cm$^{-2}$, respectively. The authors interpret the low-density mobility rise ($\\alpha\\approx2.47$) as remote-impurity Coulomb scattering and the low percolation threshold as evidence of a low-disorder landscape in the quantum well, comparable to or better than state-of-the-art field-effect stacks used for quantum devices. Structural imaging supports the electrical picture: no residual Si cap, no significant Ge pile-up at the gate interface, abrupt quantum-well interfaces, and the expected $1.29\\%$ tensile strain in the Si channel.","pith_inferences":["Editorial inference: the single edge-device outlier with suppressed peak mobility hints that practical yield could be increased by excluding the wafer edge, but the sparse sampling cannot establish an edge-exclusion rule.","Editorial inference: transport metrics are proxies; the decisive follow-up is to build quantum dots on these wafers and measure charge noise, valley splitting, and single-electron operation directly.","Editorial inference: the observed saturation of carrier density at high gate voltage, attributed to electrons tunneling into dielectric-interface traps, may cap the usable density range for devices; whether this limits qubit operation is not addressed here.","Editorial inference: a direct next experiment is growing the same stack with an isotopically enriched $^{28}$Si quantum well and checking that mobility and percolation density survive the isotope switch, as the authors expect from the growth chemistry."],"forward_implications":["Across multiple fabrication runs a single 200 mm line can deliver peak 2DEG mobility above $4\\times10^5$ cm$^2$/Vs, matching the best field-effect stacks reported for quantum devices.","Percolation densities near $6\\times10^{10}$ cm$^{-2}$ mean the quantum well stays low-disorder at the low carrier densities where gate-defined quantum dots are operated.","The few-percent spread in capacitance, mobility, and percolation density across a wafer means many nominally identical qubit devices can be made side by side, a prerequisite for high-yield processor fabrication.","Because the growth chemistry is standard silane and germane reduced-pressure CVD, the same process can be applied to isotopically purified $^{28}$Si to obtain nuclear-spin-free heterostructures for long-coherence qubits.","The gate-stack characterization shows the sacrificial Si cap is fully consumed without leaving Ge pile-up, removing a known source of disorder at the dielectric interface."],"supporting_citations":[{"why":"Provides the earlier process-integration baseline on the same pilot line whose transport performance this work improves.","marker":"20"},{"why":"Supplies the comparison state-of-the-art low-disorder Si/SiGe quantum wells whose transport values are matched or exceeded.","marker":"17"},{"why":"Establishes the benchmark for industrially manufactured field-effect stacks used in quantum transport measurements.","marker":"19"},{"why":"Identifies remote-impurity Coulomb scattering as the expected low-density mobility-limiting mechanism, grounding the exponent interpretation.","marker":"31"},{"why":"Supplies the critical exponent $p=1.31$ used in the percolation fit that extracts $n_p$.","marker":"35"},{"why":"Explains the high-gate-voltage saturation of carrier density by electron tunneling into dielectric-interface traps.","marker":"29"},{"why":"Provides the no-epitaxial-Si-cap wafer-scale low-disorder reference and the Ge pile-up comparison for the gate interface.","marker":"26"},{"why":"Shows the same Hall-bar FET platform being used to probe undoped silicon quantum-well field-effect devices.","marker":"18"}],"fun_headline_variants":["BiCMOS line yields uniform low-disorder Si/SiGe for spin qubits","200 mm BiCMOS line produces low-disorder Si/SiGe for qubits","Uniform high-mobility Si/SiGe wafers from BiCMOS pilot line","Spin qubit wafers show low disorder and high mobility in pilot line","BiCMOS line yields reproducible low-disorder Si/SiGe for qubits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim of reproducibility and uniformity rests on assuming that one device from the center of each of six wafers, plus twelve devices from one wafer, represent the full wafers; if defective or nonuniform regions fell outside those probed spots, the narrow reported distributions would overstate process stability.","fun_headline_variants_meta":{"raw":{"variants":["BiCMOS line yields uniform low-disorder Si/SiGe for spin qubits","200 mm BiCMOS line produces low-disorder Si/SiGe for qubits","Uniform high-mobility Si/SiGe wafers from BiCMOS pilot line","Spin qubit wafers show low disorder and high mobility in pilot line","BiCMOS line yields reproducible low-disorder Si/SiGe for qubits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000891,"raw_usage":{"total_tokens":3921,"prompt_tokens":1103,"completion_tokens":2818,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":719,"completion_tokens_details":{"reasoning_tokens":2710}},"tokens_in":719,"tokens_out":2818,"duration_ms":20063,"temperature":1.0,"reasoning_tokens":2710,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:49:05.752790+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure transport on a dense map of devices across whole wafers from several runs, including edge dies and multiple radii. If a significant fraction of devices shows peak mobility below the reported distribution or percolation density above roughly $1\\times10^{11}$ cm$^{-2}$, the claimed wafer-scale homogeneity and run-to-run reproducibility would be refuted.","supporting_citations":[{"cited_title":"Li , author L","cited_arxiv_id":null,"evidence_quote":"Provides the earlier process-integration baseline on the same pilot line whose transport performance this work improves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the comparison state-of-the-art low-disorder Si/SiGe quantum wells whose transport values are matched or exceeded."},{"cited_title":"Xue , author M","cited_arxiv_id":null,"evidence_quote":"Establishes the benchmark for industrially manufactured field-effect stacks used in quantum transport measurements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies remote-impurity Coulomb scattering as the expected low-density mobility-limiting mechanism, grounding the exponent interpretation."},{"cited_title":"\\ Su , author Y","cited_arxiv_id":null,"evidence_quote":"Supplies the critical exponent $p=1.31$ used in the percolation fit that extracts $n_p$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Explains the high-gate-voltage saturation of carrier density by electron tunneling into dielectric-interface traps."},{"cited_title":"Wild , author J","cited_arxiv_id":null,"evidence_quote":"Provides the no-epitaxial-Si-cap wafer-scale low-disorder reference and the Ge pile-up comparison for the gate interface."},{"cited_title":"Künne , author A","cited_arxiv_id":null,"evidence_quote":"Shows the same Hall-bar FET platform being used to probe undoped silicon quantum-well field-effect devices."}],"review_version":2}