{"id":"762ddfd9-397c-48f6-b8ca-0b4ea92bec67","arxiv_id":"2509.10726","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Ptychography tolerates large scan jumps and irregular trajectories, while EELS and ADF maps remain sensitive to scan-induced artifacts, so scan design should depend on the imaging modality.","lead":"This paper compares different electron beam scan patterns in a transmission electron microscope, finding that a computational imaging method called ptychography keeps atomic resolution even with irregular scan steps, while chemical maps from EELS show scan-related artifacts. It offers practical guidance for low-dose and cryogenic microscopy.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Residual thickness/defocus variation across scan-pattern datasets is the weak link in the trajectory-attribution claim; it needs a per-dataset control.","rationale":"The paper does valuable work: it directly benchmarks multiple programmable scan trajectories on a model perovskite, uses a standard reconstruction package (py4DSTEM), and openly identifies several limitations, including residual thickness/defocus variation and the need for dose-quantification maps and real-time drift correction. However, the evidence for the headline differential claim is qualitative, and the one confound the authors themselves disclose is exactly the kind that could generate the observed pattern without any true trajectory effect. The reader's weakest-assumption analysis pointed to this same issue. A concrete per-dataset thickness/defocus measurement from the existing 4D-STEM data would settle whether the concern lands, so no change to the conditional verdict is warranted.","tokens_in":11873,"tokens_out":7273,"duration_ms":91916,"concrete_test":"Use the recorded 4D-STEM datasets for each scan pattern (raster, N=2/4/8, spiral) to fit per-dataset thickness and defocus from the same data, e.g., via the py4DSTEM multislice reconstruction or position-averaged CBED patterns, and compare these fitted values across scan modes. If fitted thickness or defocus differences exceed roughly one slice (2 nm) or otherwise correlate with the scan-mode ordering, the trajectory attribution is not established; a matched-defocus control on a single DyScO3 region would then be needed to separate trajectory effects from sample/optical drift.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—ptychography tolerates trajectory irregularities while EELS/ADF do not—rests on comparing datasets acquired with different scan patterns. The Methods ('4D-STEM Data Acquisition') state: 'Variations in sample thickness and defocus were minimized, but not eliminated, across the different scan pattern datasets.' EELS and 4D-STEM were also acquired separately (different probe currents, step sizes, dwell times, number of passes), with defocus manually optimized per dataset in the 10–20 nm range. If thickness or defocus varied systematically with scan mode, it would produce exactly the reported pattern: ptychography, which reconstructs the probe and can absorb some aberrations in a mixed-state multislice model, would look robust, while EELS/ADF, which depend on direct pixel registration and local signal statistics, would show non-uniform contrast or replicated features. The conclusion that the scan trajectory itself causes the observed differences requires that these residual variations be negligible or uncorrelated with scan mode. No per-dataset thickness/defocus measurements, repeated acquisitions, or quantitative error metrics are provided to exclude this confound. Since the abstract and conclusion offer practical guidelines based on this attribution, this is the load-bearing soft spot.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a systematic comparison of programmable STEM scan trajectories (raster, inward/outward spiral, and N-sequential) applied to electron energy-loss spectroscopy (EELS), annular dark-field (ADF) imaging, and ptychography. Using DyScO3 as a model perovskite, the authors evaluate atomic-resolution EELS maps, ADF-derived FFTs, and ptychographic phase reconstructions. They conclude that ptychography is largely robust to large probe jumps and non-raster trajectories, whereas EELS and ADF are sensitive to trajectory-induced artifacts such as probe-settling delays, flyback streaks, and non-uniform dose delivery. A cryogenic BaTiO3 experiment is used to argue that outward spiral scanning improves oxygen-column visibility and reveals subtle oxygen displacements consistent with a low-temperature ferroelectric phase. The paper proposes practical guidelines for choosing scan strategies in low-dose 4D-STEM.","tokens_in":12119,"tokens_out":2893,"duration_ms":37834,"significance":"If the conclusions hold, the paper provides useful practical guidance for the growing use of programmable scan trajectories in dose-sensitive electron microscopy. The claim that ptychography tolerates