{"id":"8441120e-7685-4a05-a6a7-b13e0c677b21","arxiv_id":"2607.07273","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":4,"one_line_summary":"Rotating FIB-SEM tomography to image porous GaN perpendicular to the film surface significantly mitigates the pore back effect, and new voxel-intensity metrics quantify this improvement.","lead":"This paper shows that rotating a 3D microscopy experiment 90 degrees for porous gallium nitride films dramatically reduces image artifacts from electrons scattering through pores. The method and new metrics for quantifying artifacts could improve structural characterization of porous semiconductors used in electronics.","discovery_kind":"unclear","skeptic_critique":{"model":"glm-5.2","headline":"The intensity-based formalisms lack independent validation (no ground truth, no simulation, no scalar metric), and the generalizability claim to 'a range of pore morphologies' is only strongly supported for vertically aligned pores.","rationale":"I agree with the reader that the generalizability claim is overstated for Sample 3, but I identify a more fundamental concern: the metrics themselves lack independent validation. The reader's weakest_assumption focuses on whether the method generalizes to non-vertical pores; my concern is whether the intensity-based formalisms validly measure the pore back effect at all, even for Samples 1-2. The two setups differ in multiple ways beyond pore orientation (mounting, charging, curtaining, tracking geometry), and no control or simulation isolates the pore back effect as the causal driver of the observed intensity differences. That said, the internal consistency is reasonably compelling: identical beam conditions produce darker reconstructed pores in plan-view than in cross-sectional as-taken frames, and line profile anisotropy appears along the milling direction but not perpendicular to it — a signature hard to explain by other artifacts. The physics is also sound: aligning the milling direction with the long pore axis maximizes distance to the back wall. So the concern weakens the 'quantitatively assess' claim but does not undermine the core qualitative finding. The reader's CONDITIONAL verdict with MODERATE confidence is appropriate; my concern reinforces rather than shifts it. The paper would be strengthened by (a) scoping the abstract claim to vertically aligned pores, (b) defining a scalar metric from the histograms (e.g., void-peak prominence or bimodality index), and (c) running the simulation the authors themselves propose. None of these are present, which justifies the conditional assessment but not rejection.","tokens_in":24178,"tokens_out":4594,"duration_ms":248922,"concrete_test":"Implement the Monte Carlo simulation the authors propose in §5: construct a 3D pore geometry matching Sample 1 (wide columnar pores, ~1 µm depth, from cross-sectional SEM dimensions in Fig 11a), simulate both cross-sectional and plan-view FIB-SEM acquisition with identical beam parameters (2 kV, 65 pA, 5.7 nm slice thickness), and compare the resulting simulated voxel intensity histograms to the experimental distributions in Figure 14. If the simulated cross-sectional distribution lacks a low-intensity void peak while the simulated plan-view distribution shows one — matching the experimental pattern — the attribution to the pore back effect is validated and confounding factors are ruled out. If the simulated distributions do not reproduce the experimental pattern, the intensity differences may arise from setup-dependent artifacts rather than the pore back effect.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim has two components: (1) plan-view tomography mitigates the pore back effect, and (2) voxel intensity-based formalisms can quantitatively assess it. For (1), the evidence is physically motivated and internally consistent for vertically aligned pores (Samples 1-2): reconstructed cross-sections from plan-view data show darker pores than as-taken cross-sectional frames under identical beam conditions (§4.1, comparing Fig 13c to Fig 12a), and line profiles show anisotropy along the milling direction but not perpendicular to it (Fig 15). However, for Sample 3 (branched pores), the authors concede the advantage is 'contentious' and 'marginal' (§4.3), and the plan-view dataset suffered overexposure — yet the abstract claims 'a range of pore morphologies.' For (2), the 'formalisms' are qualitative comparison tools: the intensity histograms (Figs 14, 19, 24) are compared by eye with no scalar metric or statistical test defined, and the line profile anisotropy is demonstrated on individual pores without aggregation across the dataset. Crucially, no ground truth — simulation, phantom sample, or comparison with a pore-back-free technique — validates that the intensity distribution differences are attributable to the pore back effect specifically, rather than to other systematic differences between the two setups (sample mounting geometry, charging behavior, curtaining, surface quality, tracking structure dimensions). The authors themselves acknowledge in §5 that 'it may be of significant value to develop a