{"id":"568ded6f-d2a5-4068-932c-afd13d956d43","arxiv_id":"1908.03881","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Molten tin droplets stick strongly to warm silicon and coated-mirror surfaces, peel off spontaneously on cold surfaces, and can be converted to brittle gray tin for easy removal.","lead":"This paper tests how molten tin droplets stick to silicon wafers and coated mirrors in vacuum, varying substrate temperature from -100°C to 95°C. It finds that cooling strongly reduces sticking and that a tin-pest phase change can loosen deposits that do stick, which matters for cleaning EUV lithography optics.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The collector-cooling recommendation relies on a temperature threshold that was not tested: non-sticking is shown only at -50°C, while super-cooling to -20/-30°C is recommended.","rationale":"The paper's core observations are direct and well illustrated; the tin-pest removal sequences are convincing demonstrations, and the qualitative sticking matrix is a useful empirical contribution. The weakest link is not the acknowledged model approximation but the quantitative jump from one deep-cooled point to a recommended operating range. The industry-facing implication depends on this threshold. The proposed test would settle whether the recommendation is warranted. The reader's model concern is valid but is explicitly acknowledged in the text and does not undermine the empirical sticking matrix; the threshold issue is unacknowledged and directly actionable. The verdict remains CONDITIONAL, as the original reader stated, because the paper needs additional measurements before the collector-cooling recommendation can be treated as established.","tokens_in":14216,"tokens_out":11784,"duration_ms":123898,"concrete_test":"Run tin-drip tests at Ts = -20, -30, -40, and -50°C on ZrN- and ZrO2-capped Mo/Si ML samples (same setup, at least 3 drops per temperature), and measure the shear force needed to dislodge each splat with a calibrated force gauge rather than the ~1 N manual push. If any splat remains attached at -30°C with a measurable adhesion force, the collector super-cooling recommendation is not supported by the data and must be revised to colder temperatures or qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4 recommends operating EUV collectors with inlet temperatures below -20°C or -30°C to reduce tin sticking. The experimental support (Table 3, §3.2) consists only of drips at Ts≈-50°C, 23°C, and ~90°C. No data exist between 23°C and -50°C, so the threshold below which self-peeling reliably dominates is unknown. For the relevant ZrN/ZrO2-capped ML coatings, Table 3 shows mixed behavior at 23°C (sticks or peels), and the adhesion test is a manual ~1 N push, not a quantitative measurement. The contact-temperature model cannot fill this gap because the authors state it is not valid for coated samples, and Table 2 shows nearly equal thermal resistance for uncapped vs capped ML despite different sticking, so surface chemistry also matters. Thus the -20/-30°C recommendation is an untested extrapolation; if the actual threshold is colder, the proposed operating condition would not prevent sticking.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental study in which molten tin droplets (3.0–3.6 mm diameter, ~250 °C) are dripped under vacuum onto silicon wafers and Mo/Si multilayer-coated mirror samples with various cap layers. The authors characterize whether the solidified splats stick, self-peel, or self-contract as a function of substrate temperature (-100 °C to ~95 °C) and coating type, and they demonstrate an in-situ cleaning method based on infection with gray tin powder to induce the β-Sn to α-Sn (tin pest) transformation. The central claims are that droplet adhesion is strong at ~90 °C, largely non-sticking at -50 °C (except on glass), and that transformation to gray tin always leads to embrittlement and detachment, enabling contamination removal without coating damage. The paper further proposes that super-cooling EUV collector mirrors below -20 °C or -30 °C would reduce tin sticking during source operation.","tokens_in":14352,"tokens_out":4114,"duration_ms":38551,"significance":"If the reported temperature-dependent sticking behavior holds, the study has practical value for EUV lithography source development, where tin debris contamination of collector mirrors is a known lifetime issue. The paper contributes a useful observational matrix of sticking and detachment behavior across relevant collector-mirror coatings (ZrN- and ZrO2-capped Mo/Si multilayers), and demonstrates that tin-pest transformation is a reproducible cleaning route for strongly adhering deposits, with photographic time series and profilometry supporting the morphological claims. The work is largely free of fitted parameters: the sticking matrix is observational and the contact-temperature values are computed from literature material properties, which is a strength. The main limitations are the qualitative adhesion classification (a manual ~1 N push) and the extrapolation of a simplified heat-transfer model to coated samples for which the model is acknowledged to be invalid; these limit the strength of the causal explanation and the proposed operating recommendation.","major_comments":[{"comment":"The recommendation to operate EUV collectors at inlet temperatures below -20 °C or -30 °C is an extrapolation beyond the presented data. The sticking matrix (Table 3) contains only substrate temperatures near -50 °C, 23 °C, and ~90 °C (plus one glass point at -100 °C); no measurement exists in the interval between -50 °C and 23 °C. Since the proposed operating window lies in this unmeasured interval, the statement that surface temperatures 'sufficiently low for the dominance of splat self-peeling' would be reached is not directly supported. Please either add dripping experiments at intermediate substrate temperatures (e.g., -20 °C and -30 °C) on at least the ML-coated samples, or explicitly restrict the conclusion to temperatures at or below -50 °C.","section":"Section 4 and Table 3"},{"comment":"The contact-temperature model is acknowledged to be invalid for coated samples ('the assumption of a homogeneous semi-infinite body below the tin splat is no longer valid'), yet the paper uses Eq. (1) together with the one-dimensional coating thermal resistance ranking to interpret the sticking trends. This is internally problematic: Table 2 gives nearly equal thermal resistance per area for uncapped Mo/Si (319), ZrN-capped (318), and ZrO2-capped (323) coatings, but Table 3 shows qualitatively different behavior at 23 °C (peels vs. sticks or peels). The thermal-resistance ranking alone therefore cannot explain the differences among the capped and uncapped ML samples, indicating that surface chemistry (wetting and adhesion) contributes as well. The paper should either develop a layered heat-transfer model that accounts for finite coating thickness and contact resistance, or explicitly present the thermal-resistance ranking as a qualitative tendency rather than as the explanatory mechanism.","section":"Section 3.1, Eq. (1) and Table 2"},{"comment":"The claim that the observed sticking and detachment trend is 'generally consistent with the sample ranking based on thermal resistance per unit area' is weakened by the Si-ZrO2 case. This sample has R_t/A of only 5.3 m²K GW⁻¹ (Table 2) and yet shows self-contraction at -50 °C and sometimes even at 23 °C (Table 3), whereas the uncapped Mo/Si sample (R_t/A = 319) only peels or contracts. The very thin ZrO2 layer has negligible thermal resistance, so the observed self-contraction on Si-ZrO2 cannot be explained by heat conduction alone. Please address this counterexample explicitly or adjust the proposed ranking.","section":"Section 3.2, last paragraph"},{"comment":"Adhesion is classified by a manual push force of up to about 1 N, yielding only a binary 'strong' versus 'weak/zero' distinction. The Abstract's statement that 'the adhesion strength of solidified tin splats decreased strongly with decreasing substrate temperature' is therefore stronger than the measurement supports. I recommend rewording to 'sticking tendency' or 'sticking probability', or adding a quantitative pull-off measurement to substantiate the strength claim.","section":"Section 2.2 and Abstract"}],"minor_comments":[{"comment":"Two subsections are numbered 3.2 ('Tin splat morphology' and 'Delamination behavior of non-sticking tin droplets'); the second should be renumbered as 3.3 and the subsequent 'Transformation behavior' as 3.4.","section":"Section 3.2"},{"comment":"The caption of Fig. 5 states that the tin drop was dripped at Ts = 90 °C, while the text in Section 3.3 says 'dripped (at Ts = 82 °C)'. Please make the values consistent.","section":"Fig. 5 caption and text"},{"comment":"The caption gives 'mass: 0.154 mg', which is likely a typo for 0.154 g (the text elsewhere reports drop masses of ~100–170 mg). Please correct the unit.","section":"Fig. 6 caption"},{"comment":"The reported substrate temperatures are rounded ('~90 °C', '-50 °C'), but no uncertainty or measured range is provided. Adding the spread or uncertainty of the thermocouple readings would make the comparison among samples more quantitative.","section":"Table 3 and Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the scope of Applied Physics A and reports a reproducible cleaning phenomenon of potential industrial relevance. The main revision needed is to bring the conclusions and the proposed cooling recommendation in line with the actual data: either add intermediate-temperature experiments or temper the claims. The