{"id":"a6a78063-8a16-4fd1-b07f-03048b950252","arxiv_id":"1908.01604","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"A random-surface adsorption model fitted to nitrogen isotherms extracts silica nanoroughness, but the fractal-dimension formula reduces to a power-law fit to the model's own output.","lead":"Researchers fit a random-surface model to nitrogen adsorption data for silica materials to extract roughness, correlation length, and specific surface area. They then claim the model naturally produces the fractal dimensions seen in experiments, but that 'derivation' is a fit to the model's own output.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed fractal-dimension derivation Df = 2 + 1/γ relies on an unproved cl ∝ 1/d assertion; without it Eq. (7) does not follow, so the central claim is a fitted exponent presented as a prediction.","rationale":"Good-faith reading: the paper's purpose is to characterize silica rough surfaces from low-pressure nitrogen adsorption using RS-DFT, and to show that the surface fractal dimension seen in experiments follows from the correlated random surface model rather than from an assumed self-similarity. For that second purpose, Eq. (7) is the load-bearing result. The reader's weakest_assumption identifies precisely the vulnerable link: the transformation from a fitted power-law exponent γ to a physical fractal dimension depends on cl ∼ 1/d, which is asserted, not derived, and the power law itself is fitted to the inversion procedure's internal scan over A. I checked the text following Fig. 4 and the derivation of Eqs. (6)-(7); the assertion 'for molecules with different sizes near a certain rough surface ... cl ∼ 1/d' carries the entire argument. No derivation, reference, or multi-adsorbate experiment supports it. Without it, Df = 2 + 1/γ is a restatement of a fit, not a prediction of the model. The paper also lacks error bars and released data, but those are secondary; the fractal-dimension claim is the central contribution and it is not supported. The roughness parameters and ARS comparisons in Table II may have standalone value, but they do not rescue the headline result. Hence the reader's REJECT verdict is appropriate; no change is needed. The concrete test I propose, multi-probe inversion, would settle whether cl ∼ 1/d holds within the model; if it does, the concern would be resolved and the claim would merit reconsideration.","tokens_in":8442,"tokens_out":4623,"duration_ms":46014,"concrete_test":"Use the same RS-DFT inversion workflow to generate synthetic adsorption isotherms for one of the reported silica samples with a known (var, cl) pair and probe diameters d = 0.30, 0.36, and 0.42 nm, then recover cl from each isotherm using the paper's minimization (5) with the same fitting range. If the recovered cl does not scale as 1/d across the three diameters, the assertion in the paragraph after Fig. 4 fails, and Df = 2 + 1/γ should not be reported as a model prediction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Within the paper's own logic, the central claim that the surface fractal dimension observed in experiments is natural for the correlated random surface model reduces to Eq. (7), Df = 2 + 1/γ. That reduction requires two unsecured steps. First, the paragraph following Fig. 4 fits cl = kL^γ to the curve generated by scanning A over 0.75ABET to 1.25ABET. This is a fit to the inversion procedure's own output, and the scanned L range is only a factor sqrt(1.25/0.75) ≈ 1.29 wide, with cl constrained to [3d, 7d]; a power-law exponent extracted from such a narrow, bound-limited window is not an independent material property. Second, and decisively, the text asserts 'for molecules with different sizes near a certain rough surface ... cl ∼ 1/d' with no derivation. This step is essential: substituting cl ∼ 1/d into L ∼ (cl/k)^{1/γ} gives L ∼ d^{-1/γ}, which turns Eq. (6) into Df = 2 + 1/γ. But cl is a property of the solid surface, not of the adsorbate molecule; no physical mechanism or multi-probe measurement supports the inverse proportionality. All inversion results use a single adsorbate (N2 at 77 K), so the d-dependence is never tested. Without cl ∼ 1/d, Eq. (7) is not a consequence of RS-DFT; it is a fitted exponent relabeled as a fractal dimension. The agreement of the resulting numbers with SAXS values is therefore a consistency check, not independent confirmation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an inversion workflow in which low-pressure nitrogen adsorption isotherms on five silica materials are fitted with random-surface density functional theory (RS-DFT). The fitted parameters are the normal-direction roughness variance