{"id":"e6fefc17-a000-4474-adac-e9bc065c178a","arxiv_id":"2608.05749","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A single 4D-STEM scan per setting reconstructs the mixed quantum state of a programmable electron vortex phase plate, showing purity falling from 0.47 to 0.24 with bias.","lead":"Electron beams shaped by programmable phase plates are usually assumed to be pure quantum states; this team directly measured the delivered beam and found it substantially mixed, with purity falling from about 0.47 to 0.24 as the voltage increased. The same single scan also yields the beam's coherence, angular momentum content, and dose efficiency, turning one acquisition into a quantum-state acceptance test for these devices.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Detector sampling and finite-rank mode reconstruction could produce the observed purity decline with |ℓ|; the purity trend is not yet separable from measurement bias.","rationale":"The reader correctly identified two load-bearing missing controls: the contiguous acquisition confound and the detector-sampling/rank-model bias. I single out the second as the more fundamental threat to the central claim because it directly targets the ℓ-dependence of the purity trend and is explicitly left unresolved by the paper: the round-beam control lacks the ℓ-dependent phase structure that could trigger the bias, and the injection–recovery tests use the same idealized forward model. The paper's own statement that detector point-spread is mathematically absorbed into probe-mode mixing, together with its admission that finite pixel integration and detector MTF can bias the recovered mode rank at high |ℓ|, makes this a genuine soft spot rather than a manufactured one. I credit the paper for the round-beam reference, mode-count sweeps, mask-width sensitivity, and explicit limitations section, all of which support the methodology but do not close this particular gap. The randomized interleaved re-sweep is also needed to rule out drift/contamination, but the detector-sampling control is the one check that would most directly decide whether the headline purity-versus-charge trend is a delivered-beam property. Since the authors themselves present the result as system-level and bound device attribution as an upper limit, the existing CONDITIONAL verdict already reflects this concern; a conclusive test could upgrade it, but no new failure is established here.","tokens_in":773,"tokens_out":701,"duration_ms":64660,"concrete_test":"Acquire the same physical probe (fixed source, aperture, and bias) for ℓ = 0, ±8, and ±17 at native detector sampling, 2×2 binning, and a second camera length, reconstruct each with the same eight-mode model and identical settings, and compare recovered purity versus |ℓ|. If the purity slope changes by more than the stated ≈0.03–0.05 model-dependence bound between sampling conditions, the monotonic purity decline is at least partly a detector-sampling artifact rather than a delivered-beam property.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—purity falling from ≈0.47 to ≈0.24 and inconsistent with a fixed lateral source-blur model—requires that the reconstructed mode-rank falloff with |ℓ| reflect the delivered beam, not the measurement. The paper itself (Discussion, detector-side contribution; Sec. F.) concedes that finite pixel integration, detector MTF, and undersampling are mathematically absorbed into probe-mode mixing and can bias the recovered mode rank at high |ℓ|, because the azimuthal phase gradient of a charge-ℓ vortex produces finer scan-dependent interference structure as ℓ grows. A charge-dependent detector-sampling bias would produce exactly the observed purity decline. The round-beam reference (purity 0.90–0.98) does not control for this, since it lacks that ℓ-dependent structure; the injection–recovery test uses the same idealized forward model and so does not certify the real detector. Unless this channel is quantified, the purity trend is a system-level reconstruction result, not a property of the delivered beam.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a method for reconstructing the transverse density matrix of a programmable electron optical element from a single four-dimensional STEM scan per state, using mixed-state ptychography. The method is demonstrated on a MEMS electrostatic spiral phase plate across programmed charges ℓ = −17…+17. The authors report that the delivered beam is substantially mixed, with purity falling from ≈0.47 to ≈0.24 as the applied bias grows, that the real-space coherence width stays near 1 nm, and that the same data provide an in-situ voltage-to-OAM calibration, virtual OAM sorting, and partial-coherence-aware dose-efficiency metrics. The manuscript includes an unusually extensive limitations section, round-beam reference scans, injection–recovery