simulation workflow' using Monte Carlo electron scattering, but this validation has not been performed. Without it, the causal attribution of intensity differences to the pore back effect remains an inference from internal consistency rather than an independently verified result.","agreement_with_reader":"partial"},"referee_report":{"model":"glm-5.2","summary":"This manuscript presents a new 'plan-view' FIB-SEM tomography methodology for porous GaN thin films, in which the experiment is rotated 90 degrees so that the milling direction is perpendicular to the film surface and aligned with the long axis of vertically propagating pores. The approach is compared against conventional 'cross-sectional' tomography on three samples with different pore morphologies. Two voxel-intensity-based metrics are introduced to assess the pore back effect: intensity distribution histograms and feature anisotropy line profiles. The authors find that plan-view tomography mitigates the pore back effect for vertically aligned pores (Samples 1 and 2) and offers marginal improvement for branched pores (Sample 3). The paper is well-motivated, the experimental comparison is internally consistent, and the methodology is described in sufficient detail for replication.","tokens_in":24602,"tokens_out":1509,"duration_ms":263623,"significance":"The pore back effect is a well-known but poorly quantified artefact in FIB-SEM tomography of porous materials, and this work represents a serious attempt to address it through both experimental geometry and quantitative assessment tools. The plan-view setup is a practical contribution that other groups working on mesoporous thin films can adopt. The intensity histogram comparison (e.g., Fig. 14) provides a visually compelling and physically motivated demonstration of the difference between the two approaches. The line-profile anisotropy metric (Fig. 15) is a useful diagnostic. The work is relevant to the porous GaN community and, more broadly, to FIB-SEM characterisation of mesoporous thin films on substrates.","major_comments":[{"comment":"§3 and §4: The intensity-based 'formalisms' are presented as quantitative tools, but no scalar metric or statistical test is defined. The histograms in Figs. 14, 19, and 24 are compared visually ('the more the distribution resembles the simple two-peak ideal case, the less the effect of the pore back has been'). The line-profile anisotropy (Fig. 15) is demonstrated on a single pore per sample without aggregation. To support the claim that the pore back effect can be 'quantified,' the authors should define at least one scalar quantity (e.g., a bimodality index for the histogram, or an asymmetry ratio for the line profiles) and report it for both orientations across all three samples. Without this, the assessment remains qualitative despite the language of 'formalisms' and 'quantification.'","section":null},{"comment":"§4.3 and Abstract: The abstract claims the approach 'significantly mitigates the pore back effect in porous GaN thin films with a range of pore morphologies.' However, for Sample 3 (branched pores), the authors concede the advantage is 'contentious' and 'marginal' (§4.3), the plan-view dataset suffered overexposure (Fig. 24), and the reconstructed cross-section shows 'very poor contrast' in the top half. The abstract should be revised to accurately scope the claim — e.g., 'for porous GaN thin films with vertically aligned pore morphologies, with more limited improvement for branched morphologies' — or the Sample 3 data should be re-acquired without overexposure to support the broader claim.","section":null},{"comment":"§3: No ground-truth validation is provided to confirm that the intensity distribution differences between the two orientations are attributable specifically to the pore back effect rather than to other systematic differences between the setups (mounting geometry, charging, curtaining, tracking structure dimensions, surface quality). The authors acknowledge several such confounders (curtaining in Fig. 13b, mounting tilt, software glitching in Figs. 18c and 23c). While the physical argument is sound, at least one of the following would strengthen the causal claim: (a) Monte Carlo electron scattering simulation of the expected pore back signal for each geometry, (b) a phantom sample with known pore geometry, or (c) a more explicit discussion of why the confounders cannot account for the observed intensity differences. The authors gesture toward simulation in §5 ('It may therefore be of знач","section":null},{"comment":"§4.1, Fig. 15: The line-profile anisotropy metric is noted to be inapplicable to the plan-view tomograph of Sample 1 because 'there is no front wall of the pore with GaN' and the back wall is up to 1 µm away. The authors state the metric is introduced 'only to be applied to suitable morphologies.' This limits the generality of the metric and should be stated more prominently, ideally in §3 where the metric is defined, rather than only at the end of §4.1.","section":null}],"minor_comments":[{"comment":"§2.2.2: The sentence beginning 'The total range of the m-axis of the stage in the Crossbeam 