var, the lateral correlation length cl, and the specific surface area A, the last of which is scanned over a range centered on the BET value. The authors report best-fit roughness parameters, compare the resulting surface areas with BET values, propose a storage-capacity metric, and argue that the correlated random surface model yields a surface fractal dimension Df = 2 + 1/γ through the power-law cl = kL^γ with L = sqrt(A). The central claim is that the fractal dimensions observed in SAXS experiments are natural consequences of the model.","tokens_in":8830,"tokens_out":7149,"duration_ms":70511,"significance":"If the fractal-dimension derivation were rigorous, the paper would connect routine adsorption measurements to independently observed fractal dimensions and provide a route to realistic explicit surfaces for atomistic simulation. The use of a correlated random process model, the low-pressure fitting protocol, and the material-specific parameter tables are genuine strengths, and the comparison with published roughness parameters is useful. However, the central Df claim rests on two assumptions that are not derived: the power-law scaling cl = kL^γ extracted from a narrow scanning interval, and the inverse proportionality cl ∼ 1/d. Without these, Eq. (7) is a fitted exponent relabeled as a prediction rather than a model-derived result. The inversion and material characterization parts of the paper are more incremental but could stand after the fractal-dimension claim is either properly supported or removed.","major_comments":[{"comment":"The step from cl = kL^γ to Df = 2 + 1/γ requires the assertion cl ∼ 1/d, which is stated without derivation. Since cl is a property of the solid surface, not of the adsorbate, the inverse proportionality to the molecular diameter is not a consequence of the RS-DFT model as presented. All fits use nitrogen at 77 K, so the d-dependence is never tested. Without a derivation of cl(d) or a multi-adsorbate experiment, Eq. (7) is an assumption, not a prediction of the model.","section":"Paragraph following Fig. 4 and Eq. (7)"},{"comment":"The power-law cl = kL^γ is fitted over A ∈ [0.75ABET, 1.25ABET], so L = sqrt(A) spans only a factor of about 1.29, while cl is restricted to [3d, 7d]. A two-parameter power-law exponent extracted from such a narrow, bound-limited window is not robust. No uncertainties, residuals, or goodness-of-fit measures are reported for γ, yet Df is quoted to two decimals. Provide a sensitivity analysis with respect to the fitting range and the imposed bounds on cl.","section":"Fig. 4 and Table I"},{"comment":"The definition Df = -lim log N_d/log d + 1 with N_d = L/d assumes a smooth one-dimensional measure. For a rough line one expects the number of segments of length d to scale as (L/d)^{D_line}. As written, Eq. (6) gives Df = 2 for any finite L that does not depend on d, so the entire fractal content is carried by the unproved relation L ∼ d^{-1/γ}. The derivation should be reconciled with the standard yardstick definition of fractal dimension.","section":"Eq. (6) and surrounding text"},{"comment":"The specific length L = sqrt(A) is defined from the fitted projected area A, which is itself a free parameter scanned over an arbitrary range. The scaling cl = kL^γ is therefore a relation between two outputs of the inversion procedure rather than a directly measured geometric relation of the material. The authors should clarify what physical length L represents and justify why a power-law relation between cl and sqrt(A) is expected from the model, rather than being an artifact of the chosen scanning interval.","section":"Section 'Our analysis is similar...' and Fig. 2"},{"comment":"The reported coincidence with SAXS values (Df ≈ 2.4 for silica glass) is presented as validation, but the extracted Df values all fall in the narrow range 2.30–2.54, so the comparison is not very discriminative. Moreover, because the Df values derive from fits to adsorption data on the same samples, agreement with independent SAXS measurements is only a consistency check; it does not validate the cl ∼ 1/d assumption.","section":"Table I and Fig. 5A"}],"minor_comments":[{"comment":"The caption refers to 'the colors of the materials are the same as those in Fig. A', but Fig. A is not defined; the intended reference appears to be the left panel of Fig. 5.","section":"Fig. 5 caption"},{"comment":"There is a duplicated paragraph beginning 'The best fit roughness parameters var and cl corresponding to the specific surface area ARS...'; one copy should be deleted.","section":"Text between Fig. 5A and Fig. 5B"},{"comment":"Typographical errors: 'o Additionally' in the introduction and 'thet