contact-temperature model should be used more cautiously for coated samples, with explicit acknowledgment of the surface-chemistry contribution. I do not see grounds for rejection: the observed sticking/non-sticking trend at the three tested temperatures is directly supported by the photographs and Table 3, and the tin-pest transformation is demonstrated convincingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth knowing: this is a solid, well-scoped experimental paper. The central observation—tin splats stick strongly on these substrates near 90 °C and generally do not stick at -50 °C—is directly supported by a consistent matrix of samples, photographs, and profilometry. The most useful new results are the vacuum dripping data at substrate temperatures below room temperature, the observation of self-contraction rather than only self-peeling, and the demonstration that tin-pest transformation can detach tin splats from ZrN- and ZrO2-capped Mo/Si multilayer coatings without visible coating damage. There are no fitted parameters and no circular steps; the tin-pest cleaning demonstration stands on its own.\n\nThe paper also builds fairly on prior work. It extends de Ruiter et al.'s room-temperature self-peeling results into vacuum and onto EUV-relevant cap layers, and it is honest about the limits of the contact-temperature model. The authors state that Eq. (1) is not valid for coated samples, then use the coating thermal-resistance ranking of Table 2 to interpret sticking trends. That is a real soft spot: Table 2 gives nearly identical thermal resistance for uncapped and capped ML coatings, yet their sticking behavior at 23 °C differs, so surface chemistry is doing work the model does not capture. The qualitative ~1 N push classification is another limit; adhesion is not measured quantitatively. These issues do not overturn the main temperature trend, but they weaken the mechanism story.\n\nThe biggest gap is the operating recommendation. Section 4 proposes collector super-cooling with inlet temperatures below -20 °C or -30 °C. The experiments only test roughly 90 °C, 23 °C, -50 °C, and one glass case at -100 °C. There is no data between 23 °C and -50 °C, so the threshold below which self-peeling reliably dominates is unknown. The model cannot fill that gap for coated samples, and the extrapolation from millimeter-size drops to EUV microdroplets is explicitly hedged—appropriately so.\n\nOverall, this is an honest, useful paper for the EUV collector community and for people studying droplet impact and solidification. It deserves a serious referee. Before the cooling recommendation is treated as established, it needs quantitative adhesion measurements, more substrate temperatures between -50 °C and room temperature, and a clearer statement of how far the contact-temperature model can be pushed for coated samples. With those changes, I would be comfortable seeing it published.","headline":"Useful experimental addition on tin drop adhesion and tin-pest cleaning, but the collector super-cooling recommendation goes beyond the measured temperature range.","tokens_in":14906,"tokens_out":2292,"would_cite":true,"duration_ms":26294,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Substrate temperature controls whether molten tin droplets adhere to silicon and multilayer mirror surfaces, with cooling to about -50 °C making them self-peel, while any adhering deposit can be removed by inducing the tin-pest phase…","keywords":["droplet impact","phase transformation","optics cleaning","multilayer coating","extreme ultraviolet light","tin pest","self-peeling","thermal effusivity"],"falsifier":"Measure the sticking outcome of identical tin drops on two Mo/Si multilayer samples with the same ZrO2 cap and the same surface chemistry but coating thicknesses differing by, say, a factor of two: if the thicker coating does not stick more often at the same substrate temperature, the thermal-resistance mechanism is wrong. Alternatively, embed a thin-film thermocouple or use high-speed thermal imaging to check whether the interfacial contact temperature under the splat actually changes with coating thickness as Eq. (1) plus Table 2 predicts.","tokens_in":13961,"feed_emoji":"🧊","tokens_out":6847,"duration_ms":69745,"temperature":0.7,"pith_summary":"The paper tries to establish that the sticking of molten tin droplets onto silicon and multilayer-coated mirror surfaces is controlled mainly by substrate temperature: at roughly 90 °C droplets adhere strongly, while at -50 °C they do not stick on any tested surface except glass. This matters for extreme ultraviolet lithography, where tin debris from the plasma source contaminates collector mirrors and shortens their life. The paper also shows that tin deposits that do stick can be removed in-situ by seeding them with gray tin powder and