{"id":"861a9c78-be84-403d-a771-e1e033981681","arxiv_id":"2607.10378","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"MOCVD β-Ga2O3 on (001) substrates grows predominantly (−401)-oriented, imposing ~+4.1% tensile strain along [104] that drives cracking above ~1.8 µm.","lead":"MOCVD β-Ga2O3 films on the cheapest large (001) wafers grow as (−401)-oriented layers, not (001), creating ~4% tensile mismatch that cracks films above ~1.8 µm. This crystallographic mismatch, not thermal stress, is the main barrier to thick vertical power-device drift layers on industry-standard substrates.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified to the central crystallographic-mismatch claim.","rationale":"The Reader correctly isolates the reconstruction hypothesis as the softest point while recognizing that the main experimental chain (orientation + anisotropic mismatch + crack orientation/thickness) stands on solid multi-technique data. That assessment matches a careful re-read of the full manuscript, including Supporting Information figures S1–S3. Because the load-bearing claim does not depend on the unconfirmed facet indices, no verdict adjustment is warranted. The suggested TEM check is the natural next experiment already flagged by the authors as beyond scope; it would tighten the nucleation picture without threatening the crystallographic origin of cracking.","tokens_in":14017,"tokens_out":558,"duration_ms":7742,"concrete_test":"Acquire a single cross-sectional TEM or STEM image (or selected-area diffraction) across the epilayer–substrate interface of the thinnest (0.3 µm) sample; if the interface shows continuous (−401) lattice planes matching the bulk XRD orientation (rather than a thin (001) interlayer or mixed domains), the central claim is further locked; any other result would only refine the nucleation story, not reverse the mismatch-driven cracking conclusion.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's strongest claim is that MOCVD epilayers on (001) are predominantly (−401)-oriented from the earliest stages, with the in-plane relation [010]epi ∥ [010]sub and [104]epi ∥ [100]sub imposing a theoretical tensile strain ε[104] ≈ +4.1% that drives channeling cracks parallel to [010]epi above ~1.8 µm. This is directly supported by the thickness series (0.3–3.5 µm), HRXRD ω–2θ (emergence and intensity growth of −401/−802/−1203 with no broad 002 epilayer peak), φ-scans establishing the single-domain in-plane alignment, rocking-curve trends, AFM striation orientation, and optical microscopy of crack direction and critical thickness. The lattice-mismatch arithmetic (L[104] = 23.464 Å vs 2 × L[100] = 24.42 Å) is standard and the progressive −401 peak shift to lower 2θ is consistent with progressive tensile relaxation. The reconstruction-as-nucleation hypothesis (Fig. 7) is weaker and is already labeled as such by the authors and the Reader; it is not required for the cracking-origin claim to hold. No internal inconsistency or missing control that would overturn the orientation–mismatch–crack chain is present.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"This manuscript reports a systematic thickness series (0.3–3.5 µm) of MOCVD β-Ga₂O₃ epilayers grown at ~3.5 µm h⁻¹ on (001) native substrates. AFM shows striated morphology aligned with [010]sub and increasing RMS roughness beyond coalescence; optical microscopy shows channeling cracks parallel to [010]epi above ~1.8 µm. HRXRD ω–2θ scans establish that the epilayers are predominantly (−401)-oriented from the earliest stages (with −401/−802/−1203 intensity growing and no broad 002 epilayer peak), while φ-scans of the −201 reflection establish the single-domain in-plane relation [010]epi ∥ [010]sub and [104]epi ∥ [100]sub. Rocking-curve FWHMs indicate higher twist than tilt, both decreasing with thickness. The authors calculate a geometric tensile mismatch ε[104] ≈ +4.1% and argue that this anisotropic strain drives the observed cracking; they further hypothesize that oxygen-rich annealing produces a faceted (001) surface that templates (−401) growth.","tokens_in":14347,"tokens_out":1032,"duration_ms":13072,"significance":"If the orientation–mismatch–cracking chain holds, the work supplies a concrete crystallographic explanation for a known industrial bottleneck: thick MOCVD layers on the largest, lowest-cost (001) β-Ga₂O₃ wafers crack and roughen. That explanation is actionable—it identifies the epitaxial relationship itself, rather than generic thermal stress or growth-rate effects, as the limiting factor—and therefore guides nucleation strategies (temperature/ambient ramps, low-T nucleation layers) that could enable device-relevant thick (001) films without AlGaO buffers that impede vertical transport. Strengths include a well-controlled thickness series, mutually consistent AFM/optical/HRXRD datasets, and a