correlation length' in the Fig. 3 caption should be corrected.","section":"Introduction and Fig. 3 caption"},{"comment":"The sentence 'the random surface model results in a fractal dimension without assumptions about spatial scale self-similarly or self-affinity' has a grammatical error; 'self-similarly' should be 'self-similarity'.","section":"Section following Eq. (7)"},{"comment":"The fitting parameters k and γ are used in Table I before being introduced in the text; define them when the power-law ansatz cl = kL^γ is first presented.","section":"Table I"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the RS-DFT inversion and material characterization portions are publishable in principle, but the fractal-dimension claim is presented as a derivation while relying on an unproved cl ∼ 1/d assumption. If the authors cannot supply a derivation or multi-adsorbate experimental test, I would either reject the paper or require it to be substantially reframed without the Df prediction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nBottom line: this one has a useful applied core and a load-bearing flaw in the headline result. The authors take their RS-DFT model, fit it to low-pressure nitrogen adsorption on several silica materials, and report roughness variance var, correlation length cl, and an 'ARS' specific surface area that can differ from BET. That part is fine in principle, and the new experimental isotherms for CPG-500/1000/3000 and Varapor-100 are a real plus. The roughness parameters do seem to track synthesis conditions (CPG-3000 smoother than CPG-500), which gives the work some credibility.\n\nThe problem is the fractal dimension. The paper claims that the correlated random surface model 'results in' Df = 2 + 1/γ, but this is not a derivation from the physics. The step cl ~ 1/d, introduced in the paragraph after Fig. 4, is asserted with no support, and all the data are for a single adsorbate (N2 at 77 K), so the d-dependence is never probed. Without cl ~ 1/d, Eq. (7) does not follow from Eq. (6). On top of that, the power-law cl = k L^γ is fit to a curve generated by scanning A over 0.75 to 1.25 ABET. That is only a factor ~1.29 in L, and the fit is constrained by the bounds 3d ≤ cl ≤ 7d used in the minimization. A power-law exponent from such a narrow, bound-limited window is not an independent material property. The Df values then match SAXS literature numbers, but that is a consistency check, not confirmation.\n\nI agree with the stress-test note: the fractal dimension is essentially a fitted parameter relabeled as a prediction. The paper also lacks error bars, sensitivity analysis, and released data, so the reproducibility of the inversion is hard to assess.\n\nWhere I'd push back on the reject verdict is the rest of the paper. The ARS vs ABET discussion and the methane storage contours are useful, and the measured isotherms have standalone value. If the fractal dimension section were removed or reframed as a speculative scaling observation with a proper caveat, the paper could be a solid methods contribution to an adsorption journal.\n\nFor peer review: this deserves a serious referee, not a desk reject, because the methodological question is technical and the authors should have the chance to defend or withdraw the cl ~ 1/d step. But in its current form I would recommend rejection. I would not cite the fractal dimension result.","headline":"The roughness-parameter inversion is plausible, but the paper's headline fractal dimension is a fitted exponent presented as a derivation, and the unproved cl ~ 1/d step breaks it.","tokens_in":9376,"tokens_out":3123,"would_cite":false,"duration_ms":30062,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Nanoscale roughness of silica can be read from low-pressure adsorption isotherms, and the fractal dimension near 2.4 emerges from the correlated random surface model without self-similarity assumptions.","keywords":["surface roughness","adsorption isotherm","random surface density functional theory","fractal dimension","silica glass","correlated random process","specific surface area","BET method"],"falsifier":"Measure adsorption isotherms for the same silica sample with two molecular sizes (e.g., N2 and Ar) and apply the paper's fitting scheme to each. Because the underlying surface is unchanged, Eq. (7) must yield the same fractal dimension for both; a systematic $D_f$ shift between adsorptives would falsify the $c_l \\propto 1/d$ assumption and hence the derived fractal dimension. Alternatively, direct AFM or GISAXS imaging of the same samples should reproduce the best-fit $\\mathrm{var}$ and $c_l$; if real-space correlation lengths disagree, the