{"id":"63f14849-f48d-4aeb-a310-b5ce82b5b501","arxiv_id":"2608.06693","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Tilting the ion beam changes the optimal dose for activating visible defects in thick hBN flakes, but not in thin ones.","lead":"Ion beams are usually fired straight down at materials, but this paper shows that firing them at an angle changes how many light-emitting defects appear in hexagonal boron nitride. The angle sets the best ion dose for making bright emission, giving a new geometric knob for tuning optical materials.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The geometric-dilution mechanism in §II D is quantitatively inconsistent with the experimental fluences: ion cascades overlap by orders of magnitude, so lateral straggle cannot reduce the average defect density by the claimed factors.","rationale":"The reader's weakest_assumption is that integrated PL may be a biased proxy for emitter density because swelling, roughening, and milling alter collection efficiency. That is a legitimate concern, but it is unlikely by itself to produce a two-order-of-magnitude shift in the fluence optimum, and its directional effect is ambiguous. A more load-bearing problem is that the paper's quantitative mechanism—the 'geometric dilution' of the collision cascade—is internally inconsistent with the fluence range of the experiment. The per-ion areal-vacancy-density argument in Section II D assumes isolated cascades; however, at the optimal fluences the ion spacing is far smaller than the reported lateral straggle. A simple calculation of the coverage fraction Φ·A_XY shows that cascades overlap massively, at both the normal and oblique optima. Once overlapping, the average through-thickness vacancy density is set by Φ·N_v, which is nearly angle-independent according to the paper's own Table S2. The paper's depth-confinement numbers actually point in the opposite direction: the shorter range at oblique incidence would increase the volume density for fixed fluence. Thus the claimed factor-64 lateral dilution cannot explain the measured 100× optimal-fluence shift. This does not invalidate the empirical observation that the PL optimum changes with angle and thickness, but it removes the stated physical explanation. The verdict should remain conditional: the data may be correct, but the central mechanism must be rederived with overlap-aware simulations and quantitative treatment of sputtering before the claims about geometric control of defect activation can be accepted.","tokens_in":16274,"tokens_out":23879,"duration_ms":231128,"concrete_test":"Compute the coverage fraction Φ·A_XY using Table S2 parameters at each PL optimum: A_XY = π(ΔR_lat)²/cosθ, Φ = 2.5×10^14 (0°) and 2.5×10^16 (30°/60°) cm^-2. If Φ·A_XY > 1 (expected ≈70 at 0° and ≈7×10^5 at 60°), the per-ion dilution factors (13 and 64) do not apply; replace them with the overlap-corrected average density Φ·N_v and show it is angle-independent. Alternatively, run a multi-ion TRIM simulation at high fluence and compare the predicted vacancy-density profile with the single-cascade footprint prediction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central mechanistic claim (Section II D) is that oblique incidence dilutes the collision cascade laterally, reducing the areal defect density per ion by ~13× (30°) and ~64× (60°), and that this explains the ~100× shift of the PL optimum. This per-ion dilution argument treats cascade footprints as isolated. But at the PL-optimum fluences the mean ion spacing is 1/√Φ ≈ 0.63 nm at 2.5×10^14 cm^-2 and ≈0.063 nm at 2.5×10^16 cm^-2, far smaller than the reported lateral straggle (3–21 nm). The coverage fraction Φ·A_XY (A_XY = πΔR_lat²/cosθ) is ≈70 at the 0° optimum and ≈7×10^5 at the 60° optimum; even the normal-incidence optimum already has cascades overlapping by two orders of magnitude. In this regime the average through-thickness vacancy density is Φ·N_v, independent of ΔR_lat, and Table S2 gives N_v ≈ 505 for all angles. Spreading the cascade cannot lower the average damage density when ion impacts are packed more closely than the cascade width. The depth effect even points the wrong way: the 60° range (11.9 nm) is half the 0° range (23.8 nm), which would raise the volume density at fixed fluence. The claimed geometric dilution therefore does not operate at the experimental fluences, and the angle-dependent optimum must be caused by sputter-driven erosion (yield 22.8 vs 3.68 atoms/ion) or other unmodeled processes. The central explanation is quantitatively unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental study of visible photoluminescence (PL) activation in hexagonal boron nitride (hBN) flakes irradiated with 30 keV Xe+ ions at normal, 30°, and 60° incidence. For thick flakes (130–240 nm), the ion fluence that maximizes integrated visible PL shifts from ~2.5×10^14 ions/cm^2 at normal incidence to ~2.5–5×10^16 ions/cm^2 at oblique incidence, a shift of nearly two orders of magnitude; thin flakes (40 nm) show a largely angle-independent optimum near 10^14 ions/cm^2. TRIM/Iradina simulations show angle-dependent