{"id":"fe8f86f4-7a63-450e-acd7-72fc67b522a5","arxiv_id":"1908.08780","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For a single 2D domain fed by diffusing adsorbates, the Stefan problem reduces to a closed equation whose solution predicts area growing roughly linearly with time, with a logarithmic correction.","lead":"This paper derives a scaling theory for how a circular solid domain grows on a surface as diffusing adsorbates attach to it, including deposition and desorption. It gives growth-rate formulas that can be used to interpret chemical vapor deposition of graphene and other two-dimensional materials.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Stefan condition drops the cumulative deposition term gt without quantitative justification, and with the paper's own parameter estimates that term is O(1) in the same regime where the Kummer expansion used for Eq. (28) is applied.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the neglect of gt in going from Eq. (5) to Eq. (6). I re-derived this step and could not find an error in the algebra itself; the issue is the validity of the neglect. My reading sharpens the concern by connecting it to the paper's own asymptotic approximations. The road from Eq. (23) to Eq. (28) requires ψ[1+q/(4α)] ≈ ln(q/(4α)), i.e. kd t large. But then gt/ρ = (C∞/ρ) kd t is not small; with C∞/ρ ≈ 2.6×10^-4, kd t = 4×10^3 makes gt/ρ of order unity. Thus the same time regime in which the closed-form growth law is derived is the regime in which the dropped term should be retained. There is also a related time-dependence of q = kdR²/D = 4αkd t, so the scaling solution with constant α is at best adiabatic; the error of that approximation is not estimated. These are correctness risks rather than internal algebraic failures, and they are addressable by a direct numerical check. They do not, by themselves, force a rejection, because there may be parameter windows where gt remains negligible over the times of interest; the paper simply has not established such a window. The reader's CONDITIONAL verdict is therefore appropriate, and my stress-test does not move it.","tokens_in":12170,"tokens_out":20601,"duration_ms":237849,"concrete_test":"Solve the full moving-boundary problem numerically: Eq. (2) with the Stefan condition (5) retaining the gt term, for the paper's graphene parameters D = 10^-6 m²/s, ρ = 3.82×10^19 m^-2, C∞ = g/kd = 10^16 m^-2, and target growth rate R²/t = 10^-12 m²/s. Infer kd self-consistently from Eq. (28) at R² = 10^-6 m² (order 0.5 s^-1), then integrate over t = 10^3 to 10^4 s. Compare the resulting R²(t) against 4αDt with α from Eq. (23). If retaining gt changes R²(t) by more than about 10% at these times, the neglect is quantitatively invalid in the claimed regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central derivation passes from Eq. (5) to Eq. (6) by dropping the cumulative-deposition term 2πR(t)gt dR/dt, justified only by the assertion that gt is small compared with ρ and C(R(t),t). This is load-bearing because Eq. (6) is the Stefan condition that fixes α in Eq. (23), and hence the predicted growth constant and the diffusion-limited versus reaction-limited diagnosis. No quantitative regime for the neglect is given. The problem is sharpened by the later asymptotics: to reach Eq. (28), ψ[1+q/(4α)] is replaced by ln[q/(4α)], which requires q/(4α) = kd t to be large. At such times, gt/ρ = (C∞/ρ) kd t with C∞ = g/kd. Using the paper's own estimates C∞/ρ ≈ 10^-3, kd t ~ 10^3 already makes gt/ρ of order 0.1-1, so the neglected term is not small in exactly the regime where the analytical expansion is used. Relatedly, q = kdR²/D = 4αkd t is time-dependent, so α obtained from Eq. (23) is not a true constant and R²(t) = 4αDt is an adiabatic approximation rather than an exact asymptotic scaling relation. The paper does not state or justify this adiabatic approximation. The printed sign issue in Eq. (24) appears typographical and is compensated in Eq. (26), so it is not the primary concern.