{"id":"949f7b54-a52b-46f7-82b2-0c8b4b0b131a","arxiv_id":"2608.10226","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Atomic-resolution imaging of NiO/Ga2O3 interfaces on three substrate orientations shows the (100) interface is the most abrupt and best-matched to calculated models.","lead":"Using atomic-resolution electron microscopy and computer models, this paper maps exactly how nickel oxide crystals connect to three different faces of gallium oxide crystals. It finds that the (100) face forms the cleanest, most orderly junction, which could guide the design of better power electronic devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'abrupt, low-defect' (100)-interface claim rests on interpreting weak extra columns as a step-edge projection artifact; the authors explicitly cannot rule out Ga interstitials, so the defect-density conclusion is underdetermined.","rationale":"The reader's verdict is CONDITIONAL and correctly identifies the weak extra-column contrast at the (100) interface as the key unverified assumption. I agree with that identification. The paper is otherwise internally consistent: the main lattice registry, XRD orientation relationships, and simulated STEM images of the primary interface support the crystallographic conclusions, and the authors are transparent about the (001) model's limitations and about the beam sensitivity that prevents EELS/EDS. The load-bearing concern is not that the interpretation is impossible, but that it is underdetermined: a forward simulation with an assumed step-edge geometry matches the contrast, but the alternative explanation (Ga interstitials) is explicitly noted to match the same sites and is not chemically excluded. Because the low-defect-density recommendation depends on which explanation is correct, the central claim should remain conditional pending chemical identification or an independent geometric test. I therefore recommend no change to the reader's verdict.","tokens_in":14238,"tokens_out":5372,"duration_ms":55377,"concrete_test":"Perform atomically resolved EELS/EDS spectrum imaging at the weak-column sites of the NiO/Ga2O3(100) interface, using low-dose, dose-fractionated acquisition to minimize beam damage. If the columns are Ni, the step-edge projection interpretation is supported; if Ga, the interface contains interstitials and the low-defect claim requires revision. As a complementary check, simulate the step-edge model with thickness fractions of 1/4 and 1/2 and compare the resulting column intensities quantitatively to the experimental images; if only the 1/3 fraction reproduces the observed contrast, the interpretation remains underdetermined. Additionally, prepare a specimen from the same sample along a second zone axis, since a step edge along [010] and isolated interstitials produce distinguishable projected patterns.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the NiO/Ga2O3(100) interface is nearly atomically abrupt and low in defect density depends on the interpretation of the weak 'extra' atomic columns in the top 1–2 Ga2O3 unit cells (Fig. 4(a)). The paper states that this contrast matches a common Ga interstitial site (Fig. S8) and that 'closer inspection reveals that it is also explained by a step edge' (Fig. 4(b,c)). The step-edge model is a forward multislice simulation with an assumed geometry—a NiO layer shifted down one unit cell over 1/3 of the specimen thickness—and no reported sensitivity analysis. It demonstrates that projection overlap can produce weak columns, but it does not establish that the actual interface has this geometry. The authors explicitly note that Ni and Ga cannot be distinguished at these sites without atomically resolved EELS or EDS. Since the low-defect-density recommendation is a central deliverable, and since real Ga interstitials would directly contradict that recommendation, the argument is underdetermined at exactly the point that separates 'abrupt and clean' from 'abrupt but defective.'","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an atomic-scale study of NiO/Ga2O3 interfaces formed by pulsed-laser deposition on (100), (-201), and (001) oriented β-Ga2O3 substrates. Using aberration-corrected HAADF-STEM, the authors identify epitaxial orientation relationships for each substrate orientation and compare the experimental images with interface models generated by a Lennard-Jones structure-matching algorithm followed by DFT relaxation, along with multislice STEM image simulations. For NiO/Ga2O3(100) they report a nearly atomically abrupt interface with NiO(001) || Ga2O3(100) and in-plane NiO[110] || Ga2O3[010], and they interpret weak 'extra' column contrast at the interface