transparent geometric strain calculation using standard lattice periodicities. The reconstruction hypothesis is weaker and is already labeled as such, but it is not required for the central claim.","major_comments":[{"comment":"The central claim (orientation + anisotropic mismatch → cracking) is well supported by the thickness series, ω–2θ, φ-scans, rocking curves, AFM striation orientation, and optical crack direction/critical thickness. No load-bearing inconsistency is present. The only material gap is the absence of cross-sectional TEM/STEM of the interface (Discussion / Fig. 7), which would confirm facet indices and the nucleation pathway; the authors already present the reconstruction as a hypothesis, so this is not required to accept the cracking-origin claim, but a brief statement of what TEM would (and would not) change would strengthen the Discussion.","section":null}],"minor_comments":[{"comment":"Abstract and Results: the phrase “in theory, a maximum tensile in-plane strain of approximately +4.1%” is slightly ambiguous; state explicitly that this is the geometric (unrelaxed) mismatch and that the actual elastic strain is partially relaxed by defects and cracks.","section":null},{"comment":"Fig. 4(a) and associated text: the progressive −401 peak shift to lower 2θ is used as evidence of tensile relaxation; a short quantitative estimate of Δd/d (or residual strain) from the measured 2θ values would make the argument more precise.","section":null},{"comment":"Fig. 5: the anomalous smaller FWHM of the −401 reflection measured coplanar with [104]epi for the 0.3 µm sample is noted but only briefly discussed; a sentence linking this to incomplete defect-mediated relaxation before cracking would help the reader.","section":null},{"comment":"Supporting Information Fig. S2 is useful; consider moving a condensed version of the “no broad 002 epilayer peak” argument into the main text so that the single-domain (−401) conclusion is self-contained.","section":null},{"comment":"Materials and Methods: growth rates are taken from Si-doped calibration runs; a short note that undoped layers under the same conditions give the same rate (or that SIMS marker runs were co-loaded) would remove any residual ambiguity.","section":null},{"comment":"References: a few recent (001) MOCVD morphology papers are cited; ensuring the most recent HVPE vs MOCVD comparison reviews are included would help place the work for non-specialists.","section":null}],"recommendation":"accept","confidential_remarks":"The manuscript is a solid, industrially relevant materials-science contribution. The reconstruction hypothesis is the weakest link but is already framed as such and is not load-bearing for the cracking claim. I see no reason to delay publication for TEM; that can be a natural follow-up. Fit for a materials/condensed-matter journal is good."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The useful result here is straightforward: MOCVD epilayers on (001) β-Ga2O3 are not crystallographically continuous (001) films. From the thinnest layers they grow as single-domain (−401), with [010]epi || [010]sub and [104]epi || [100]sub. That registry is lattice-matched along [010] but imposes a geometric tensile mismatch of ~4.1% along [104], which orients and drives the channeling cracks that appear above ~1.8 µm. Prior papers already saw −401 peaks and cracking; this one closes the loop with a thickness series, φ-scans that pin the in-plane relation, rocking-curve trends, AFM striation direction, and optical crack orientation, all mutually consistent.\n\nWhat they do well is the experimental chain. ω–2θ shows −401/−802/−1203 growing with thickness and no broad 002 epilayer peak; the progressive −401 shift to lower 2θ tracks tensile relaxation; cracks run parallel to [010]epi exactly as the mismatch predicts. The arithmetic uses standard lattice periodicities, not fitted parameters. Growth is in an industry-scale CCS reactor at a usable rate (~3.5 µm/h), so the observation is relevant rather than niche.\n\nThe soft spot is the nucleation story. They hypothesize that oxygen-rich annealing facets the (001) surface and templates (−401). That is labeled as a hypothesis, rests on AFM of the annealed control plus surface-energy literature and an MBE analogy, and lacks TEM or in-situ diffraction of the interface. It is not required for the cracking-origin claim to stand. Missing TEM of defects and residual strain is an ordinary limitation for this class of paper, not a hole that sinks the main argument.