tests, mode-count and mask-width robustness checks, and openly deposited data and code.","tokens_in":22915,"tokens_out":6208,"duration_ms":58002,"significance":"If the central quantitative claim survives scrutiny, this is a significant advance: it would turn a standard 4D-STEM acquisition into a quantum-state acceptance test for programmable electron optics, replacing assumed purity and coherence with measured, per-state specifications. The paper's strengths include the round-beam reference reconstructions (purity 0.90–0.98), the injection–recovery self-consistency test (six of eight purities reproduced within ≈0.04), the mode-count plateau at 6–8 modes, the mask-width insensitivity (<0.02), and the branch-symmetric purity slopes. The manuscript is also commendably explicit about its own limitations. However, the headline purity trend is not yet separable from a charge-dependent detector-sampling bias, and the current data do not disentangle voltage from acquisition-time drift. These are load-bearing gaps for the main claim, and they require additional experimental or forward-model controls rather than mere rewording.","major_comments":[{"comment":"The central claim that purity falls from ≈0.47 to ≈0.24 with charge, and is 'inconsistent with a fixed lateral source-blur model,' is not yet separable from a charge-dependent detector-sampling bias. The manuscript itself states in Sec. III and Sec. F that finite pixel integration, detector MTF, and undersampling are mathematically absorbed into probe-mode mixing and can bias the recovered mode rank at high |ℓ|, because the azimuthal phase gradient of a charge-ℓ vortex produces progressively finer scan-dependent interference structure. The round-beam control (purity 0.90–0.98) does not exercise this channel, since it lacks the ℓ-dependent structure, and the injection–recovery test uses the same idealized forward model and therefore cannot certify the real detector. To support the headline claim, the authors should provide either a forward-model study that includes an independently measured detector MTF and pixel integration for the actual detector, or experimental detector-sampling controls such as native versus 2×2 binning and a second camera length at fixed probe. Without such a bound, the purity trend is a system-level reconstruction result, not an established property of the delivered beam.","section":"Sec. III / Sec. F"},{"comment":"The monotonic purity decrease is inferred from one contiguous acquisition per charge state, in which the applied bias is ordered in acquisition time. The manuscript is explicit that voltage is confounded with acquisition time, drift, hysteresis, contamination, and device charging history, but this confound is load-bearing for the abstract's statement that purity falls 'as the applied bias grows.' A randomized, interleaved re-sweep with repeated zero-bias and repeated charge states, analyzed with bootstrap confidence intervals on the purity slope, is needed to separate a genuine voltage dependence from a time-dependent artifact. This is listed in Sec. F as a required control; the headline result should be re-derived from such data before publication.","section":"Sec. F"},{"comment":"The adjacent-OAM degree of coherence g is used in Sec. C to conclude that the state is a partially coherent mixed-OAM vortex with a charge-independent plateau g = 0.30 ± 0.03. However, Sec. F.1 reports per-state reconstruction-seed standard deviations up to 0.14 (a factor of about three) at the highest charges, and the plateau spans exactly those high-charge states. A three-seed mean does not reduce this scatter for a ratio built from a single off-diagonal element. The paper's own caution against interpreting individual high-charge g values should be extended to the plateau claim, or the plateau should be reported with per-state seed uncertainty propagated, rather than as a global ±0.03.","section":"Sec. F.1 and Sec. C"}],"minor_comments":[{"comment":"The DOI '10.5281/zenodo.CODE' is a placeholder and should be replaced with a working DOI before acceptance.","section":"Code Availability"},{"comment":"The phrase 'the two device-in ℓ = 0 scans' appears to be a typo; consider rewording to 'the two ℓ = 0 scans acquired with the device installed' for clarity.","section":"Sec. D"},{"comment":"The text states that the ±0.1 ℏ/V regression standard error is a lower bound on total calibration uncertainty; the figure caption should state this as well, since the plotted error bar may otherwise be misread as the total uncertainty.","section":"Fig. 4(c)"},{"comment":"The electrode-charging model would be more reproducible if the explicit expressions for the potentials N1 and N2, including the convolution with ln|k|, were written out in the text or in an appendix rather than left to the code.","section":"Sec. E, Eq. (11)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is unusually candid about its limitations, which is a genuine strength. The principal risk is that the headline purity-versus-charge trend may be dominated by detector-sampling bias or by the voltage/time confound; both are explicitly acknowledged by the authors but not yet bounded by data. Because the necessary controls are concrete and feasible, I regard this as a major-revision situation rather than a rejection, but the paper should not be accepted until the central trend is shown to survive those controls."