{"id":"3ec8441f-4e4e-46a6-ad0f-57c8324caedc","arxiv_id":"2411.13658","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Direct x-ray measurements show that capping-layer strain in monolayer MoS2 depends mainly on seed-layer adhesion and geometry, with stressed metal stacks imparting up to ~2% tensile strain.","lead":"This paper uses x-ray diffraction to directly measure strain in monolayer MoS2 caused by capping layers and metal contact patterns. It finds that seed-layer adhesion and device geometry, rather than oxide thickness or deposition temperature, control the strain, and reports tensile strains up to about 2% from stressed metal stacks.","discovery_kind":"extension","skeptic_critique":null,"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports grazing-incidence x-ray diffraction (GIXRD) measurements of the in-plane (100) lattice spacing of monolayer MoS2 on SiO2/Si and extracts strain directly from diffraction peak positions via Bragg's law. It presents three main results: (i) ALD Al2O3 capping (with a thin Al seed layer) imparts roughly 0.1% compressive strain that is attributed to the seed layer and to substrate adhesion rather than to ALD temperature or oxide thickness; (ii) transistor-like patterned metal lines capped with AlOx impart about six times more strain (about -0.44%) than blanket AlOx films; and (iii) a blanket Au/Ti/Ni metal stack on transferred MoS2 imparts tensile strain of roughly 2 to 2.5%, claimed to be among the largest values reported on a rigid substrate. All 2θ values, d-spacings, and strain values are tabulated in Tables S1-S6, so the Bragg conversion is transparent and independently checkable; the strain extraction itself involves no fitted model or prediction loop, and the only assignment-dependent step is the identification of the diffraction peaks as arising from 1H MoS2.","tokens_in":10745,"tokens_out":15765,"duration_ms":144358,"significance":"Direct x-ray measurement of strain in monolayer TMDs is a genuine methodological strength in a field where strain is usually inferred from Raman or photoluminescence with model-dependent calibrations. The tabulated peak positions permit independent verification of the strain arithmetic, and the negative results (ALD temperature and thickness independence) plus the geometry comparison are useful for device-level strain engineering. The central claims are plausible and the paper is refreshingly candid about its unassigned diffraction peak. However, the headline 2 to 2.5% strain currently rests on a single spectrum whose key peak assignment is not uniquely established against intermetallic phases listed in the paper's own Table S7, and no statistical or fitting uncertainties are reported anywhere, so the strength of the comparative claims (seed-layer control, sixfold geometry enhancement) cannot presently be assessed. These gaps are addressable with targeted control samples and uncertainty reporting, within the manuscript's scope, which is why I recommend major revision rather than rejection.","major_comments":[{"comment":"The assignment of the peak at 2θ = 18.170° (d = 2.80 Å) to 1H MoS2 is not secure. Table S7 itself lists AuTi (hexagonal, d = 2.832 Å, (100)) and Au3Ni (cubic, d = 2.828 Å, (110)), both within about 1% of the measured 2.80 Å spacing, and the same spectrum contains an acknowledged unassignable peak at d = 2.894 Å. Because the stack places Au, Ti, and Ni in intimate contact, intermetallic formation is chemically plausible, yet the discussion of assignment ambiguity is limited to the left-most peak and does not address the identical risk for the peak identified as 1H MoS2. If the 18.170° peak is an intermetallic reflection, the headline strain claim is unsupported. I request: (a) a control GIXRD scan of the same Au/Ti/Ni stack deposited on bare SiO2/Si without MoS2, showing the absence of a peak near 18.170°; (b) the relative integrated intensity of the 18.170° peak versus the Au(111) peak, together with a scattering-volume estimate demonstrating that a single MoS2 monolayer can produce the observed intensity; and (c) an explicit statement of why the candidate phases in Table S7 are excluded for this peak.","section":"§4, Fig. 4(b), Table S6, Table S7"},{"comment":"No uncertainty is reported for any 2θ, d, or strain value, and each condition is represented by a single spectrum. This is load-bearing because several comparative conclusions rest on 2θ differences of only a few thousandths of a degree: the ALD temperature series spans -0.06% to -0.09% strain, and the Al2O3 thickness series is non-monotonic (-0.12%, -0.08%, -0.15%, -0.07% for seed-only, 15, 20, and 25 nm), with the largest magnitude appearing at 20 nm rather than at the largest thickness. As reported, these numbers are equally consistent with the stated conclusions and with fitting noise. A quantitative statement of peak-center precision (Gaussian fit uncertainty, step size, counting statistics, and sample-to-sample reproducibility) is required before the 'no effect of temperature/thickness' claims and the sixfold geometry