\n\nThis is for people growing vertical Ga2O3 power devices or developing MOCVD processes on the large-diameter (001) orientation. It tells them the crystallographic bottleneck and why AlGaO buffers or nucleation tricks matter. I would send it to peer review; the data support the central claim cleanly enough that referees can focus on the open mechanism questions rather than basic soundness.","headline":"Solid crystallographic explanation for why thick MOCVD Ga2O3 on (001) cracks: the film is actually (−401) with ~4.1% tensile mismatch along one in-plane axis.","tokens_in":14989,"tokens_out":549,"would_cite":true,"duration_ms":7675,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Thick MOCVD beta-Ga2O3 films on (001) substrates grow as (-401) layers and crack from ~1.8 um because of a ~4.1% tensile mismatch along [104].","keywords":["beta-Ga2O3","MOCVD","homoepitaxy","(001) substrates","(-401) orientation","cracking","lattice mismatch","surface reconstruction"],"falsifier":"Cross-sectional TEM or STEM of the earliest stages of growth (or of the annealed-only control) that either identifies low-index facets matching the (-401)/(001) relation or shows continuous (001) registry would confirm or refute the proposed selection mechanism.","tokens_in":14918,"feed_emoji":"🔬","tokens_out":730,"duration_ms":8579,"temperature":0.7,"pith_summary":"Thick, crack-free gallium oxide layers grown by industry-standard MOCVD are needed for fully vertical power devices, yet films on the largest and cheapest (001) native substrates roughen and crack as thickness increases. This paper shows that the films never grow as true (001) homoepitaxy: from the earliest stages they adopt a predominantly (-401) orientation locked to the substrate by the in-plane relation [010]epi parallel to [010]sub and [104]epi parallel to [100]sub. That alignment is lattice-matched along [010] but imposes a theoretical tensile strain of about +4.1% along [104], which drives channeling cracks parallel to [010] once the film exceeds roughly 1.8 micrometres. The authors link the unexpected orientation to oxygen-driven faceting of the (001) surface that occurs during the routine pre-growth anneal. The result explains why previous thick MOCVD layers on (001) cracked and points to nucleation strategies that could stabilise true (001) growth for device-relevant thicknesses.","feed_headline":"Why thick Ga2O3 films on (001) crack: they grow as (-401)","feed_subtitle":"A 4.1% tensile mismatch along [104] forces channeling cracks above ~1.8 micrometres","key_machinery":"The crystallographic orientation relationship between the (-401) epilayer and (001) substrate: [010]epi || [010]sub and [104]epi || [100]sub. It converts an intended homoepitaxial growth into a lattice-mismatched heteroepitaxy whose anisotropic strain drives cracking.","core_discovery":"Despite growth on nominally (001) substrates under conventional oxygen-rich MOCVD conditions, the epilayers adopt a predominantly (-401)-oriented structure from the earliest stages. The resulting epitaxial alignment [010]epi || [010]sub and [104]epi || [100]sub produces zero lattice mismatch along [010] but a theoretical tensile in-plane strain of approximately +4.1% along [104], which relaxes by forming channeling cracks parallel to [010]epi once thickness exceeds ~1.8 um.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Thick Ga2O3 cracks as (-401) not (001) from 4.1% strain","(-401) growth on (001) Ga2O3 drives cracks above 1.8 um","Epilayers adopt (-401) orientation, imposing tensile cracks","Why Ga2O3 films crack: (-401) alignment with 4.1% mismatch","(-401) epi on (001) substrates forces [104] channeling cracks"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The claim that oxygen-annealed faceting of the (001) surface is what selects the (-401) orientation rests on AFM topography and surface-energy arguments; the paper does not show the actual nucleation interface or facet indices by cross-sectional imaging.","fun_headline_variants_meta":{"raw":{"variants":["Thick Ga2O3 cracks as (-401) not (001) from 4.1% strain","(-401) growth on (001) Ga2O3 drives cracks above 1.8 um","Epilayers adopt (-401) orientation, imposing tensile cracks","Why Ga2O3 films crack: (-401) alignment with 4.1% mismatch","(-401) epi on (001) substrates forces [104] channeling cracks"]},"model":"grok-4.5","effort":"low","cost_usd":0.0066,"raw_usage":{"total_tokens":1786,"prompt_tokens":933,"num_sources_used":0,"completion_tokens":97,"cost_in_usd_ticks":66000000,"prompt_tokens_details":{"text_tokens":933,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":756,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":933,"tokens_out":97,"duration_ms":9842,"temperature":1.0,"reasoning_tokens":756,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T12:10:49.291985+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Cross-sectional TEM or STEM of the earliest stages of growth (or of the annealed-only control) that either identifies low-index facets matching the (-401)/(001) relation or shows continuous (001) registry would confirm or refute the proposed selection mechanism.","supporting_citations":[],"review_version":1}