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the first measured transverse density matrix, OAM coherences, and delivered purity for a programmable electron phase plate, and the experimental/reconstruction work is careful. But the headline purity trend (0.47 to 0.24) is a system-level reconstruction result, not yet a property of the delivered beam; the paper says so itself, and the abstract gets ahead of that.\n\nWhat's new: the two-stage pipeline—direct bright-field-shift probe estimation followed by mixed-state ptychography—is a sensible way to initialize and refine the probe state. The virtual OAM sorter with per-dataset axis refinement and virtual aberration correction goes beyond what a hardware log-polar sorter gives, since it accesses off-diagonal coherences. The partial-coherence-aware SSNR/DQE extension is a useful generalization and reduces properly to known limits. The internal controls are genuinely good: round-beam reference scans reconstruct nearly pure, injection-recovery reproduces six of eight purities within 0.04, mode-count plateaus, mask-width insensitivity. Data are on Zenodo with a manifest. That's reproducible practice.\n\nSoft spots, in order of importance. First, the central quantitative claim—purity falling with |ℓ|, inconsistent with fixed source blur—is confounded with detector sampling. The azimuthal phase gradient of a vortex gets finer as ℓ grows; finite pixel integration, MTF, and undersampling get absorbed into probe-mode mixing and can bias the recovered mode rank exactly as observed. The round-beam control doesn't share that ℓ-dependent structure, and injection-recovery uses the same idealized forward model, so neither certifies the real detector. The paper acknowledges this in the Discussion and Sec. F, calling it a limitation of measurement and reconstruction, not physical decoherence—but then the abstract still calls the beam \"substantially mixed\" and says the trend is \"inconsistent with a fixed lateral source-blur model.\" That's overreach relative to their own caveats.\n\nSecond, the contiguous acquisition confounds voltage with time: drift, contamination, charging history. The authors note this and call for a randomized interleaved re-sweep. Fine, but it means the trend is not yet established as a voltage dependence.\n\nThird, the virtual sorter refines its axis and aberration correction by maximizing spectral concentration on the same data; the paper partially bounds the effect but the off-diagonal coherence g is not calibrated. Fourth, no pointwise error bars or bootstrap intervals, and code is not yet released. These are all stated.\n\nOverall: this is a serious, honest paper with a real methodological contribution. The physics story is plausible but not yet separated from measurement bias. It deserves a serious referee and, if published, should carry the caveats in the abstract. Recommended: send to review, with emphasis on asking the authors to add the randomized re-sweep and detector-sampling controls, or at minimum to reframe the abstract as system-level.","headline":"First measured density matrix for programmable electron optics, backed by unusually honest controls, but the headline purity trend is not yet separable from detector-sampling and time-confounding biases.","tokens_in":23545,"tokens_out":2539,"would_cite":true,"duration_ms":22288,"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":"One four-dimensional STEM scan per programmed state reconstructs the transverse density matrix of an electron beam from a programmable electrostatic phase plate, showing the beam is substantially mixed — purity falling from about 0.47 to…","keywords":["electron vortex beams","mixed-state ptychography","transverse density matrix","orbital angular momentum","programmable phase plate","spatial coherence","dose efficiency","4D-STEM"],"falsifier":"Run the same MEMS device through a randomized-order re-sweep with repeated zero-bias and repeated charge states; if purity no longer decreases monotonically with delivered charge, the reported trend is an artifact of acquisition order. Separately, fixing the physical probe and comparing reconstructions at native versus 2×2 binned detector sampling at a second camera length would test whether detector point-spread biases high-|ℓ| reconstructions toward