lateral cascade spreading and increased sputter yields, and AFM measurements identify swelling, thinning, and substrate-milling regimes. Spectrally resolved PL, time-resolved PL, and post-irradiation annealing are used to argue that oblique incidence preserves the same broad defect family while shifting its relative populations. The paper's central claim is that incidence angle is a geometric control parameter for defect activation in van der Waals materials.","tokens_in":16584,"tokens_out":10032,"duration_ms":106326,"significance":"The empirical dataset is a strength: PL maps across 22 doses, three incidence angles, and several flake thicknesses, combined with correlative AFM, provide a systematic picture that is not common in the hBN defect-engineering literature. The observation that incident angle can shift the optimal activation fluence by ~100× in thick flakes is novel and potentially useful for photonic integration, and the paper does not fit any parameter to the fluence-shift data. However, the mechanistic explanation—that lateral redistribution of the collision cascade dilutes the areal defect density per ion and thereby shifts the required fluence—is quantitatively not supported at the fluences used in the experiments. The paper also acknowledges that the calculated 13× dilution at 30° cannot explain the observed ~100× shift and appeals qualitatively to sputtering. Because the cascade-overlap argument is central to the abstract, introduction, and conclusions, the manuscript needs substantial revision before the mechanistic claim can be accepted.","major_comments":[{"comment":"The geometric-dilution mechanism is not operative at the experimental fluences. For uniform irradiation at fluence Φ, the expected depth-integrated areal density of vacancies is Φ·N_v, which is independent of the lateral cascade width; the paper's use of N_v/(πΔR_lat^2/cosθ) as a per-ion 'projected areal vacancy density' implicitly assumes isolated, non-overlapping cascades. At the normal-incidence PL optimum, Φ = 2.5×10^14 cm^-2 gives a nominal ion spacing of ~0.63 nm, already smaller than the reported ΔR_lat of 3.0 nm, and at the 60° optimum Φ = 2.5×10^16 cm^-2 the spacing is ~0.063 nm against ΔR_lat = 20.6 nm. The coverage fraction Φ·A_XY is ~70 at the 0° optimum and ~7×10^5 at the 60° optimum. In this regime the average damage density is set by Φ·N_v, with N_v ≈ 505 for all angles (Table S2), so lateral spreading cannot lower the mean through-thickness vacancy density by factors of 13 or 64. The central explanation for the fluence shift is therefore quantitatively unsupported; it should be replaced by a model that includes cascade overlap and possibly local-dose saturation, or the fluence shift should be presented as an empirical observation without attributing it to lateral redistribution.","section":"Section II.D, Table S2"},{"comment":"Even if one accepts the sputtering contribution, the paper does not quantitatively account for the 30° case. The computed lateral dilution factor is only ~13, the sputter-yield increase is 6.91/3.68 ≈ 1.9, and at the 30° optimum (2.5–5×10^16 ions/cm^2) the estimated milling depth is ~17–34 nm for a flake of 130 or 240 nm thickness. Combining these factors does not reach the observed ~100× shift in optimal fluence. The statement that 'the additional influence of material removal contributes' to the larger-than-predicted shift is qualitative; a quantitative model, or at least a clear scaling argument, is needed before the proposed two-mechanism explanation can be regarded as established.","section":"Section II.D, Figure 5(b)"},{"comment":"The use of integrated PL intensity as a proxy for activated emitter density is potentially confounded by the morphological changes documented in the same samples. The paper reports that the PL maximum coincides with the swelling fluence, and at oblique incidence the optimum lies in a regime with measurable sputter thinning; AFM shows milling depths up to ~140 nm at the highest fluences. Swelling, roughening, and thinning can change the collection efficiency, the emitting volume, and the local optical interference condition—the paper itself notes Fabry–Pérot interference in the thicker flake—so the apparent fluence shift could arise in part from morphology-dependent optical artifacts rather than from a true change in defect creation efficiency. I ask for a control or quantitative correction, for example PL normalized by remaining hBN volume, cross-section measurements, or a demonstration that the PL optimum is insensitive to these morphology changes at fixed defect density.","section":"Section II.A and Section II.E"}],"minor_comments":[{"comment":"The text refers to 'Figure 4(a)' and 'Figure 4(b)' when discussing the TRIM results; these should be 'Figure 5(a)' and 'Figure 5(b)'.","section":"Section II.D"},{"comment":"The caption contains a redundant sentence: 'The distribution of vacancy defect states in hBN is illustrated by their atomic lattice configurations relative to their emission wavelengths. Different defect states in hBN with respect to their emission wavelengths.' This