ratio in Table S5 can be evaluated.","section":"Tables S1–S6, Figs. 2–4"},{"comment":"The experimental geometry and calibration are under-reported. The text states a grazing incidence angle of 0.1–0.2° 'depending on the sample' but does not list the angle used for each sample, nor the critical angle of the corresponding surface. For the Au/Ti/Ni-capped sample, the critical angle of the Au layer at 14 keV is roughly 0.13°; incidence below this value confines the beam to an evanescent wave with a penetration depth of only a few nanometers, which cannot reach a MoS2 monolayer buried under approximately 27.5 nm of metal. The paper should report the incidence angle per sample, the resulting penetration condition through the metal stack, and an angular calibration check such as the measured Au(111) position compared with its bulk value; the latter is particularly important because the absolute d values, and hence the 2.5% strain, depend on the angular accuracy of the instrument.","section":"§2 (Fig. 1(b)–(d)) and §4 (Fig. 4(b))"},{"comment":"The sixfold geometry enhancement compares a patterned MoS2 strip sample with Ti/Au metal lines and an AlOx cap against a blanket AlOx-capped sample. The two samples differ not only in geometry but in MoS2 morphology (etched strips versus continuous film), in layer stack, and in the fraction of the probed area covered by metal, so the attribution of the enhanced strain to geometry is not uniquely established. The argument would be materially strengthened by a third control, such as a blanket metal film without patterning or patterned MoS2 without metal lines. In addition, the increased FWHM is attributed to non-uniform strain along the channel, but the FWHM values are never reported, so the explanation cannot be checked.","section":"§3, Fig. 3(d), Table S5"}],"minor_comments":[{"comment":"The headline strain value is quoted inconsistently: the abstract says 'up to 2%', the Fig. 4(b) caption says '>2%', and Table S6 reports 2.45%. Strictly, 2θ = 18.170° with λ = 0.8856 Å gives d ≈ 2.804 Å, which is about 2.6% relative to the d = 2.733 Å reference, so all quoted values should be reconciled with the quoted 2θ and d.","section":"Abstract, Fig. 4(b) caption, Table S6"},{"comment":"The text states that as-grown monolayer MoS2 has built-in tensile strain of '~0.5%' for CVD growth at 750 °C, whereas Table S1 reports 0.33% for the as-grown sample; these values should be reconciled.","section":"§2 and Table S1"},{"comment":"The caption contains a garbled fragment ('The change in strain compared ε = 33% -grown') that appears to be a typographical or OCR artifact; it should be rewritten as complete sentences stating the strain values for as-grown and transferred MoS2.","section":"Fig. 2(a) caption"},{"comment":"The reflection is referred to inconsistently as the '(01) peak', '(0l) peak', and (in the figure) the d = √3a/2 spacing; this is the (100) reflection of the hexagonal lattice, and the notation should be unified.","section":"Fig. 1(e) and text"},{"comment":"The strain reference baseline is not stated uniformly: Table S1 references as-grown and transferred films to the theoretical value (d = 2.738 Å), while Tables S2–S5 reference capped films to the measured as-grown value and Table S6 references the capped film to the measured transferred value. Each table should state explicitly which reference d was used, to avoid reader confusion about the strain arithmetic.","section":"Tables S1–S6"},{"comment":"The claim that 2% strain is 'one of the largest experimental values to date on a rigid substrate' is not contextualized by a comparison list or citations to other measured values; either add such a comparison or soften the claim to a supported range, given that the value originates from a single spectrum.","section":"Abstract, §4"}],"recommendation":"major_revision","confidential_remarks":"For the editor: I do not think this is a rejection candidate. The methodological contribution, direct in-plane XRD strain measurement on monolayer TMDs, is real, and the missing controls and error bars are obtainable within a normal revision cycle. However, I expect the headline 2–2.5% strain claim to remain contested until the peak-assignment control (same metal stack on bare SiO2/Si) is shown: the paper's own Table S7 demonstrates that the assigned 18.170° peak shares d-spacing with AuTi and Au3Ni intermetallics, and the authors' admitted inability to assign the 2.894 Å peak means two of the three peaks in Fig. 4(b) are of uncertain origin. I would also ask the editor to request that the authors state, per condition, how many independent samples and how many repeated scans were used, since the single-spectrum nature of the data is the source of most of the fragility."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a useful direct-measurement catalog of capping-induced strain in monolayer MoS2, but the headline 2% value rests on one peak assignment in one spectrum, and that needs to be nailed down before the paper's big claim is taken seriously.