540 instrument presented another limitation...' is grammatically awkward and could be clarified.","section":null},{"comment":"Figures 12-13, 17-18, 22-23: The axis labels on the bottom-right of the images are referenced but not always legible in the figure descriptions. Ensure axis labels are clearly readable in the final figures.","section":null},{"comment":"§4.2, Fig. 19 caption: The caption states 'the value of intensity plotted against the fraction of total voxels' but the y-axis label is not explicitly described. Clarify whether the histograms are normalised.","section":null},{"comment":"§4.3: 'the contrast is also sharper than it is in the input frames for this tomograph, such as in Figure 17a' — this reference to Fig. 17a (Sample 2) appears to be an error; it should likely reference a figure from Sample 3 (Fig. 22a).","section":null},{"comment":"§5: The suggestion of a simulation workflow is valuable but somewhat underdeveloped. Consider moving this to an outlook section or expanding briefly on what inputs would be needed.","section":null},{"comment":"Reference [13] is cited as 'J. Appl. Phys. 139 (17) (2026) 175702' — verify this is not a preprint or in-press reference and that the year/volume are correct.","section":null},{"comment":"The term 'pore back effect' is introduced in §1 but the alternative 'shine-through artefacts' (ref [15]) is mentioned only once. Consider using both terms initially for searchability.","section":null}],"recommendation":"major_revision","confidential_remarks":"The reader's stress-test concern about the lack of ground-truth validation for the intensity metrics is well-founded and is the primary reason for the major revision recommendation. The paper's central experimental claim (plan-view mitigates pore back for vertical pores) is sound and well-demonstrated, but the 'quantification' claim is not yet supported by the analysis as presented. The Sample 3 overexposure issue is a separate, fixable experimental problem that also weakens the generalisability claim. If the authors define a scalar metric, re-acquire the Sample 3 data, and add either a simulation or a more rigorous discussion of confounders, this would be a strong contribution suitable for Ultramicroscopy."},"author_rebuttal":null,"desk_editor":{"model":"glm-5.2","letter":"The main thing to know: this paper takes a known idea (align your milling direction with the long axis of pores to minimize shine-through) and actually implements it for porous GaN thin films, with a detailed practical protocol and an honest A/B comparison against the conventional cross-sectional setup on three samples with different pore morphologies. The plan-view approach clearly works for vertically aligned pores — the intensity histograms (Fig. 14) show a distinguishable low-intensity peak for plan-view vs. a single smeared peak for cross-sectional, and the line profiles (Fig. 15) show the expected anisotropy along the milling direction that disappears in the as-taken plan-view frames. That's a real methodological contribution, and the setup protocol is described in enough detail to be reproducible by someone willing to invest the effort. The authors also deserve credit for acknowledging where the method struggles — they openly concede that for Sample 3 (highly branched pores) the advantage is 'contentious' and 'marginal.' The stress-test concern about lack of ground truth is partially valid but slightly overplayed. The causal chain is physically sound: the pore back effect is a known electron-scattering phenomenon, the two setups use identical beam conditions, and the observed anisotropy along the milling direction but not perpendicular to it is exactly what the mechanism predicts. You don't need a Monte Carlo simulation to find that convincing at a basic level. That said, the 'formalisms' are really qualitative comparison tools. The histograms are compared by eye with no scalar metric or statistical test, and the line profiles are shown for individual pores without aggregation. A scalar anisotropy index or at least a population-level summary would strengthen the claim of 'quantitative assessment.' And the abstract's 'range of pore morphologies' is overstated — the evidence supports vertically aligned pores (Samples 1–2), and for branched pores (Sample 3) the plan-view dataset had overexposure issues and the authors themselves call the result marginal. The claim should be scoped accordingly. This is a solid methods paper for people doing FIB-SEM tomography on mesoporous thin films with anisotropic pores. It deserves a serious referee who can push the authors to tighten the generalization claim and add at least one quantitative scalar metric to the otherwise qualitative intensity analysis.","headline":"Useful rotated-geometry method for FIB-SEM tomography of porous GaN; metrics are qualitative, generalization claim overstated","tokens_in":25007,"tokens_out":559,"would_cite":false,"duration_ms":109105,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"glm-5.2","headline":"Rotating FIB tomography 90° cuts pore back effect in porous GaN","keywords":["FIB-SEM tomography","pore back effect","porous