as a step-edge projection artifact rather than a Ga interstitial. The (-201) and (001) interfaces are described as more complex, with additional atomic columns and reconstructed layers that are only partially captured by the calculated models. The paper concludes that (100)-oriented Ga2O3 is a promising substrate for high-quality, low-defect-density NiO/Ga2O3 heterojunctions and emphasizes the importance of accounting for 3D-to-2D projection effects in interpreting such interfaces.","tokens_in":14499,"tokens_out":6946,"duration_ms":73206,"significance":"If the central claim is correct, the paper provides a concrete atomic-registry model for a technologically important heterojunction and gives a usable rationale for preferring (100)-oriented Ga2O3 substrates. The study has clear strengths: the (100) and (-201) orientation relationships emerge from an independent structure search rather than being read off the images; the relationships are corroborated by XRD; the multislice simulations are matched to the experimental imaging parameters; and the discussion of projection artifacts, including the step-edge model, is a valuable methodological contribution. The explicit negative result that the as-deposited interfaces do not show NiGa2O4 or γ-Ga2O3 spinel contrast is also useful for the community. However, the load-bearing conclusion that the (100) interface is low-defect-density relies on distinguishing a benign step-edge projection from real Ga interstitials, and the manuscript does not provide the chemical or quantitative evidence needed to make that distinction. The paper's comparative qualitative assessment of interface complexity is more robust than the quantitative 'low defect density' claim.","major_comments":[{"comment":"The central conclusion that the NiO/Ga2O3(100) interface is 'nearly atomically abrupt' and has 'limited defect density' hinges on interpreting the weak extra column contrast in the top 1-2 Ga2O3 unit cells as a step-edge projection artifact rather than as Ga interstitials. The text states that this contrast matches a common Ga interstitial site (Figs. S8a-b) and that it is 'also explained by' a step-edge model in which one-third of the NiO layer is shifted down one unit cell (Fig. 4b-c). The authors explicitly note that Ni and Ga cannot be distinguished without atomically resolved EELS or EDS, and no such data are provided. No quantitative sensitivity analysis is given: the shifted fraction is fixed at 1/3, no simulation of the competing Ga-interstitial configuration is presented, and the occurrence rate of this contrast is not correlated with specimen thickness. Because real Ga interstitials would directly contradict the low-defect-density recommendation, the evidence is underdetermined. The authors should either provide a chemical signature via EELS/EDS, present quantitative simulations and statistics that discriminate between the two interpretations, or soften the conclusion to 'abrupt registry with unresolved point-defect-like contrast'.","section":"Fig. 4(a-c) and 'Projection effects' section"},{"comment":"The main text states that the calculated interface models were obtained by the structure-matching algorithm 'followed by density functional theory (DFT) structure relaxation,' but Supplementary Section IV describes only the LJ-based surface generation, the p2ptrans structure matching, and the chemical-potential values. No DFT functional, pseudopotentials, plane-wave cutoff, k-point sampling, or convergence criteria are reported, and the relaxed interfacial atomic structures are not given. This omissions make it impossible to reproduce or independently assess the calculated (100) interface model that is central to the claimed 'excellent agreement' with experiment. Please add the missing computational details, or explicitly state that the models were not DFT-relaxed and discuss how that affects the model-experiment comparison.","section":"Supplementary Section IV and 'Interface structure modeling'"},{"comment":"The NiO(1-11)/Ga2O3(101) model for the (001) interface was selected after measuring the experimental mistilt between the NiO growth plane and the Ga2O3(001) normal, and the authors acknowledge that the couple atomic layers at the interface are not well captured by this model. This post-hoc selection should be explicitly labeled as a candidate relationship rather than a predicted interface structure. The XRD data in Figs. S1(c)-(d) are consistent with the (1-11)/(101) alignment, but they constitute a targeted confirmation of a hypothesis derived from the STEM images, not an independent prediction. The manuscript would be clearer if this distinction were stated, and if the text