should be reduced to a single clear sentence.","section":"Figure 1(b) caption"},{"comment":"The word 'irrdiated' should be 'irradiated'.","section":"Figure 7 caption"},{"comment":"The sentence 'The IRF was independently measured across the visible spectral range using a tunable, spectrally filtered supercontinuum source generated from the same excitation laser' is confusing because the excitation source is described as a Ti:sapphire laser, not a supercontinuum source; please clarify the relationship between the two sources.","section":"Methods (time-resolved PL)"},{"comment":"The statement that 'the effects of incidence angle and absolute fluence are therefore not fully separable' is important and well placed, but the same caveat should be applied to the annealing and TRPL comparisons, where the spectra are also acquired at geometry-specific fluences.","section":"Section II.B"}],"recommendation":"major_revision","confidential_remarks":"The empirical observation of an angle-dependent fluence shift is likely publishable, and the dataset is systematic. The main risk is overclaiming the mechanism. I would encourage the editor to require either an overlap-aware estimate of damage density or a clear reframing of the fluence shift as an empirical geometry effect with the mechanism left as a qualitative hypothesis. The paper's self-citation pattern is not problematic, and I see no circularity in the central measurement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The empirical core here is solid and worth having, but the proposed mechanism in Sec. II D doesn't survive contact with the reported fluences. The new observation is that in thick hBN flakes (130, 240 nm), the fluence that maximizes visible PL shifts by nearly two orders of magnitude when the Xe+ beam is tilted to 30-60 degrees, while thin flakes (35, 40 nm) show almost no angle dependence. That is a clean, systematic result: 22 doses per angle, three thicknesses, PL maps at four wavelengths, AFM on the same boxes, TRPL and annealing as consistency checks. The AFM identification of swelling, thinning, and substrate milling regimes is useful on its own. If I were processing hBN for quantum photonics, I would want this dataset in the literature.\n\nThe soft spot is the mechanism. The paper argues that oblique incidence spreads the collision cascade laterally, reducing the areal defect density per ion by ~13x at 30° and ~64x at 60°, and that this geometric dilution explains the ~100x fluence shift. But at the PL-optimum fluences, cascades overlap massively. The mean ion spacing at 2.5e14 cm^-2 is about 0.6 nm, far smaller than the reported lateral cascade dimensions (3-21 nm). Using the paper's own A_XY values, the coverage fraction Φ·A_XY is ~70 at the 0° optimum and even larger at the oblique optima. In that regime, the average through-thickness vacancy density is simply Φ times vacancies-per-ion, independent of lateral straggle. The 13x/64x factors apply to isolated cascades, not to the overlapping regime the experiment actually operates in. The depth effect even points the wrong way: the 60° range (11.9 nm) is half the 0° range (23.8 nm), which would raise the volume density at fixed fluence. So geometric dilution as stated cannot be the main driver. The sputter-yield difference (22.8 vs 3.68 atoms/ion) is the more plausible explanation, and the paper does invoke it as a secondary contribution, but it does not quantitatively separate the two. This is an addressable flaw: the empirics stand, but the mechanism section needs a rewrite.\n\nTwo smaller quibbles: the integrated PL per 2x2 µm box is the sole proxy for emitter density, and the AFM shows swelling and roughening over the same fluence range, which could modulate collection efficiency. And the thickness dependence rests on one flake per thickness in the main comparison. Neither is fatal, but both lower confidence in the quantitative claims.\n\nWho is this for? Experimentalists working on defect activation in hBN and other vdW materials. The angle-dependent fluence shift is a practical knob and the data are a useful reference. The mechanism overreach should be fixed in review, not desk-rejected. I'd send it to a serious referee.","headline":"Solid empirical dataset on angle-dependent defect activation in hBN, but the geometric dilution mechanism in Sec. II D is quantitatively inconsistent with the high-fluence regime.","tokens_in":17145,"tokens_out":7053,"would_cite":true,"duration_ms":63807,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["78.55.