large scan jumps while EELS requires precise, temporally uniform illumination is of broad interest and would help experimentalists allocate dose and choose scan patterns. The use of py4DSTEM for reconstructions and the focus on cryogenic low-dose conditions are strengths. However, the support is primarily qualitative: resolution is judged from visual inspection of FFT peaks, EELS maps come from single acquisitions without error bars, and the Methods explicitly acknowledge residual sample-thickness and defocus variations across scan-pattern datasets. These limitations are load-bearing for the central attribution of observed differences to scan trajectory. The paper also gives limited side-by-side ptychographic comparisons for raster and spiral scans in the same material system, so the strongest claim of general ptychographic robustness is only partially supported.","major_comments":[{"comment":"The Methods state: 'Variations in sample thickness and defocus were minimized, but not eliminated, across the different scan pattern datasets.' This residual variation is a direct confound for the central claim that observed differences in EELS, ADF, and ptychography arise from the scan trajectory. Since each scan-pattern dataset is acquired at a different location with manually optimized defocus (10–20 nm), systematic thickness or defocus differences could produce exactly the reported pattern: ptychography, which reconstructs the probe via mixed-state multislice, would appear robust, while EELS/ADF would show non-uniform contrast or replicated features. The authors need to provide per-dataset thickness/defocus estimates, repeat acquisitions, or otherwise demonstrate that these variations are uncorrelated with scan mode.","section":"Materials and Methods, 4D-STEM Data Acquisition"},{"comment":"The claim that 'Bragg peaks remained visible to ~0.7 Å' and that ptychographic reconstructions show 'minimal differences in achievable resolution' is based solely on visual inspection of FFTs. No quantitative resolution metric (e.g., Fourier ring correlation, peak width, or signal-to-noise threshold) is provided, and no uncertainty estimate is given. Furthermore, Fig. 4 compares only N=2, N=4, and N=8 sequential scans; raster and spiral ptychography of DyScO3 are not shown in the main text. Thus the paper's headline conclusion that 'ptychography tolerates large pixel jumps' and is robust across trajectory types is only directly demonstrated for one family of trajectories (N-sequential) in the primary benchmarking material. The cryogenic BaTiO3 comparison (Fig. 5) is a single raster vs. spiral pair and is also not quantitatively assessed.","section":"Fig. 4 and discussion of ptychographic resolution"},{"comment":"The EELS elemental maps are presented from single acquisitions (15 passes with drift correction between passes) with no repeat measurements, no error bars, and no quantitative measure of contrast non-uniformity. The conclusion that sequential and spiral scans 'introduce non-uniform elemental contrast' is inferred visually. Drift correction between passes, which may itself vary with scan pattern, could generate patchiness or discontinuities that are then attributed to the trajectory. The authors should provide either repeated acquisitions with variance estimates or a quantitative metric of map uniformity across the different scan modes.","section":"EELS Data Acquisition and Fig. 3"}],"minor_comments":[{"comment":"The caption lists 'raster, random, N=8, N=4, N=2 sequential, and spiral' scan modes, but the Materials and Methods state that programmable scan patterns scripted were 'raster, spiral (inward and outward), and N-sequential scans (N = 2, 4, 8)'. There is no description of a 'random' scan mode in the methods. Please clarify whether a random trajectory was used and, if so, provide details; if not, correct the caption.","section":"Fig. 3 caption"},{"comment":"The ADF images used for FFT analysis were processed with 'Hanning filtering, quartile-based thresholding, and cropping' before the FFTs were computed. These operations can alter periodic features and introduce or suppress artifacts. Showing the raw or minimally processed FFTs, or discussing the impact of these filters, would strengthen the interpretation of the FFT features.","section":"Materials and Methods, EELS Data Acquisition and Fig. 3"},{"comment":"The Fig. 4 caption states '2 probe modes, 18 slices, 2 nm thickness, 20 nm defocus', whereas the Methods state '10–20 slices (2 nm per slice)' and 'defocus was varied between 10 and 20 nm, optimized manually for each dataset'. Please reconcile these numbers and report the actual parameters for each reconstruction, including the number of slices and defocus used for the N=2, N=4, and N=8 datasets.","section":"Fig. 4 caption vs. Materials and Methods, Ptychographic Reconstructions"},{"comment":"The paper would benefit from a data availability statement and a list of the exact code and versions used (e.g., py4DSTEM version) to facilitate reproducibility. The current text only names the software package.