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a Stefan-problem theory for the growth of a single circular two-dimensional solid domain fed by diffusing adsorbates, including deposition and desorption on the substrate. The central result is Eq. (23), a closed transcendental equation for the growth-rate constant α defined by R²(t)=4αDt, together with approximate closed-form expressions for R²(t) in the small-α, small-q regime [Eqs. (28) and (37)-(38)] and a generalization to finite attachment rate k. The theory is benchmarked against numerical solution of the exact transcendental equations in Figs. 2-4, and the results are compared with graphene chemical-vapor-deposition growth data. The paper also discusses the shrinking of domains when deposition is stopped.","tokens_in":12462,"tokens_out":10536,"duration_ms":104773,"significance":"If the derivation is valid, the paper provides a useful, parameter-free (up to physical constants) closed equation for the area-growth constant of a single 2D domain, and it clarifies the crossover between diffusion-limited and reaction-limited growth. The numerical benchmark of the approximate formulas against the exact Eq. (23) is a genuine strength, as is the explicit treatment of desorption to regularize the two-dimensional diffusion problem. The finite-attachment-rate extension in Sec. 3 is a natural and valuable generalization. The comparison with graphene CVD is suggestive, though the experimental uncertainties are acknowledged.","major_comments":[{"comment":"The neglect of the cumulative deposition term gt in the Stefan boundary condition is unquantified and is load-bearing because Eq. (6) determines α through Eq. (23). At time t this term is of order (g/ρ)t = (C∞/ρ)kd t. Using the paper's own estimate C∞/ρ ≈ 10⁻³, the term is O(0.1-1) when kd t ≈ 10³, which is precisely the regime in which the digamma expansion leading to Eq. (28) requires q/(4α)=kd t to be large. The paper should state the validity window for dropping gt (e.g., kd t << ρ/C∞) and demonstrate that the experimental and benchmark regimes lie inside it; otherwise the α obtained from Eq. (23) is not the growth constant of the original problem.","section":"§2, Eqs. (5)-(6)"},{"comment":"The scaling solution assumes that α is a constant, but q = kdR²/D = 4αkd t is time-dependent. Therefore Eq. (23) is an equation for α(q(t)), not for a fixed constant, and R²(t)=4αDt is an adiabatic approximation rather than an exact asymptotic scaling relation. The paper does not state or justify this approximation. The actual growth law would be dR²/dt = 4D α(kdR²/D), and Eq. (28) is an implicit equation for R²(t) rather than a direct statement that area is proportional to time. The author should either quantify the adiabatic error or reformulate the central result as the solution of this differential equation.","section":"§2, Eqs. (11)-(23)"},{"comment":"The numerical benchmarks in Figs. 2-4 compare approximate and exact expressions by plotting 4α/q versus 1/q, treating q as an independent parameter. In the physical problem, however, q and α are linked through q = 4αkd t. The figures therefore demonstrate agreement of the algebraic formulas at fixed q, but they do not validate the time-dependent trajectory R(t) predicted by the theory, which is the quantity used in the experimental comparison. The author should provide a benchmark of the full time evolution, or state explicitly that only the instantaneous algebraic relation is being tested.","section":"§2 and Figs. 2-4"}],"minor_comments":[{"comment":"Equation (24) has a sign error as printed: the standard small-z form is U(a,1,z) ≈ [ln z - ψ(a) - 2γ]/Γ(a). The subsequent Eq. (26) uses the corrected sign, so this appears typographical, but it should be fixed.","section":"Eq. (24)"},{"comment":"The statement that 'the right-hand side of equation (28) yields an estimate 10⁻⁹ m²/s whereas the left-hand side of equation (28) is 10⁻¹² m²/s' is confusing because Eq. (28) is an implicit equation for R²(t), so the two sides are not independent estimates. Please clarify the intended comparison.","section":"§4, numerical estimate"},{"comment":"The symbol q is defined as the dimensionless desorption rate kdR²/D, but R depends on time, so q is time-dependent. This should be stated explicitly at first use to avoid the impression that q is a constant parameter.","section":"Eq. (10) and throughout"},{"comment":"The figure captions contain apparent label corruption (e.g., '/s32/s33/s34'), which makes the figures difficult to read. The figures should be regenerated with proper axis labels.","section":"Figure captions"},{"comment":"Reference [16] is incomplete; the publisher, year, and place of publication for the Gupta book should be added.","section":"Reference [16]"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious theoretical contribution with a clean derivation of the transcendental equation for the growth constant, and the numerical checks of the algebraic approximations are valuable. The main risk is that the central scaling claim R²∝t is not fully established because of the unquantified neglect of the gt term and the implicit adiabatic treatment of the time-dependent q. These issues are fixable within the manuscript's scope by adding quantitative validity estimates and, ideally, a numerical solution of the full time-dependent