inversion is not recovering the actual geometry.","tokens_in":8208,"feed_emoji":"📏","tokens_out":9877,"duration_ms":82740,"temperature":0.7,"pith_summary":"The paper claims that a low-pressure nitrogen adsorption isotherm contains enough information to reconstruct the nanoscale geometry of a rough surface—its normal-direction variance, lateral correlation length, and true covering area—by fitting a random-surface density functional theory (RS-DFT) model. It goes on to derive the surface fractal dimension $D_f = 2 + 1/\\gamma$ from the same model, where $\\gamma$ is the power-law exponent tying correlation length to the square root of the fitted area, and shows that this yields $D_f \\approx 2.4$ for silica glass, matching SAXS measurements. The fitted roughness parameters also reproduce the experimentally observed 15–20 Å roughness region and vary with synthesis conditions in the expected direction. This matters because it turns a routine adsorption measurement into a source of full 3D surface morphology, ready for atomistic simulation and for screening gas storage materials.","feed_headline":"One scaling exponent from adsorption yields silica's fractal dimension","feed_subtitle":"Random-surface theory turns nitrogen isotherms into roughness maps that match silica-glass experiments.","key_machinery":"The load-bearing object is the RS-DFT adsorption model for a correlated Gaussian random surface, whose geometry is fixed by the outward variance $\\mathrm{var}$, the lateral correlation length $c_l$, and the flat-plane area $A$. The effective wall potential $U_{\\mathrm{eff}}(z,\\mathrm{var},c_l)$ and the available-surface function $S(z) = A\\left(1 - \\tfrac{1}{2}\\operatorname{erfc}\\left(z/\\sqrt{2\\,\\mathrm{var}}\\right)\\right)$ turn an isotherm into fitted $(\\mathrm{var}, c_l, A)$ triplets. The identity $D_f = 2 + 1/\\gamma$, obtained by fitting the inversion curve $c_l = k L^\\gamma$ and assuming $c_l \\propto 1/d$, converts a fitted exponent into a fractal dimension without any self-similarity assumption.","core_discovery":"On the paper's own terms, the central discovery is that the fractal dimension seen in adsorption experiments on silica does not require a self-similar or self-affine surface; it emerges naturally from a correlated Gaussian random surface once the correlation length scales as $c_l = k L^\\gamma$ with $L = \\sqrt{A}$, and as $c_l \\propto 1/d$ for the adsorbate molecular diameter $d$. Combining these scalings with the box-counting definition of fractal dimension gives $D_f = 2 + 1/\\gamma$. Fitting low-pressure ($P/P_0 < 0.1$) nitrogen isotherms at 77 K for five silica materials yields $D_f$ values from 2.3 to 2.54, in good agreement with published values near 2.4, and the best-fit roughness region $2\\delta \\approx 17$ Å lies inside the 15–20 Å range reported for silica glass. The same fits return a specific surface area $A_{\\mathrm{RS}}$ that can be larger or smaller than the BET area, which the authors interpret as a direct signature of how monolayer adsorption is distorted by surface roughness.","pith_inferences":["A direct test of the model's core assumption would be to measure the same silica sample's isotherm with several adsorptives (N2, Ar, CH4): if $c_l \\propto 1/d$ is physical, Eq. (7) must return the same $D_f$ for each; any variation would locate the failure in the molecular-size scaling, not the density functional.","If the $c_l = k L^\\gamma$ scaling generalizes beyond silica, the same three-parameter fit could map roughness and fractal dimension for carbons, oxides, or polymers from routine adsorption data, making fractal analysis a lab-bench measurement rather than a synchrotron one.","The paper's inversion treats each isotherm independently; coupling the fits across samples of the same material family (e.g., all CPG glasses) could tighten the constraint on $\\gamma$ and reveal whether the power-law relation is a genuine material law or a numerical envelope of the fitting procedure."],"forward_implications":["Low-pressure adsorption isotherms ($P/P_0 < 0.1$) can be inverted into variance, correlation length, and specific surface area, yielding a complete rough-surface geometry for atomistic modelling.","For silica glass, the correlated random surface model predicts fractal dimensions near 2.4, in line with SAXS/Porod experiments, without invoking scale self-similarity.","The BET surface area is not a neutral measure: the model predicts $A_{\\mathrm{BET}} > A_{\\mathrm{RS}}$ on very rough surfaces and $A_{\\mathrm{BET}} < A_{\\mathrm{RS}}$ on smoother ones, quantifying when the monolayer assumption fails.","The storage criterion $C = A_{\\mathrm{RS}} N_{\\mathrm{ads}}/N_0$ ranks mesoporous silica samples for methane uptake, showing that roughness enhances storage but the effect is offset by the true surface area."],"supporting_citations":[{"why":"Establishes RS-DFT and provides the density equation used to fit the isotherms.","marker":"[17]"},{"why":"Derives the effective solid–fluid potential for correlated random surfaces used in the RS-DFT equation.","marker":"[25]"},{"why":"Defines surface fractal dimension from line length measured in molecular diameters, the basis of Eq. (6).","marker":"[34]"},{"why":"Reports SAXS fractal dimension 2.4 and roughness region 15–20 Å for silica glass, the experimental benchmark.","marker":"[7]"},{"why":"Provides published nitrogen adsorption data for LiChrospher Si-1000, one of the five test materials.","marker":"[31]"},{"why":"Defines BET surface area and the allowed range $0.75A_{\\mathrm{BET}} \\le A \\le 1.25A_{\\mathrm{BET}}$.","marker":"[28]"},{"why":"Supports treating specific surface area as a fitting parameter, with BET underestimating the true area.","marker":"[30]"},{"why":"Independently found a specific surface area of 24 m²/g for Si-1000, matching the paper's $A_{\\mathrm{RS}}$ value.","marker":"[35]"}],"fun_headline_variants":["Adsorption isotherms decode silica's nanoroughness","Random-surface theory fits silica's fractal dimension","Gas adsorption maps silica's surface roughness","One exponent from isotherms yields silica roughness","Silica's fractal dimension emerges from adsorption"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The fractal dimension formula depends on the assumed power-law growth of correlation length with $\\sqrt{A}$ and its inverse proportionality to the adsorbent molecule's diameter, both introduced as fits or assumptions rather than derived from the physics; if either scaling is a numerical artifact of the inversion, the predicted fractal dimension is not a real surface property.","fun_headline_variants_meta":{"raw":{"variants":["Adsorption isotherms decode silica's nanoroughness","Random-surface theory fits silica's fractal dimension","Gas adsorption maps silica's surface roughness","One exponent from isotherms yields silica roughness","Silica's fractal dimension emerges from adsorption"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001052,"raw_usage":{"total_tokens":4439,"prompt_tokens":984,"completion_tokens":3455,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":3386}},"tokens_in":600,"tokens_out":3455,"duration_ms":25962,"temperature":1.0,"reasoning_tokens":3386,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:07:43.588986+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure adsorption isotherms for the same silica sample with two molecular sizes (e.g., N2 and Ar) and apply the paper's fitting scheme to each. Because the underlying surface is unchanged, Eq. (7) must yield the same fractal dimension for both; a systematic $D_f$ shift between adsorptives would falsify the $c_l \\propto 1/d$ assumption and hence the derived fractal dimension. Alternatively, direct AFM or GISAXS imaging of the same samples should reproduce the best-fit $\\mathrm{var}$ and $c_l$; if real-space correlation lengths disagree, the inversion is not recovering the actual geometry.","supporting_citations":[{"cited_title":"Aslyamov and A","cited_arxiv_id":null,"evidence_quote":"Establishes RS-DFT and provides the density equation used to fit the isotherms."},{"cited_title":"Khlyupin and T","cited_arxiv_id":null,"evidence_quote":"Derives the effective solid–fluid potential for correlated random surfaces used in the RS-DFT equation."},{"cited_title":"Farin, S","cited_arxiv_id":null,"evidence_quote":"Defines surface fractal dimension from line length measured in molecular diameters, the basis of Eq. (6)."},{"cited_title":"Levitz, G","cited_arxiv_id":null,"evidence_quote":"Reports SAXS fractal dimension 2.4 and roughness region 15–20 Å for silica glass, the experimental benchmark."},{"cited_title":"Jaroniec, M","cited_arxiv_id":null,"evidence_quote":"Provides published nitrogen adsorption data for LiChrospher Si-1000, one of the five test materials."},{"cited_title":"Thommes, K","cited_arxiv_id":null,"evidence_quote":"Defines BET surface area and the allowed range $0.75A_{\\mathrm{BET}} \\le A \\le 1.25A_{\\mathrm{BET}}$."},{"cited_title":"Ustinov, D","cited_arxiv_id":null,"evidence_quote":"Supports treating specific surface area as a fitting parameter, with BET underestimating the true area."},{"cited_title":"Ustinov, D","cited_arxiv_id":null,"evidence_quote":"Independently found a specific surface area of 24 m²/g for Si-1000, matching the paper's $A_{\\mathrm{RS}}$ value."}],"review_version":1}