cooling, triggering the tin-pest phase transformation that embrittles and detaches the deposit within hours without damaging the mirror coating. A sympathetic reader would take away a practical recipe: run the collector cold to avoid sticking, and use tin pest as a cleaning step.","feed_headline":"At -50 °C, molten tin drops stop sticking to mirror surfaces","feed_subtitle":"Cooling collector mirrors could stop tin debris buildup, and tin pest can clean whatever sticks.","key_machinery":"The load-bearing theoretical object is the interfacial contact temperature $T_c = T_s + (T_d - T_s)(1+e_s/e_d)^{-1}$ from Eq. (1), where $e_i=(k\\rho c_p)^{1/2}$ is thermal effusivity, plus the per-area thermal resistance $R_t/A$ of the coating stack listed in Table 2. This pair of quantities converts a materials list into a predicted sticking ranking: low effusivity or high coating resistance keeps the interface hot and promotes adhesion, while high effusivity and thin coatings cool the interface and promote self-peeling. The second mechanism is the allotropic $\\beta$-tin to $\\alpha$-tin tin-pest transformation, induced by seeding with gray tin powder at low temperature, which expands the deposit by about 25 percent, embrittles it, and drives detachment.","core_discovery":"On the paper's own terms, the central experimental discovery is a temperature-controlled adhesion switch. Tin drops at about 250 °C falling 41 cm onto smooth horizontal samples stick strongly when the surface is held near 90 °C, peel off or contract and detach at 23 °C on low-adhesion coatings, and do not stick at all at -50 °C except on soda-lime glass, which still holds drops even at -100 °C. The order of sticking tendency across coatings follows the one-dimensional thermal resistance of the coating stack: thicker or more insulating layers (thick oxide, Mo/Si multilayers with ZrO2 cap) keep the interfacial contact temperature higher and favor sticking, while bare silicon with its thin native oxide cools the interface fastest and favors self-peeling. For every case where a splat did stick, infection with alpha-tin powder followed by cooling to -30 to -40 °C converted the beta-tin to gray alpha-tin, with roughly 25 percent volume expansion and embrittlement, cracking and detaching the deposit; full conversion occurred within 24 hours, usually in less than 12 hours, with no visible damage to the multilayer coating.","pith_inferences":["If the thermal-resistance picture transfers to smaller droplets, super-cooling should also suppress adhesion of tin microdroplets and vapor-condensed tin, which would make the EUV debris problem much smaller; this is a testable extension the authors only hint at.","A direct falsification of the model would be to compare two coatings with identical top-layer chemistry but very different layer thickness: if sticking does not track $R_t/A$, the assumed heat-flow mechanism is wrong and the collector-mirror recommendation would rest on correlation rather than cause.","The tin-pest cleaning step could be combined with hydrogen-radical etching, with pest handling thick splats and radicals handling thin films, potentially covering the full size range of tin contamination in one in-situ procedure.","Because the paper notes coated samples violate the homogeneous semi-infinite assumption, deriving a corrected contact temperature for the full layer stack rather than using substrate bulk values would let the model make quantitative predictions worth testing."],"forward_implications":["Cooling an EUV collector mirror surface to roughly -50 °C should make most incident molten tin drops self-peel or contract instead of adhering, reducing contamination buildup during source operation.","Any tin splat that nevertheless sticks can be removed in situ within about a day by sprinkling gray tin seed powder on it and cooling to -30 to -40 °C, because the tin-pest transformation embrittles and lifts the deposit without damaging the Mo/Si multilayer.","The cleaning step works on every substrate tested, including bare Si, oxidized Si, uncapped Mo/Si, ZrN-capped and ZrO2-capped Mo/Si, and glass, so it does not depend on the cap-layer chemistry.","Coating stacks with higher thermal resistance, such as thick oxide or Mo/Si with a ZrO2 cap, tend to hold tin splats even at room temperature, while bare silicon and thin-oxide surfaces spontaneously shed them.","The observed self-contraction in vacuum shows that recoil of spread tin splats is not caused by trapped air; surface tension alone can pull the splat back into a ball."],"supporting_citations":[{"why":"Supplies the self-peeling mechanism and the flat-surface droplet-impact phenomenology this study extends to vacuum and sub-zero temperatures.","marker":"[9]"},{"why":"Supplies the splat-freezing timescale and spread-factor baseline used to interpret the