problem. I therefore recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper. First, it derives a closed transcendental equation, (23), for the growth-rate constant of an isolated circular 2D domain growing by attachment of diffusing adsorbates, with deposition and desorption. That equation, and its finite-attachment generalization (36), are genuinely new relative to the cited inside-diffusion Stefan problems. The algebra from the diffusion equation to the Kummer-function form is consistent, and the approximate growth laws (28), (38), and (39) are sensible and reduce to the expected diffusion- and reaction-limited limits. Second, the paper rests on a scaling ansatz that is not exact for this problem, and the authors do not say so plainly. The quantity q = kdR^2/D = 4αkd t is time-dependent, so the profile C(ξ) with ξ = r/R(t) cannot solve the original PDE for all t unless desorption is handled in a quasi-static way. The paper moves from Eq. (9) to Eq. (13) as if q were a constant, then later lets R appear in the logarithmic terms of (28). This is an adiabatic approximation, not an exact asymptotic scaling law, and the paper never states or justifies it. The stress-test note sharpens this: the ψ[1+q/(4α)] ≈ ln[q/(4α)] expansion used to reach (28) requires kd t large, and with the paper's own estimate C∞/ρ ≈ 10^-3, the dropped gt term in the Stefan condition (Eq. (5) to (6)) becomes O(1) at exactly those times. The statement 'we assume that the concentration of accumulated adsorbates on the substrate is small compared with ρ' is thus not harmless; it is load-bearing and unquantified. The 'exact numerical' curves in Figures 2–4 solve (23) with constant q, so they validate the transcendental algebra, not the full time-dependent PDE. A sign issue in Eq. (24) seems typographical, since Eq. (26) is consistent. The experimental comparison is semiquantitative, as the authors acknowledge. So: the paper is a solid, honest extension of Stefan-scaling methods, not a breakthrough, and its central results are likely correct in some regime, but that regime is not defined. A serious referee could usefully ask for a quantitative criterion for neglecting gt and for a statement of the adiabatic approximation, possibly with a direct comparison to a numerical solution of the coupled PDE and moving boundary. I would send it to review rather than desk-reject: the closed equation is new, the derivations are checkable, and the weaknesses are addressable rather than fatal.","headline":"A careful Stefan-problem derivation of 2D domain growth, new in its outside-diffusion geometry with deposition/desorption, but the scaling ansatz is a quasi-static approximation whose validity is not quantified.","tokens_in":12968,"tokens_out":4230,"would_cite":false,"duration_ms":47589,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35R35","80A22","82C24"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper derives a closed transcendental equation for the growth-rate constant α in the asymptotic law R²(t)=4αDt for a circular 2D domain growing by attachment of diffusing adsorbates with deposition and desorption.","keywords":["Stefan problem","domain growth","two-dimensional diffusion","deposition and desorption","Kummer confluent hypergeometric function","reaction-limited growth","diffusion-limited growth","graphene CVD"],"falsifier":"Run a kinetic Monte Carlo simulation of a circular island grown by diffusing adsorbates with deposition and desorption, and compare the simulated area growth constant with the solution of equation (23) for the same parameters; if the simulated $\\alpha$ deviates systematically as $gt$ grows, the approximation of dropping the $gt$ term is falsified.","tokens_in":83,"feed_emoji":"📈","tokens_out":10183,"duration_ms":170815,"temperature":0.7,"pith_summary":"The paper derives the asymptotic growth law for an isolated circular two-dimensional domain that grows by capturing adsorbates diffusing on a substrate while deposition and desorption continue. The central claim is that the area grows linearly with time, $R^2(t) = 4\\alpha D t$, where the growth-rate constant $\\alpha$ is fixed by a closed transcendental equation that depends on the two-dimensional diffusion constant, the deposition and desorption rates, the equilibrium boundary concentration, the solid density, and (for finite attachment) the attachment rate constant. Approximate analytical expressions are obtained for both diffusion-limited and reaction-limited regimes, and the reaction-limited form reduces to $R^2(t) \\approx (k/\\pi)(C_\\infty - C_0)/\\rho\\,t$. Because the proportionality constant is expressed in terms of measurable kinetic parameters, the theory