noted that the (001) interface structure remains largely unresolved.","section":"Fig. 3(e-f) and 'NiO/Ga2O3(001)' section"}],"minor_comments":[{"comment":"There are typographical errors: 'candidat' should be 'candidate' in the abstract, and 'It's pseudocubic' should be 'Its pseudocubic' in the introduction.","section":"Abstract and Introduction"},{"comment":"In the Supplementary Material reference list, entry 3 reads 'F. Therrien, P. Graf, and V. Stevanović, .' with the article title and journal missing. The full citation should be provided.","section":"Supplementary References"},{"comment":"The overlays of the calculated models and simulated images on the experimental HAADF-STEM images would be easier to evaluate if the simulated-image overlay were shown with a color or intensity scale that is more distinguishable from the experimental image; the current presentation makes the claimed 'excellent agreement' harder for the reader to verify independently.","section":"Fig. 3 and Fig. 4 overlays"},{"comment":"The comparison of the hexagonal motifs in the (-201) and (001) images with the γ-Ga2O3/spinel models is qualitative. A quantitative measure such as the spacing and orientation of the hexagon motifs, or a cross-correlation between the experimental and simulated images, would strengthen the negative conclusion that no NiGa2O4 interlayer has formed in the as-deposited samples.","section":"Fig. 4(d-i)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and reports a useful comparative study with several strong methodological elements. The main issue is the underdetermination of the 'low defect density' conclusion for the (100) interface, as detailed in Major Comment 1; this is fixable either by additional experiments or by appropriately qualifying the claim. The missing DFT details in the Supplementary Material (Major Comment 2) are also readily remedied. I do not see grounds for rejection, but the load-bearing interpretation needs to be supported or circumscribed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a careful experimental STEM paper that provides the first atomic-scale structural models for NiO/Ga2O3 interfaces on (100), (-201), and (001) substrates. The (100) structural determination is probably right, but the 'low defect density' claim outruns the evidence.\n\nWhat's new is genuinely useful: direct atomic-resolution imaging of all three interfaces, interface models from an independent structure search, and a clear demonstration that certain STEM contrast features are projection artifacts rather than real defects. The identification of NiO(10 10 3) on (001) is a nice observational result. The specimen preparation is meticulous—inert PFIB with backside thinning, avoiding Ga implantation—and the multislice simulations match the experimental images to a convincing degree.\n\nCredit where due: the (100) orientation relationship and atomic registry come out of a Lennard-Jones structure search that did not use the experimental images as input, and XRD independently corroborates the orientation. That is not circular. The step-edge model in Fig. 4 is a plausible explanation for the extra weak columns, and the authors are honest about not being able to distinguish Ni from Ga at those sites without atomically resolved EELS or EDS.\n\nThe soft spot is exactly where the stress-test note lands. The recommendation that (100) gives 'low defect density' interfaces rests on interpreting those weak columns as a projection artifact from a step edge. The simulation shows that such a geometry can produce this contrast, but it does not prove the actual specimen has that geometry—the 1/3-thickness assumption is not tested against alternatives, and no sensitivity analysis is reported. If those columns were real Ga interstitials, the defect-density conclusion would reverse. The authors' caveat is appropriate, but the abstract and conclusion still assert the stronger claim.\n\nSecondary issues: the (001) model is partly post hoc, chosen after measuring the experimental mistilt, and explicitly fails to capture the interface layers. Interface abruptness is assessed qualitatively, without quantitative roughness or defect-density metrics. These are real but minor compared to the interstitial ambiguity.\n\nBottom line: the paper deserves a serious referee. The central (100) structural result is well supported; the interpretation of the extra columns needs either supporting spectroscopy or a more conditional wording. A good reviewer should push for atomically resolved EELS/EDS on the (100) interface, or at least a revision that explicitly flags the defect-density recommendation as contingent on the step-edge interpretation.