lower purity.","tokens_in":22547,"feed_emoji":"🌀","tokens_out":6363,"duration_ms":50233,"temperature":0.7,"pith_summary":"This paper establishes that the transverse quantum state of an electron beam delivered by a programmable electrostatic phase plate can be fully reconstructed from a single four-dimensional STEM scan per setting, using mixed-state ptychography with no extra hardware. The central experimental result is that the delivered beam is far from pure: its purity falls from about 0.47 to about 0.24 as the applied bias grows, contradicting a simple fixed source-blur model, while the real-space coherence width stays near one nanometre. The same scans yield the beam's orbital-angular-momentum density matrix, an in-situ voltage-to-charge calibration, and dose-efficiency metrics, turning one acquisition into a quantum-state acceptance test for programmable electron optics. This matters because proposals for dose-efficient phase imaging, shaped-electron X-ray sources, and chiral spectroscopy all assume a pure, fully coherent delivered wave that until now had never been measured.","feed_headline":"Purity of twisted electron beams falls from 0.47 to 0.24 with bias","feed_subtitle":"One ptychographic scan per setting yields the beam's density matrix, OAM spectrum, coherence widths, and dose efficiency.","key_machinery":"The argument rests on a two-stage reconstruction pipeline. Stage one extracts the complex aperture function directly from the parallax shifts of the bright-field disk, using a loop integral whose circulation separates the vortex winding (effective topological charge) from scan drift; stage two refines the probe as an effective finite-rank density matrix by mixed-state ptychography, jointly optimizing eight physical probe modes and a single object transmission. The accompanying partial-coherence-aware transfer theory extends the sideband-overlap SSNR formalism to arbitrary mixed probes: the coherence envelope D_coh = L_ρ/L_up multiplies the shot-noise-limited SSNR, reducing to known source-size and chromatic envelopes in the appropriate limits.","core_discovery":"A microelectromechanical electrostatic spiral phase plate does not deliver the pure helical wavefront that programmable-electron-optics proposals assume; the delivered beam is a substantially mixed, partially coherent state. Mode-power analysis of eight reconstructed probe modes shows the dominant-mode power and purity Tr ρ² declining linearly with delivered orbital-angular-momentum charge, from a measured purity of about 0.47 at zero bias to about 0.24 at the largest programmed charge, while the von Neumann entropy rises from about 1.1 to about 1.6 nats. The real-space coherence envelope stays flat near 1 nm, and the Fourier-space envelope narrowing with charge is reproduced by a pure vortex control, identifying it as a diagnostic artifact rather than loss of angular coherence; the physically meaningful OAM density matrix instead shows a coherent-to-incoherent step: adjacent-channel coherence g ≈ 0.78 for the round beam falling to a charge-independent plateau near 0.30. The delivered charge is non-integer (gain 16.8 ± 0.1 ħ/V), so part of the measured OAM spread is a coherent consequence of fractional vortex structure, and the measured mixedness implies that purifying the output could improve dose efficiency roughly threefold.","pith_inferences":["If the method transfers to other dynamic shaping devices (pixel-array plates, Ampere phase plates), the same single-scan test could become a standard acceptance measurement for any programmable electron optic, quantifying device-to-device variation.","The measured ~1 nm real-space coherence patch implies, by the coherent-area scaling used for shaped-electron X-ray sources, enhancements of order ten for beams like those delivered here; reaching 1000× would require roughly ten-nanometre coherence, which is a quantitative design target.","A randomized, interleaved voltage re-sweep with repeated zero-bias states — the control the authors state they did not perform — would settle whether the purity decrease is genuinely charge-driven or a time/drift artifact; this is the most direct next experiment."],"forward_implications":["Each programmed state gets a measured per-state specification — mode powers, OAM spectrum, coherence widths, and dose-to-SNR requirements — replacing the assumed-pure inputs used by vortex magnetometry, phase-plate imaging, and shaped-electron X-ray proposals.","The reconstructed charge tracks the programmed bias linearly (≈ 16.8 ħ/V), so a standard 4D-STEM camera becomes an in-situ calibration instrument for tunable phase plates.","Because the measured mixedness erases most of the coherent vortex's dose-efficiency advantage, purifying the delivered beam is the single largest expected improvement in dose-efficient structured illumination.","Virtual