offers a way to identify the controlling step from the temperature and source-gas-flow dependence of the area growth, which is the motivation from graphene chemical vapour deposition experiments.","feed_headline":"One equation fixes the area growth of 2D islands","feed_subtitle":"The constant α comes from diffusion, deposition, and attachment rates, not from fitting.","key_machinery":"The central object is the scaling reduction of the two-dimensional Stefan problem. Assuming the adsorbate concentration depends only on $\\xi = r/R(t)$, the consistency of the time-dependent diffusion equation forces $R(dR/dt) = 2\\alpha D$, giving $R^2(t) = 4\\alpha D t$. The spatial profile then satisfies Kummer's differential equation, with solution $C(\\xi) = C_1 \\exp(-\\alpha\\xi^2)\\,U(1+q/(4\\alpha),1,\\alpha\\xi^2) + g/k_d$, where $U$ is the confluent hypergeometric function of the second kind. The Stefan condition at the moving boundary, $[\\rho - C(1)]\\,dR/dt = D\\,\\partial C/\\partial r$ evaluated at $r=R(t)$, together with the value at the boundary $C(1) = C_0$, yields the closed transcendental equation for $\\alpha$. For finite attachment, the boundary condition $\\partial C/\\partial r = k(C - C_0)/(2\\pi R D)$ is combined with the same Stefan condition to produce a second closed equation. The use of Kummer functions converts the moving-boundary problem into a single algebraic equation, which is the mechanism that makes the growth rate computable from physical parameters.","core_discovery":"This paper establishes that for an isolated circular solid domain on a two-dimensional substrate, the asymptotic growth driven by diffusing adsorbates with deposition and desorption is $R^2(t) = 4\\alpha D t$, with $\\alpha$ determined not by a fitting procedure but by the closed equation $\\alpha(\\rho - C_0)\\,U(1+q/(4\\alpha),1,\\alpha) = [(g/k_d) - C_0]\\,U(q/(4\\alpha),0,\\alpha)$. Here $D$ is the adsorbate diffusion constant, $\\rho$ the areal density of the solid, $C_0$ the boundary concentration at local equilibrium, $g$ the deposition rate, $k_d$ the desorption rate, and $q = k_d R^2/D$. The same approach yields approximate growth laws: in the diffusion-controlled limit $R^2(t) \\approx 4Dt\\,[\\ln(4D/(k_d R^2)) - 2\\gamma]^{-1}(C_\\infty - C_0)/\\rho$, and in the reaction-controlled limit $R^2(t) \\approx (k/\\pi)(C_\\infty - C_0)/\\rho\\, t$, where $C_\\infty = g/k_d$ is the uniform far-field adsorbate concentration. For a finite attachment rate $k$, the paper shows that the Stefan boundary condition still holds and derives a companion closed equation for $\\alpha$. The paper also predicts the area decrease after stopping the source gas flow, and argues that the observed smaller shrinking rate is consistent with the theory when $C_0 < g/(2k_d)$.","pith_inferences":["The theory implies that fitting $R(t)$ versus $t$ alone cannot reliably distinguish the $R(t)\\propto t$ and $R^2(t)\\propto t$ regimes, because the Stefan-derived linear-area law and the constant-concentration law can both appear linear over short time windows; measuring the boundary concentration gradient would settle the mechanism.","The least-controlled step is the neglect of the accumulated deposition $gt$ in the Stefan condition; retaining this term would produce a time-dependent growth constant, so a direct testable extension is to check whether the growth law becomes sub-linear at long times when $gt$ grows.","If the single-domain theory is embedded in a population of domains, the effective $\\alpha$ will decrease when diffusive crowding sets in; the paper's isolated-domain law thus serves as the early-time limit of a coarsening theory, not the full surface-coverage kinetics."],"forward_implications":["The linear area-versus-time growth observed in graphene chemical vapour deposition is a direct consequence of the Stefan condition, which makes the area growth rate independent of the domain radius.","Because $\\alpha$ is a function of $D$, $k_d$, $g$, $C_0$, $\\rho$, and $k$, simultaneous measurements of the growth constant at different temperatures and source-gas flow rates can separate the activation energies of diffusion and attachment.","In the reaction-limited regime, $R^2(t)$ is proportional to the attachment rate $k$, so the theory predicts that the growth rate is linearly controlled by the attachment process, not by diffusion.","Stopping the source gas makes $g=0$, and the theory then predicts a decreasing area whose rate is smaller than the growth rate when $C_0 < g/(2k_d)$, matching the reported asymmetry in experiments.","The same scaling analysis is transferable to other near-circular two-dimensional domains, such as transition-metal