gallium nitride","electrochemical etching","voxel intensity distribution","mesoporous thin films","feature anisotropy","plan-view tomography"],"falsifier":"If the intensity-distribution histograms and feature-anisotropy line profiles showed no systematic difference between the two tomography orientations across all three samples, the claim that plan-view geometry mitigates the pore back effect would be unsupported.","tokens_in":24255,"feed_emoji":"🔬","tokens_out":1822,"duration_ms":109189,"temperature":0.7,"pith_summary":"The paper argues that the pore back effect — where electrons scatter from the back wall of a transparent pore and create spurious contrast in FIB-SEM tomography — can be both quantified and mitigated. The mitigation comes from rotating the entire tomography experiment 90 degrees so that the ion beam mills parallel to the long axis of vertically-propagating pores, maximizing the distance between the imaging plane and the back wall of each pore. The quantification comes from two new formalisms: comparing voxel intensity distributions to the ideal two-peak (GaN/void) histogram, where smearing of the low-intensity void peak signals the effect, and measuring asymmetry in line profiles extracted along versus perpendicular to the milling direction, where the pore back effect introduces a directional bias that should not exist. Across three samples with different pore morphologies, the plan-view approach consistently produced darker void voxels, sharper pore walls, and intensity distributions closer to the ideal binary form, without increasing total experiment time because fewer but larger frames are needed. The advantage is strongest for vertically aligned columnar pores and diminishes for highly branched or isotropic pore structures where no single orientation aligns with a dominant pore axis.","feed_headline":"Rotating FIB tomography 90° cuts pore back effect in porous GaN","feed_subtitle":"A simple rotation aligns the ion beam with pore axes and, with new intensity metrics, lets researchers measure the artifact they have been忽略","key_machinery":"The pore back effect is the central phenomenon: when imaging through transparent voids in SEM-based tomography, secondary and back-scattered electrons from the solid material forming the back wall of a pore are detected and contribute unwanted contrast to frames captured at shallower milling depths than the feature's true position, producing a directional smearing of features along the milling axis in the reconstructed tomograph.","core_discovery":"The plan-view tomography orientation — milling direction perpendicular to the film surface and thus parallel to the long axis of vertically-etched pores — significantly reduces the pore back effect compared to conventional cross-sectional tomography, and two voxel-intensity-based metrics (intensity distribution histograms and feature anisotropy line profiles) provide the first quantitative framework for assessing the severity of this imaging artifact in porous thin films.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Plan-view FIB tomography reduces pore back effect in porous GaN films","Sample rotation mitigates pore back effect in FIB tomography of GaN","Voxel intensity metrics quantify pore back effect in mesoporous GaN","90° rotation reduces pore back artifact in FIB tomography of porous GaN","Intensity-based metrics quantify pore back effect in FIB tomography"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The method's advantage depends on pores being predominantly vertically aligned — etching downward from the film surface. For highly branched or isotropic pore structures, the advantage shrinks because no orientation can align the milling direction with a dominant pore axis, and the paper's own most challenging sample (Sample 3) shows the weakest improvement.","fun_headline_variants_meta":{"raw":{"variants":["Plan-view FIB tomography reduces pore back effect in porous GaN films","Sample rotation mitigates pore back effect in FIB tomography of GaN","Voxel intensity metrics quantify pore back effect in mesoporous GaN","90° rotation reduces pore back artifact in FIB tomography of porous GaN","Intensity-based metrics quantify pore back effect in FIB tomography","Perpendicular milling reduces pore back effect in GaN FIB tomography","Plan-view orientation and voxel metrics quantify pore back effect in GaN","Milling along pore axes mitigates pore back effect in GaN tomography"]},"model":"glm-5.2","effort":"low","cost_usd":0.0,"raw_usage":{"total_tokens":2666,"prompt_tokens":563,"completion_tokens":2103,"prompt_tokens_details":null},"tokens_in":563,"tokens_out":2103,"duration_ms":65344,"temperature":1.0,"reasoning_tokens":2078,"cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-09T15:26:06.290125+00:00","model_set":{"reader":"glm-5.2"},"falsifier":"If the intensity-distribution histograms and feature-anisotropy line profiles showed no systematic difference between the two tomography orientations across all three samples, the claim that plan-view geometry mitigates the pore back effect would be unsupported.","supporting_citations":[],"review_version":1}