\n\nWhat's actually new: the paper extends the Schauble et al. GIXRD approach to a systematic study of ALD Al2O3 thickness and temperature, Al seed layers, patterned metal lines, and a highly stressed Au/Ti/Ni stack. The finding that ALD alumina strain is controlled by the seed layer rather than the oxide thickness or deposition temperature is plausible and practical. The ~6x strain enhancement from patterned contacts versus blanket films is also a clean, direct result. These are honest lattice measurements, which is an improvement over Raman/PL estimates that need external calibration.\n\nWhat it does well: the authors are transparent about the unassigned peak at d=2.894 in the metal-capped sample. They list candidate intermetallic phases in Table S7 and explicitly discuss the possibility of 1T' MoS2 or non-uniform strain. That is the right way to handle an unresolved feature, and it makes the paper easy to referee.\n\nSoft spots: the 2% strain comes from a single sample, a single spectrum, and no error bars. The assigned MoS2 peak at d=2.80 is uncomfortably close to AuTi (2.832) and Au3Ni (2.828) from the paper's own table. Without uncertainty quantification, misassignment cannot be ruled out. There is also a small internal inconsistency: the text says as-grown strain is ~0.5%, but Table S1 gives 0.33%. The 'ALD temperature has little impact' conclusion relies on peak shifts of 0.02-0.03%, also without error bars—plausible, but not strong. These are real weaknesses, but they are concentrated in the headline claim; the seed-layer and geometry results are on firmer ground.\n\nBottom line: this paper deserves a serious referee. The systematic data and the seed-layer/geometry findings are worth publishing after routine revision. The 2% claim should require confirmation—additional reflections, a control stack without MoS2, or a complementary technique—plus replicates and error bars. As it stands, accept conditionally with that specific request.\n\nRecommendation: send to peer review, but flag the peak assignment and repeatability as the gating issues.","headline":"Useful direct GIXRD strain data for MoS2 capping layers, but the headline 2% strain is a single unconfirmed peak assignment and should be verified before it drives any conclusions.","tokens_in":11341,"tokens_out":3156,"would_cite":true,"duration_ms":57511,"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":"X-ray diffraction directly measures how capping layers strain monolayer MoS2, including a ~2% tensile strain from a metal stack.","keywords":["monolayer MoS2","grazing-incidence x-ray diffraction","GIXRD","strain engineering","capping layers","ALD alumina","metal contacts","2D materials"],"falsifier":"Deposit the same Au (5 nm)/Ti (2.5 nm)/Ni (20 nm) stack on a bare SiO2/Si substrate without MoS2 and run the same GIXRD scan: if the $d = 2.80$ Å peak still appears, it is not from MoS2. Alternatively, cross-section the capped film with transmission electron microscopy and perform elemental mapping to see whether an intermetallic layer with that d-spacing exists at the interface.","tokens_in":10608,"feed_emoji":"🔬","tokens_out":5680,"duration_ms":51568,"temperature":0.7,"pith_summary":"This paper uses grazing-incidence x-ray diffraction (GIXRD) to directly measure the lattice spacing, and therefore strain, of monolayer MoS2 under capping layers and contact-like patterns. It finds that alumina capping strain is set by the thin aluminum seed layer and by substrate adhesion, not by oxide thickness or deposition temperature. It reports that patterned metal lines impart about six times more strain than blanket oxide, and that a highly stressed Au/Ti/Ni blanket stack stretches the MoS2 lattice to roughly 2% tensile strain, one of the largest values recorded on a rigid substrate. The work is a direct structural check on strain values that are usually inferred indirectly from Raman or photoluminescence measurements.","feed_headline":"Direct x-ray data push MoS2 to 2% strain","feed_subtitle":"Seed layers, adhesion, and contact geometry—not oxide thickness—control strain in 2D transistors.","key_machinery":"The central object is the (01) in-plane diffraction peak of monolayer MoS2, measured by grazing-incidence x-ray diffraction (GIXRD) with a 14 keV source. The d-spacing of this peak maps directly to the in-plane lattice constant $a$ through $d = \\sqrt{3}a/2$, so shifts in $2\\theta$ translate directly into strain via Bragg's law. The argument is carried by comparing this peak across as-grown versus transferred films, blanket versus patterned caps, and oxide versus metal stacks, with the peak shift itself serving as the strain meter.","core_discovery":"The central claim is that the strain in monolayer MoS2 from common fabrication steps can be measured directly by x-ray diffraction and is governed by interfacial adhesion and geometry. For ALD alumina, the ~0.1% compressive strain comes from the ~1.5 nm Al seed layer, while ALD temperature (130-300 °C) and thickness (15-25 nm) change nothing. Transferring MoS2 releases the as-grown tensile strain, and the