dichalcogenide islands grown by chemical vapour deposition."],"supporting_citations":[{"why":"Supplies the experimental scaling relation and the observed asymmetry between area increase and decrease in graphene CVD that the theory seeks to explain.","marker":"[23]"},{"why":"Introduces the scaling-solution method for a Stefan problem solved by a similarity variable, which is adapted here to the exterior diffusion problem.","marker":"[11]"},{"why":"Gives the previous two-dimensional diffusion-controlled growth law $R^2(t)\\propto t$ that this paper generalizes to include deposition and desorption.","marker":"[12]"},{"why":"Shows how to apply scaling variables to moving-boundary wetting monolayer growth, providing the mathematical template used in the derivation.","marker":"[13]"},{"why":"Supplies the confluent hypergeometric identities and asymptotic expansions used to close the equation for $\\alpha$.","marker":"[35]"},{"why":"Provides the linear boundary condition for finite attachment and detachment at the moving boundary used in the generalized model.","marker":"[33]"},{"why":"Discusses the Stefan boundary condition for growth and the screening-length issue for desorption, guiding the neglect of the $C(R(t),t)$ term and the need for desorption.","marker":"[34]"},{"why":"Shows that desorption removes the divergence of the two-dimensional concentration profile in the far field, justifying the inclusion of $k_d$.","marker":"[25]"}],"fun_headline_variants":["Closed equation fixes 2D island growth rate","Diffusion and attachment set island growth, no fit","Scaling theory predicts 2D domain expansion","Growth constant from Stefan problem, not fitting","Two limits of 2D island growth unified"],"cache_read_input_tokens":15104,"weakest_assumption_plain":"The calculation assumes that the total amount of adsorbates deposited up to time $t$ is small compared with the solid density and the boundary concentration, and it does not specify when this neglect of the accumulation term $gt$ in the Stefan condition becomes invalid at long times.","fun_headline_variants_meta":{"raw":{"variants":["Closed equation fixes 2D island growth rate","Diffusion and attachment set island growth, no fit","Scaling theory predicts 2D domain expansion","Growth constant from Stefan problem, not fitting","Two limits of 2D island growth unified"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000222,"raw_usage":{"total_tokens":1513,"prompt_tokens":1066,"completion_tokens":447,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":682,"completion_tokens_details":{"reasoning_tokens":376}},"tokens_in":682,"tokens_out":447,"duration_ms":5000,"temperature":1.0,"reasoning_tokens":376,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:30:22.235806+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a kinetic Monte Carlo simulation of a circular island grown by diffusing adsorbates with deposition and desorption, and compare the simulated area growth constant with the solution of equation (23) for the same parameters; if the simulated $\\alpha$ deviates systematically as $gt$ grows, the approximation of dropping the $gt$ term is falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the experimental scaling relation and the observed asymmetry between area increase and decrease in graphene CVD that the theory seeks to explain."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the scaling-solution method for a Stefan problem solved by a similarity variable, which is adapted here to the exterior diffusion problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the previous two-dimensional diffusion-controlled growth law $R^2(t)\\propto t$ that this paper generalizes to include deposition and desorption."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows how to apply scaling variables to moving-boundary wetting monolayer growth, providing the mathematical template used in the derivation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the confluent hypergeometric identities and asymptotic expansions used to close the equation for $\\alpha$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the linear boundary condition for finite attachment and detachment at the moving boundary used in the generalized model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Discusses the Stefan boundary condition for growth and the screening-length issue for desorption, guiding the neglect of the $C(R(t),t)$ term and the need for desorption."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that desorption removes the divergence of the two-dimensional concentration profile in the far field, justifying the inclusion of $k_d$."}],"review_version":1}