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper's practical message is plausible and likely of interest to the STEM community, but the experimental support needs to be tightened. In particular, the confound of residual thickness/defocus variation across scan-pattern datasets, the lack of quantitative resolution metrics, and the absence of EELS repeat measurements are load-bearing issues that prevent acceptance in the current form. These are addressable with additional analysis or targeted control experiments, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a straightforward, honest experimental paper that does something the field hadn't done in one place: it directly compares raster, spiral, and N-sequential scan patterns on the same material (DyScO3) for atomic-resolution EELS, ADF, and ptychography, and then adds a cryo BaTiO3 demo for spiral scanning. The central message—ptychography tolerates large probe jumps while EELS and ADF do not—is consistent with the prior literature, and the paper makes it concrete. The authors are careful to attribute most of the observed differences to instrument effects (coil hysteresis, flyback, probe settling) rather than radiation chemistry, which shows they're not overreaching.\n\nThe main caveat is exactly what the stress-test note flags. The Methods state that 'variations in sample thickness and defocus were minimized, but not eliminated' across scan-pattern datasets. Since all conclusions rest on comparing those datasets, this is load-bearing. If thickness or defocus drifted systematically, it could produce the reported pattern: ptychography looks robust because it reconstructs the probe, while EELS and ADF look bad because they depend on direct registration. There are no per-dataset thickness/defocus measurements, no repeated acquisitions, and no error bars. Resolution is assessed by visually checking Bragg peaks in FFTs. That's not fatal—the claim is plausible and the authors explicitly list dose quantification maps and real-time drift correction as future work—but it means this is a qualitative demonstration, not a quantitative guideline.\n\nI'd still send it to review. The comparison is useful, and the cryo spiral result—resolving oxygen columns in BaTiO3—is a practical data point. The fix is not huge: add a control or at least per-dataset estimates of thickness and defocus, and report a quantitative resolution metric. Without that, the guidelines should be read as 'here's what we saw', not 'here's what you should do'. The paper is honest about its limitations, and the thinking is clear. I'd take it seriously for a methods venue, but I'd ask for the control before relying on it.","headline":"Useful direct multimodal comparison of scan trajectories for EELS vs ptychography; the central claim is plausible, but residual thickness/defocus variation is a real confound and the support is qualitative.","tokens_in":12649,"tokens_out":3103,"would_cite":true,"duration_ms":30778,"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":"Non-raster beam paths spread dose unevenly, but ptychography keeps atomic resolution from jumpy scans that leave EELS elemental maps patchy.","keywords":["programmable beam scanning","4D-STEM","electron energy-loss spectroscopy","ptychography","low-dose electron microscopy","scan artifacts","cryo-STEM","perovskite thin films"],"falsifier":"Re-acquire the same DyScO3 region with raster, spiral, and N=8 sequential scans, interleaved under fixed thickness and defocus, and check whether EELS patchiness and ptychography robustness persist; also lower the dose for N=8 to see whether ptychography still reaches ~0.7 Å resolution.","tokens_in":11770,"feed_emoji":"🔬","tokens_out":4755,"duration_ms":54580,"temperature":0.7,"pith_summary":"This paper asks whether replacing the standard raster electron-beam scan with programmable paths—spirals and multi-pass sequential blocks—can make low-dose electron microscopy more dose-efficient and artifact-free. Using DyScO3 as a model oxide, the authors compare the same scan paths across three readouts: annular dark-field imaging, electron energy-loss spectroscopy (EELS) elemental mapping, and ptychographic phase reconstruction. The central result is a split: ptychography stays sharp (~0.7 Å) even when scan jumps are large and the trajectory is irregular, while atomic-resolution EELS maps develop patchy, non-uniform elemental contrast under the same conditions. Spiral scanning removes the line-by-line flyback artifacts of raster scans and, in cryogenic BaTiO3, resolves oxygen columns that raster scans miss. If right, the paper gives practical guidance: match the scan trajectory to the modality—flexible paths for ptychography, temporally uniform illumination for quantitative