-m","79.20.Rf","61.80.Jh"],"model":"deepseek-v4-flash","headline":"Tilting the ion beam shifts the fluence that maximizes visible emission in thick hBN flakes by roughly two orders of magnitude, while thin flakes show no angle dependence.","keywords":["ion irradiation","hexagonal boron nitride","photoluminescence","defect engineering","ion incidence angle","collision cascade","sputtering","van der Waals materials"],"falsifier":"Measure the absolute density of activated emitters at the PL-optimum fluences for 0$^\\circ$ and 60$^\\circ$ on a thick flake by correlating single-defect counting or cross-sectional electron microscopy with the PL maps, and compare after normalizing by the retained hBN volume from AFM; if the angle-dependent optimum disappears under this normalization, the claimed 100-fold shift in activation fluence is a collection artifact of morphology rather than a change in defect-creation efficiency.","tokens_in":1959,"feed_emoji":"💡","tokens_out":2347,"duration_ms":80313,"temperature":0.7,"pith_summary":"This paper argues that the angle at which heavy ions strike hexagonal boron nitride is itself a processing control for activating visible-light emitters. In flakes thicker than the ion penetration depth, irradiating at 30–60 degrees from normal shifts the fluence that produces the brightest photoluminescence from about $2.5\\times10^{14}$ to $2.5$\\textendash$5\\times10^{16}$ ions/cm$^2$, a nearly 100-fold change, while in thin flakes the optimum fluence stays fixed. The authors trace this to two geometric effects: oblique impacts spread the collision cascade over a larger lateral area, lowering the areal defect density generated per ion, and they also sputter away material faster, truncating the window in which emitters can accumulate. Ion-trajectory simulations and atomic force microscopy support the picture, with the PL maximum coinciding with a fluence regime where the flake swells before milling takes over.","feed_headline":"Tilted ion beams shift hBN emitter dose 100-fold","feed_subtitle":"Oblique Xe+ irradiation raises the peak-emission fluence ~100x in thick flakes; thin flakes stay put.","key_machinery":"The central object is the projected collision cascade: the lateral spread and depth-projected range of Xe$^+$ ions in hBN, computed with a Monte Carlo ion-matter interaction code. The quantity that carries the argument is the through-thickness-integrated areal vacancy density, approximated as vacancies per ion divided by $\\pi(\\Delta R_{\\rm lat})^2/\\cos\\theta$, which falls by factors of about 13 at 30$^\\circ$ and 64 at 60$^\\circ$ relative to normal incidence; together with the simulated sputter yield, this sets both the fluence needed to reach a given local emitter density and the fluence at which milling removes it. The complementary experimental machinery is integrated photoluminescence per 2 $\\mu$m$\\times$2 $\\mu$m box as the activation readout and AFM Z-height as the morphology readout, which together identify swelling near $10^{14}$ ions/cm$^2$ as the regime where the PL maximum sits.","core_discovery":"The central claim is that ion-incidence geometry, not just ion species, energy, or fluence, controls the activation of visible emitters in hBN. For a 240 nm flake, oblique Xe$^+$ irradiation at 30$^\\circ$ and 60$^\\circ$ requires a fluence roughly two orders of magnitude higher than normal incidence to maximize visible photoluminescence, whereas a 40 nm flake shows an angle-independent optimum near $10^{14}$ ions/cm$^2$. The mechanism is geometric: tilting leaves the total vacancy yield per ion nearly unchanged (about 503–506 vacancies per ion) but compresses the cascade toward the surface and spreads it laterally, reducing the through-thickness-integrated areal defect density by factors of roughly 13 at 30$^\\circ$ and 64 at 60$^\\circ$; the simulated sputter yield also rises from 3.68 to 6.91 to 22.8 atoms per ion at 0$^\\circ$, 30$^\\circ$, and 60$^\\circ$, which bounds the upper end of the useful fluence window. Time-resolved photoluminescence shows the same biexponential lifetimes for normal and oblique incidence, so the angle redistributes a common family of defect states rather than creating a new emitter species, and annealing reveals an angle-dependent bias between cyan and green-yellow emission bands. The paper concludes that incidence angle is a mask-free, materials-level knob for programming defect density, depth profile, and spectral balance.","pith_inferences":["If the geometric dilution picture is correct, the same angle-dependent fluence shift should appear for other heavy ions of similar range, whereas light ions with deeper penetration would show a suppressed effect; this is a testable prediction the paper does not make explicitly.","The coincidence of the PL maximum with the swelling onset suggests that strain or lattice disorder, not just vacancy density, may participate in activating emission; a direct measurement of local strain versus emitter density could separate these contributions.","Since the lateral cascade spread is roughly isotropic, combining oblique irradiation with pre-patterned masks could create graded lateral emitter densities in a single exposure, a route the paper leaves implicit.","Extending the measurements to single-defect spectroscopy at fixed fluence below the optimum could test whether the angle shifts only emitter density or also changes the depth distribution of emitters available for Purcell enhancement in photonic cavities."],"forward_implications":["Incidence angle becomes a practical tuning knob for emitter density: oblique irradiation can dilute the areal defect density per ion without changing ion species