\n\nWho gets value: researchers working on NiO/Ga2O3 power devices, and the broader STEM community dealing with projection artifacts at complex oxide interfaces. I'd bring it to a reading group and would cite the (100) model were I working in this area.\n\nRecommendation: accept for peer review, expect revision.","headline":"Solid atomic-scale interface study; the (100) structural determination is well supported, but the 'low defect density' claim rests on an unverified reading of weak STEM contrast.","tokens_in":15034,"tokens_out":2609,"would_cite":true,"duration_ms":25307,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Nearly atom-perfect NiO/Ga2O3 interface found on the (100) face","keywords":["NiO/Ga2O3 heterojunctions","beta-Ga2O3","interface structure","HAADF-STEM","scanning transmission electron microscopy","epitaxial registry","projection artifacts","power electronics"],"falsifier":"Atomically resolved EELS or EDS maps of the top Ga2O3 unit cells at the (100) interface would settle the matter: if the extra columns are Ga, they are interstitials and the interface is not truly abrupt; if they are Ni, the step-edge projection interpretation stands.","tokens_in":1944,"feed_emoji":"🔬","tokens_out":2101,"duration_ms":61495,"temperature":0.7,"pith_summary":"This paper uses atomic-resolution electron microscopy, interface modeling, and image simulations to establish how NiO bonds to Ga2O3 for three different substrate orientations. The central claim is that the NiO/Ga2O3(100) interface is nearly atomically abrupt, with a clear crystallographic registry that matches a calculated interface model, while the (-201) and (001) interfaces are more complex, less abrupt, and more variable. If true, this matters for power electronics because interface quality, disorder, and defects directly affect band alignment, carrier transport, and device stability in NiO/Ga2O3 heterojunctions. The paper also shows that ordinary STEM projection effects can create contrast that looks like interstitials or spinel interlayer phases, so those apparent features must be interpreted with care.","feed_headline":"Nearly atom-perfect NiO/Ga2O3 interface found on the (100) face","feed_subtitle":"Atomic-resolution images show the (100) face forms a crisp junction, while other orientations grow messy interfacial layers.","key_machinery":"The central machinery is a three-way comparison loop: atomically resolved HAADF-STEM images, calculated interface models, and multislice-simulated STEM images from those models. Interface models are generated by a structure-matching algorithm that minimizes a Lennard-Jones energy over candidate NiO surface orientations on a fixed Ga2O3 surface, followed by density functional theory relaxation. The decisive element is the crystallographic registry—for (100), NiO(001) || Ga2O3(100) with in-plane NiO[110] || Ga2O3[010]—because matching that registry simultaneously in the model, the simulation, and the experimental image is what licenses calling the interface nearly atomically abrupt. The same simulation loop is used to test competing interpretations: a step-edge projection model reproduces the extra column contrast, while spinel and gamma-Ga2O3 models fail to reproduce the hexagonal motifs observed at the other two interfaces.","core_discovery":"The paper directly visualizes the atomic structure of NiO/Ga2O3 interfaces grown on (100), (-201), and (001) oriented Ga2O3 substrates and identifies the epitaxial relationships in each case. The central finding is that the NiO/Ga2O3(100) interface is nearly atomically abrupt and consistent across roughly half a micron of interface, with NiO(001) || Ga2O3(100) and in-plane NiO[110] || Ga2O3[010], and that this registry matches a DFT-relaxed interface model and the simulated STEM image derived from it. By contrast, the NiO/Ga2O3(-201) interface is less abrupt, with a corrugated Ga2O3 surface that produces extra atomic columns not captured by the model, and the NiO/Ga2O3(001) interface is the most complex, with NiO growing in a near-(331) orientation approximated as (10 10 3) and nucleating through an epitaxial relationship between Ga2O3(101) and NiO(1-11) rather than through the substrate surface plane itself. The paper further argues that weak extra column contrast near the (100) interface and hexagonal motifs near the (-201) and (001) interfaces are projection effects from substrate step edges and overlapping lattices, not Ga interstitials or NiGa2O4/gamma-Ga2O3 interlayer phases.","pith_inferences":["Inference: If the (100) interface is as defect-poor as claimed, vertical NiO/Ga2O3(100) diodes should show measurably lower leakage, fewer trap-related deep levels, and more stable