same capping stack that compresses as-grown MoS2 by -0.23% leaves transferred MoS2 unchanged, showing that substrate adhesion gates strain transfer. Patterning MoS2 into strips with metal contact lines increases the imparted strain to about -0.44%, roughly six times the blanket value. A blanket stack of 5 nm Au / 2.5 nm Ti / 20 nm Ni on transferred MoS2 shifts the (01) diffraction peak from $d = 2.733$ Å to $d = 2.80$ Å, corresponding to about 2% tensile strain, claimed as one of the largest experimental strains reported on a rigid substrate.","pith_inferences":["If the $d = 2.80$ Å peak is truly the MoS2 reflection, then the same stack on patterned channels should produce larger or spatially non-uniform strain under contacts, which could be probed by nanobeam diffraction or by mapping the peak width.","The ~2% tensile strain approaches the regime where the 1H-to-1T' phase transition has been predicted; the unassigned low-angle peak at $d = 2.894$ Å could be a sign of a local phase mixture or of an intermetallic compound, and the paper itself leaves that identification open.","The same measurement protocol could be applied to other transition metal dichalcogenides (WS2, WSe2, MoSe2) and to other stressors to build a direct, quantitative library of strain transfer through realistic device stacks.","A testable consequence of the adhesion picture is that improving MoS2-substrate adhesion (for example by annealing or by choosing a different substrate) should increase the strain transferred from any given capping layer."],"forward_implications":["Alumina encapsulation recipes can be chosen for their electrical or thermal role without worrying about changing MoS2 strain, because the seed layer, not the oxide, controls the effect.","Metal contact geometry is a practical lever: transistor-like lines give roughly six times the strain of blanket oxide caps, so device layout can be used for strain engineering.","Substrate adhesion determines whether a capping stack's stress reaches the 2D film, so transferred films require different strain engineering than as-grown films.","Direct x-ray diffraction can replace indirect Raman or photoluminescence strain estimates for capping layers that are too thick or metallic for optical probes.","A blanket Au/Ti/Ni stack can put monolayer MoS2 on a rigid substrate near strain levels normally seen only on flexible substrates, opening band-structure tuning without bending."],"supporting_citations":[{"why":"Supplies the CVD growth process that produces the large-area monolayer MoS2 films measured here.","marker":"28"},{"why":"Introduces the Au/Ti/Ni metal stack and the device context whose blanket version is tested here.","marker":"6"},{"why":"Establishes that gold lines impart minimal strain and that diffraction is more reliable than optical strain probes.","marker":"26"},{"why":"Documents the built-in tensile strain in as-grown MoS2 from thermal expansion mismatch.","marker":"3"},{"why":"Shows that substrate adhesion controls whether stress transfers into the 2D film.","marker":"31"},{"why":"Explains non-uniform strain under contacts, used to interpret the broadened peak in patterned samples.","marker":"36"},{"why":"Provides the theoretical threshold for the 1H-to-1T' phase transition invoked for the unassigned peak.","marker":"38"}],"fun_headline_variants":["X-rays reveal 2% strain in MoS2 via cap layers","Seed layer, not thickness, controls MoS2 strain","Direct X-ray measurement: MoS2 hits 2% strain","Adhesion and geometry dictate MoS2 capping strain","2% strain in MoS2 from capping: X-ray direct proof"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 2% strain result depends on the diffraction peak at $2\\theta = 18.170^\\circ$ being the in-plane MoS2 reflection; if that peak comes from a gold-titanium intermetallic compound instead, the headline strain value would not hold.","fun_headline_variants_meta":{"raw":{"variants":["X-rays reveal 2% strain in MoS2 via cap layers","Seed layer, not thickness, controls MoS2 strain","Direct X-ray measurement: MoS2 hits 2% strain","Adhesion and geometry dictate MoS2 capping strain","2% strain in MoS2 from capping: X-ray direct proof"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000507,"raw_usage":{"total_tokens":2494,"prompt_tokens":987,"completion_tokens":1507,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":603,"completion_tokens_details":{"reasoning_tokens":1417}},"tokens_in":603,"tokens_out":1507,"duration_ms":10410,"temperature":1.0,"reasoning_tokens":1417,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:00:24.879827+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deposit the same Au (5 nm)/Ti (2.5 nm)/Ni (20 nm) stack on a bare SiO2/Si substrate without MoS2 and run the same GIXRD scan: if the $d = 2.80$ Å peak still appears, it is not from MoS2. Alternatively, cross-section the capped film with transmission electron microscopy and perform elemental mapping to see whether an intermetallic layer with that d-spacing exists at the interface.","supporting_citations":[],"review_version":1}