OAM sorting recovers the off-diagonal elements of the OAM density matrix — the inter-channel coherences — that hardware intensity-based sorters cannot access."],"supporting_citations":[{"why":"Supplies the mixed-state ptychographic method that recovers a finite-rank probe density matrix from one scanning-diffraction dataset.","marker":"[35]"},{"why":"Describes the MEMS electrostatic spiral phase plate whose delivered beam is characterized here.","marker":"[19]"},{"why":"The hardware log-polar OAM sorter that the virtual sorting emulates and extends to off-diagonal coherences.","marker":"[29]"},{"why":"Explains how a pure non-integer vortex decomposes over integer OAM channels, used to bound how much of the measured spectral spread is coherent.","marker":"[38]"},{"why":"Defines the source-size coherence envelope that the paper's partial-coherence-aware transfer theory generalizes to arbitrary mixed probes.","marker":"[44]"},{"why":"Identifies thermal magnetic field noise as a decoherence channel from nearby conductors, evaluated for the device's doped-silicon electrodes.","marker":"[45]"},{"why":"Derives the electron-optical decoherence from thermal field noise, the charge-independent baseline candidate.","marker":"[46]"},{"why":"Supplies the finite-dose spectral-SNR metric that the paper extends to mixed, pixelated probes.","marker":"[42]"}],"fun_headline_variants":["Purity of twisted electrons drops from 0.47 to 0.24 with bias","Electron vortex purity halves as bias increases","One scan quantifies purity drop in twisted electron beams","Twisted electron purity: 0.47 to 0.24 as bias grows","Vortex electrons lose purity from 0.47 to 0.24 under bias"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central trend assumes that each charge state is faithfully represented by its single contiguous acquisition, so that the applied voltage is not confounded with acquisition order, drift, contamination, or the device's charging history; the authors flag that only a randomized interleaved re-sweep can separate these.","fun_headline_variants_meta":{"raw":{"variants":["Purity of twisted electrons drops from 0.47 to 0.24 with bias","Electron vortex purity halves as bias increases","One scan quantifies purity drop in twisted electron beams","Twisted electron purity: 0.47 to 0.24 as bias grows","Vortex electrons lose purity from 0.47 to 0.24 under bias"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002284,"raw_usage":{"total_tokens":8835,"prompt_tokens":982,"completion_tokens":7853,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":598,"completion_tokens_details":{"reasoning_tokens":7757}},"tokens_in":598,"tokens_out":7853,"duration_ms":49943,"temperature":1.0,"reasoning_tokens":7757,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T00:06:39.718840+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same MEMS device through a randomized-order re-sweep with repeated zero-bias and repeated charge states; if purity no longer decreases monotonically with delivered charge, the reported trend is an artifact of acquisition order. Separately, fixing the physical probe and comparing reconstructions at native versus 2×2 binned detector sampling at a second camera length would test whether detector point-spread biases high-|ℓ| reconstructions toward lower purity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the mixed-state ptychographic method that recovers a finite-rank probe density matrix from one scanning-diffraction dataset."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the MEMS electrostatic spiral phase plate whose delivered beam is characterized here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The hardware log-polar OAM sorter that the virtual sorting emulates and extends to off-diagonal coherences."},{"cited_title":"Juchtmans and J","cited_arxiv_id":null,"evidence_quote":"Explains how a pure non-integer vortex decomposes over integer OAM channels, used to bound how much of the measured spectral spread is coherent."},{"cited_title":"Jiang et al., Electron ptychography of 2D materials to deep sub-ångström resolution, Nature 559, 343 (2018)","cited_arxiv_id":null,"evidence_quote":"Defines the source-size coherence envelope that the paper's partial-coherence-aware transfer theory generalizes to arbitrary mixed probes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies thermal magnetic field noise as a decoherence channel from nearby conductors, evaluated for the device's doped-silicon electrodes."},{"cited_title":"Leach, E","cited_arxiv_id":null,"evidence_quote":"Derives the electron-optical decoherence from thermal field noise, the charge-independent baseline candidate."},{"cited_title":"Thibault and A","cited_arxiv_id":null,"evidence_quote":"Supplies the finite-dose spectral-SNR metric that the paper extends to mixed, pixelated probes."}],"review_version":1}