or beam energy.","The useful defect-creation window is bounded by swelling and sputter-driven milling, so AFM height measurements can predict the fluence range that will produce bright visible emission on a given flake.","Flake thickness acts as a second control parameter: thin flakes lose the angle dependence because ions and milling reach the substrate, making the geometric dilution effect ineffective.","Post-irradiation annealing after oblique irradiation preferentially preserves green-yellow emission, offering a geometry-dependent way to bias the spectral composition of a common defect family.","Because the angle sets the depth and lateral extent of the damaged volume, oblique irradiation can help position emitters relative to planar photonic structures that require specific emitter depths."],"supporting_citations":[{"why":"Supplies the threshold displacement energies (B: 19.36 eV, N: 23.06 eV) used in the TRIM vacancy calculations.","marker":"[51, 52]"},{"why":"Establishes the precedent that oblique ion incidence controls defect generation in graphene, which the paper extends to optical defects in hBN.","marker":"[42]"},{"why":"Shows that oblique-angle ion bombardment improves implantation in graphene, supporting the geometric-control mechanism.","marker":"[44]"},{"why":"Provides the known angle dependence of sputter yields and erosion rates that the paper invokes to explain enhanced milling at oblique incidence.","marker":"[38]"},{"why":"Predicts from atomistic simulation that oblique carbon implantation enhances color-center yield in hBN, a precedent for geometry altering defect populations.","marker":"[50]"},{"why":"Quantifies how irradiation fluence controls vacancy-related center density, the baseline that the paper's PL-based activation readout builds on.","marker":"[22]"},{"why":"Identifies carbon as the source of visible emission in hBN, used to interpret the annealed green-yellow bands and the common defect family.","marker":"[28]"}],"fun_headline_variants":["Oblique ion angle shifts hBN emitter peak 100x","Tilting ion beam repositions hBN emitter peak by 100x","Ion angle, not just fluence, programs hBN visible emitters","Thick hBN needs 100x more tilted ions for max emission","Geometry governs hBN emitter activation under oblique ions"],"cache_read_input_tokens":19200,"weakest_assumption_plain":"The load-bearing premise is that the integrated photoluminescence from each irradiated box is a faithful measure of the density of activated emitters, unaffected by the swelling, roughening, and thinning that atomic force microscopy shows across the same fluence range.","fun_headline_variants_meta":{"raw":{"variants":["Oblique ion angle shifts hBN emitter peak 100x","Tilting ion beam repositions hBN emitter peak by 100x","Ion angle, not just fluence, programs hBN visible emitters","Thick hBN needs 100x more tilted ions for max emission","Geometry governs hBN emitter activation under oblique ions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000862,"raw_usage":{"total_tokens":3818,"prompt_tokens":1106,"completion_tokens":2712,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":722,"completion_tokens_details":{"reasoning_tokens":2620}},"tokens_in":722,"tokens_out":2712,"duration_ms":18343,"temperature":1.0,"reasoning_tokens":2620,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:19:35.027698+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the absolute density of activated emitters at the PL-optimum fluences for 0$^\\circ$ and 60$^\\circ$ on a thick flake by correlating single-defect counting or cross-sectional electron microscopy with the PL maps, and compare after normalizing by the retained hBN volume from AFM; if the angle-dependent optimum disappears under this normalization, the claimed 100-fold shift in activation fluence is a collection artifact of morphology rather than a change in defect-creation efficiency.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the precedent that oblique ion incidence controls defect generation in graphene, which the paper extends to optical defects in hBN."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that oblique-angle ion bombardment improves implantation in graphene, supporting the geometric-control mechanism."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the known angle dependence of sputter yields and erosion rates that the paper invokes to explain enhanced milling at oblique incidence."},{"cited_title":"Ren and Z","cited_arxiv_id":null,"evidence_quote":"Predicts from atomistic simulation that oblique carbon implantation enhances color-center yield in hBN, a precedent for geometry altering defect populations."},{"cited_title":"Carbone, I","cited_arxiv_id":null,"evidence_quote":"Quantifies how irradiation fluence controls vacancy-related center density, the baseline that the paper's PL-based activation readout builds on."},{"cited_title":"Mendelson, D","cited_arxiv_id":null,"evidence_quote":"Identifies carbon as the source of visible emission in hBN, used to interpret the annealed green-yellow bands and the common defect family."}],"review_version":1}