observed pancake splats.","marker":"[5]"},{"why":"Source of the contact-temperature model and of the splat-shape comparisons used in the heat-transfer discussion.","marker":"[6]"},{"why":"Describes the tin-pest embrittlement process for solidified tin drops that the in-situ cleaning method builds on.","marker":"[19]"},{"why":"Previous first tests of tin-pest cleaning on uncapped Mo/Si samples; the present work refines the procedure and extends it to capped multilayer coatings.","marker":"[20]"},{"why":"Review of the beta-to-alpha tin transformation, supplying the 13 °C stability boundary and the volume-expansion and embrittlement facts.","marker":"[21]"},{"why":"Tin-pest reliability data used for the transformation conditions and practical expectations.","marker":"[22]"},{"why":"Gives the low work of adhesion and wetting of tin on ZrO2, justifying the expectation of low sticking on ZrO2 caps.","marker":"[33]"},{"why":"Provides the cross-plane thermal conductivity of Mo/Si multilayers used to rank coating thermal resistance.","marker":"[31]"}],"fun_headline_variants":["Supercooled mirrors stop tin droplets from sticking","Tin pest infection makes tin deposits detach safely","Tin drops self-peel at room temperature on some mirrors","Chill optics to -50 °C to avoid tin contamination"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument that cooling prevents sticking and that coating ranking predicts sticking rests on a contact-temperature formula that assumes two homogeneous semi-infinite bodies with no contact resistance; the authors themselves note that coated samples are not homogeneous semi-infinite, so if layered heat flow or interfacial contact resistance behaves differently, the proposed mechanism and the collector-mirror recommendation could fail.","fun_headline_variants_meta":{"raw":{"variants":["Supercooled mirrors stop tin droplets from sticking","Tin pest infection makes tin deposits detach safely","Tin drops self-peel at room temperature on some mirrors","Chill optics to -50 °C to avoid tin contamination"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001564,"raw_usage":{"total_tokens":6274,"prompt_tokens":1000,"completion_tokens":5274,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":616,"completion_tokens_details":{"reasoning_tokens":5210}},"tokens_in":616,"tokens_out":5274,"duration_ms":49946,"temperature":1.0,"reasoning_tokens":5210,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:59:18.659681+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the sticking outcome of identical tin drops on two Mo/Si multilayer samples with the same ZrO2 cap and the same surface chemistry but coating thicknesses differing by, say, a factor of two: if the thicker coating does not stick more often at the same substrate temperature, the thermal-resistance mechanism is wrong. Alternatively, embed a thin-film thermocouple or use high-speed thermal imaging to check whether the interfacial contact temperature under the splat actually changes with coating thickness as Eq. (1) plus Table 2 predicts.","supporting_citations":[{"cited_title":"de Ruiter, D","cited_arxiv_id":null,"evidence_quote":"Supplies the self-peeling mechanism and the flat-surface droplet-impact phenomenology this study extends to vacuum and sub-zero temperatures."},{"cited_title":"Pasandideh-Fard, R","cited_arxiv_id":null,"evidence_quote":"Supplies the splat-freezing timescale and spread-factor baseline used to interpret the observed pancake splats."},{"cited_title":"Wang, H.-L","cited_arxiv_id":null,"evidence_quote":"Source of the contact-temperature model and of the splat-shape comparisons used in the heat-transfer discussion."},{"cited_title":"Böwering, Mater","cited_arxiv_id":null,"evidence_quote":"Describes the tin-pest embrittlement process for solidified tin drops that the in-situ cleaning method builds on."},{"cited_title":"Panici, D","cited_arxiv_id":null,"evidence_quote":"Previous first tests of tin-pest cleaning on uncapped Mo/Si samples; the present work refines the procedure and extends it to capped multilayer coatings."},{"cited_title":"Plumbridge, J","cited_arxiv_id":null,"evidence_quote":"Review of the beta-to-alpha tin transformation, supplying the 13 °C stability boundary and the volume-expansion and embrittlement facts."},{"cited_title":"Böwering, C","cited_arxiv_id":null,"evidence_quote":"Tin-pest reliability data used for the transformation conditions and practical expectations."},{"cited_title":"Bozorg-Grayeli, Z","cited_arxiv_id":null,"evidence_quote":"Gives the low work of adhesion and wetting of tin on ZrO2, justifying the expectation of low sticking on ZrO2 caps."},{"cited_title":"Thoroddsen, J","cited_arxiv_id":null,"evidence_quote":"Provides the cross-plane thermal conductivity of Mo/Si multilayers used to rank coating thermal resistance."}],"review_version":1}