breakdown behavior than comparable (001) devices; this is a testable electrical prediction the paper does not make.","Inference: Because NiO on (001) nucleates through the tilted Ga2O3(101) plane, deliberately miscut (001) substrates that expose (101)-like facets might template a more ordered NiO film than nominally flat (001) surfaces.","Inference: The step-edge projection interpretation implies that the intensity of the extra (100)-interface columns should scale with lamella thickness and local step density; imaging the same interface at several thicknesses could quantify how many steps are actually present.","Inference: The same projection-vs-defect caution likely applies to other rock-salt-on-monoclinic-oxide junctions, where spinel-like hexagonal motifs may appear in HAADF-STEM images without any spinel phase being present."],"forward_implications":["If the sharp-registry picture is right, (100)-oriented Ga2O3 substrates should be the preferred platform for low-defect-density NiO/Ga2O3 power devices.","The absence of NiGa2O4 or gamma-Ga2O3 interlayer phases in as-deposited films on all three orientations implies that high-temperature interlayer formation studies must be interpreted with projection effects in mind.","The (001) interface's intrinsic disorder and strain, tied to growth through a non-surface-plane Ga2O3(101)/NiO(1-11) relationship, suggests that devices on (001) may carry more interface traps and strain relaxation defects.","The sharp registry on (100) presumably leaves fewer nucleation sites for a NiGa2O4 interlayer during high-temperature device operation, suggesting better long-term interface stability.","For dissimilar heterointerfaces generally, extra atomic column contrast and phase-like motifs should be checked against thickness and projection artifacts before being assigned to point defects or new phases."],"supporting_citations":[{"why":"Documents the NiGa2O4 spinel interlayer that forms during high-temperature cycling of NiO/Ga2O3(001), the interlayer-phase concern this paper addresses.","marker":"[7]"},{"why":"Prior XRD and electron-diffraction reports of NiO crystal orientations on monoclinic Ga2O3 that the STEM observations confirm.","marker":"[9,10]"},{"why":"Establishes the epitaxial relationship of NiO on Ga2O3 and the pseudocubic oxygen-sublattice rationale used here.","marker":"[11]"},{"why":"Supplies the atom-to-atom structure-matching algorithm used to generate candidate NiO/Ga2O3 interface models.","marker":"[17]"},{"why":"Provides the multislice method and abTEM implementation used to simulate HAADF-STEM images from the interface models.","marker":"[18-20]"},{"why":"Maps known Ga interstitial sites in beta-Ga2O3 against the extra atomic columns to test the interstitial interpretation.","marker":"[21,22]"},{"why":"Documents gamma-Ga2O3 inclusions with a defective spinel structure, the structural motif compared against the hexagonal interface contrast.","marker":"[27]"}],"fun_headline_variants":["Nearly atom-perfect NiO/Ga2O3 junction on (100) face","Crisp interface on Ga2O3(100) helps NiO film grow cleanly","Why NiO/Ga2O3(100) gives the sharpest interface","STEM shows near-perfect NiO/Ga2O3(100) interface","Ga2O3(100) yields atom-abrupt NiO interface"],"cache_read_input_tokens":17152,"weakest_assumption_plain":"The sharp-interface conclusion for Ga2O3(100) rests on interpreting the weak extra atomic columns at the top of the substrate as a projection artifact from a step edge rather than as real Ga interstitials, a distinction the images alone cannot settle without atomically resolved chemical mapping.","fun_headline_variants_meta":{"raw":{"variants":["Nearly atom-perfect NiO/Ga2O3 junction on (100) face","Crisp interface on Ga2O3(100) helps NiO film grow cleanly","Why NiO/Ga2O3(100) gives the sharpest interface","STEM shows near-perfect NiO/Ga2O3(100) interface","Ga2O3(100) yields atom-abrupt NiO interface"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000311,"raw_usage":{"total_tokens":1893,"prompt_tokens":1188,"completion_tokens":705,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":804,"completion_tokens_details":{"reasoning_tokens":597}},"tokens_in":804,"tokens_out":705,"duration_ms":5817,"temperature":1.0,"reasoning_tokens":597,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:10:35.694528+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Atomically resolved EELS or EDS maps of the top Ga2O3 unit cells at the (100) interface would settle the matter: if the extra columns are Ga, they are interstitials and the interface is not truly abrupt; if they are Ni, the step-edge projection interpretation stands.","supporting_citations":[],"review_version":1}