EELS.","feed_headline":"Ptychography survives jumpy scans that corrupt EELS maps","feed_subtitle":"Spiral and block-sequential beam paths disturb EELS contrast while overlapping diffraction keeps ptychography sharp—useful for low-dose imag","key_machinery":"The central object is the programmable scan trajectory—raster line-by-line, inward/outward spiral, and N×N block-sequential paths—and how each one distributes probe position errors and dose over time. The argument's load-bearing contrast is between direct imaging (ADF/EELS), where each pixel's intensity is assigned to a commanded position so positional errors become visible artifacts, and iterative ptychography, which reconstructs the object from overlapping, redundant diffraction patterns and can absorb or correct moderate scan-path irregularities. The paper also uses Fourier transforms of ADF images as a diagnostic: replicated Bragg peaks mark large-jump settling errors, vertical streaks m","core_discovery":"The paper's central claim is that the scan trajectory itself, not just the total electron dose, determines data quality in atomic-resolution STEM, and that the two measurement modalities sit at opposite ends of sensitivity. For ptychography, iterative recovery from overlapping diffraction patterns absorbs large probe-position jumps: N=2, 4, and 8 sequential scans all give phase reconstructions with Bragg peaks visible to about 0.7 Å, even though ADF images from the same datasets show discontinuities and replicated Bragg reflections. For EELS, the elemental maps depend on precise spatial registration and uniform temporal illumination, so high-N sequential and some spiral scans produce non-uni","pith_inferences":["An untested extension: random or interleaved trajectories may inherit the same ptychographic robustness, but only if positional metadata remains reliable; at very low dose, ptychographic convergence could fail.","The EELS sensitivity implies that quantitative atomic-scale spectroscopy with programmable scans should be validated with dose-uniformity metrics, such as per-pixel exposure histograms, before relying on elemental ratios.","The cryo-BaTiO3 oxygen shifts are consistent with a polar phase, but since ptychography is a 2D projection, full 3D displacement vectors would need multislice or tilt-series ptychography.","If the temporal-dose argument generalizes, spiral or block-sequential scans may delay radiolysis in organic or biological materials; this is a direct, testable prediction beyond the oxides studied here."],"forward_implications":["Ptychographic reconstructions can be paired with non-raster trajectories that spread dose, without sacrificing the ~0.7 Å resolution demonstrated in DyScO3.","EELS elemental maps from large-jump scans (N=8 sequential) show non-uniform contrast and replicated Bragg artifacts; users should avoid large jumps when quantitative spectra are needed.","Spiral scans suppress flyback streaks and produce cleaner Fourier-space data than raster; the mild radial dose gradient is a secondary trade-off.","At cryogenic temperature, outward spiral scans recover oxygen-column contrast in BaTiO3 that raster scans miss, enabling detection of subtle displacements.","FFT inspection of ADF frames is a useful diagnostic: replicated peaks indicate probe-settling or large jumps; vertical streaks indicate flyback."],"fun_headline_variants":["Ptychography immune to scan jumps that mar EELS maps","Spiral and block scans: bad for EELS, fine for ptychography","Scan trajectory matters: ptychography robust, EELS sensitive","Non-raster scans corrupt EELS, not ptychography"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that differences in sample thickness and defocus between datasets were small enough not to cause the observed scan-mode differences; the paper says these variations were 'minimized, but not eliminated'.","fun_headline_variants_meta":{"raw":{"variants":["Ptychography immune to scan jumps that mar EELS maps","Spiral and block scans: bad for EELS, fine for ptychography","Scan trajectory matters: ptychography robust, EELS sensitive","Non-raster scans corrupt EELS, not ptychography"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000198,"raw_usage":{"total_tokens":1179,"prompt_tokens":690,"completion_tokens":489,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":434,"completion_tokens_details":{"reasoning_tokens":414}},"tokens_in":434,"tokens_out":489,"duration_ms":5862,"temperature":1.0,"reasoning_tokens":414,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T17:35:30.712517+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-acquire the same DyScO3 region with raster, spiral, and N=8 sequential scans, interleaved under fixed thickness and defocus, and check whether EELS patchiness and ptychography robustness persist; also lower the dose for N=8 to see whether ptychography still reaches ~0.7 Å resolution.","supporting_citations":[],"review_version":1}