{"id":"79a46b28-1953-4a6e-a560-5a49dc7e84fe","arxiv_id":"2411.13747","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Interfacial water next to graphene, rather than dissolved ions, sets the electric potential and can reverse its polarity, a result incorporated into an extended Poisson-Boltzmann model.","lead":"Molecular dynamics simulations show that the first water layers next to graphene are strongly oriented, and this surface-induced water polarization, not the ions, dominates the electric potential at the interface. The finding challenges the classic electric double layer picture and leads to an extended model that improves predictions of interfacial voltages and capacitance.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'surface-induced water polarization' mechanism depends on untested transferability of pure-water phi_pzc,H2O^0 to 0.8 M electrolytes; without it, chi_H2O in Eq. (1) is not cleanly a linear response to ions and the extended PB model's subtraction is unjustified.","rationale":"The reader's weakest assumption—transferability of phi_pzc,H2O^0 from pure water to electrolyte—is exactly the most load-bearing point. The direct Maxwell decomposition robustly shows that water contributes more than ions to the potential at graphene interfaces, and that result does not depend on the questionable decomposition. What the decomposition adds is the causal attribution: water polarization that is 'quenched' by the surface and does not screen ions. That attribution requires the pure-water surface polarization to survive at 0.8 M salt, which is not established by the presented evidence. The near-cancellation of chi_H2O and phi_pzc,ion is partly a consequence of the total potential being close to phi_pzc,H2O^0, so it does not independently confirm linear dielectric response. A direct orientational-profile comparison would settle the issue. Since the reader's CONDITIONAL verdict already flags this assumption and the paper is otherwise solid, no change to the verdict is needed.","tokens_in":9088,"tokens_out":7330,"duration_ms":77618,"concrete_test":"Re-analyze the existing MD trajectories: compute the distance-resolved water dipole orientation <cos theta_OH>(z) and the water z-dipole density for Gr-water and Gr-NaBF4 over the same 10 ns windows, with block-averaged error bars, and integrate the absolute difference over z < 10 Å. If the integrated difference is statistically nonzero, phi_pzc,H2O^0 is not transferable and chi_H2O in Eq. (1) mixes ion-induced restructuring of interfacial water with surface-induced polarization; the extended PB model would then need a concentration-dependent phi0 rather than the pure-water input.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The pivotal step is Eq. (1): phi_pzc,H2O = phi_pzc,H2O^0 + chi_H2O, with phi_pzc,H2O^0 taken from the pure Gr-water system. The paper's conceptual conclusion—that a significant portion of interfacial water responds directly to graphene rather than screening ions—depends on interpreting chi_H2O as the ion-induced perturbation. At 0.8 M, especially for Gr-NaBF4 where BF4- adsorbs at roughly three times the Na+ density, ions can restructure the interfacial water layer. Any such restructuring changes the surface-induced water polarization, and because chi_H2O is defined as the difference between electrolyte and pure-water water potentials, that change is silently absorbed into chi_H2O. The mirror-image relation between chi_H2O and phi_pzc,ion (Fig. 2c-d) is not decisive evidence for linear dielectric screening: it is partly guaranteed by the decomposition and by the observation that the total potential is close to phi_pzc,H2O^0. The authors support transferability by citing similar total-potential oscillations and similar O-H orientation maps, but they do not report a quantitative, uncertainty-aware comparison of water orientational/dipole distributions in pure water versus electrolyte. If transferability fails, the attribution of water polarization to the graphene surface is not established, and the extended PB model's subtraction of phi_pzc,H2O^0 (Eqs. 2-6) is unjustified, even though the direct Maxwell decomposition still shows water contributing more than ions to the potential.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript uses classical molecular dynamics (MD) simulations of graphene in pure water and in 0.8 M NaCl and NaBF4 electrolytes, for uncharged and charged graphene, to decompose the interfacial electric potential into ionic and water contributions by linear superposition of Maxwell's equations. The authors find that the water contribution to the potential at the potential of zero charge (PZC) is larger in magnitude than the ionic contribution, can reverse the sign of the total PZC (e.g., +1.71 V water vs -1.45 V ions in Gr-NaBF4), and that the total potential profiles are dominated by a component they attribute to surface-induced water polarization present even in the absence of ions. They then propose an extended Poisson-Boltzmann (PB) framework that explicitly subtracts the pure-water surface potential, and report improved agreement with MD potential profiles compared to classical PB and Gouy-Chapman-Stern models.","tokens_in":9502,"tokens_out":3363,"duration_ms":30921,"significance":"If the central interpretation holds, the paper challenges the long-standing treatment of water as a passive dielectric in electric double layer theory and provides a tractable modification of the PB equation for weakly polarizable electrodes. The direct Maxwell decomposition of potential into ionic and water contributions is a valid and useful diagnostic, and the comparison with experimental PZC values and with ab initio results strengthens the plausibility of the simulated water structure. The paper also delivers a falsifiable prediction (the dominance of the surface-induced water potential that persists in electrolytes) and a concrete model that can be tested on other surfaces and ion types. However, the significance is conditional on the transferability of the pure-water surface potential to concentrated electrolyte systems, which is currently supported only by qualitative evidence.","major_comments":[{"comment":"The decomposition phi_pzc,H2O = phi_pzc,H2O^0 + chi_H2O assumes that the pure-water surface-induced water polarization phi_pzc,H2O^0 is unchanged when 0.8 M ions are added. This is load-bearing for the conclusion that a significant portion of water polarization responds directly to the graphene surface rather than screening ions, because any ion-induced restructuring of interfacial water would be silently absorbed into chi_H2O. The paper supports transferability by noting similar total potential oscillations and O-H orientation maps (Fig. 1c and Fig. 2a-b), but these are qualitative comparisons. I request a quantitative, uncertainty-aware comparison of the water dipole/orientational distributions in pure water versus the electrolytes, or an explicit test in which phi_pzc,H2O^0 is recomputed in the electrolyte system after removing the ions, to establish that chi_H2O is a clean ion-induced perturbation.","section":"Interfacial water beyond a passive dielectric medium, Eq. (1)"},{"comment":"The mirror-image relation between chi_H2O and phi_pzc,ion is largely a tautological consequence of the decomposition. Because the text states that phi_pzc,total is close to phi_pzc,H2O^0 (e.g., the paragraph following Eq. (1)), Eq. (1) directly implies chi_H2O ≈ -phi_pzc,ion. Similarly, the ratio phi_pzc,ion/(phi_pzc,ion + chi_H2O) being close to epsilon_r is not an independent test of linear dielectric screening, since it is derived from the same approximate relation. The authors should provide a more direct test of linear response, for example by varying ion concentration or surface charge and checking whether chi_H2O scales proportionally with phi_pzc,ion in a parameter-free manner.","section":"Interfacial water beyond a passive dielectric medium, Fig. 2c-d"},{"comment":"The model's prediction of the PZC (0.34 V) is dominated by the MD-derived input phi_pzc,H2O^0 (0.33 V), so the close agreement with the MD value of 0.33 V is substantially built into the model rather than being an independent validation. Additionally, Eq. (6) relies on the assumption that non-electrostatic changes in the ion PMF compensate Delta_phi_H2O,II^0, which is asserted from Becker et al. but not directly verified for the present systems; the manuscript itself concedes 'More work should be done in future for this non-electrostatic interaction.' To support the claim of predictive capability, the model should be tested on a system or observable not used in constructing the inputs, or the sensitivity of the predictions to the transferability and compensation assumptions should be quantified.","section":"The extended Poisson-Boltzmann EDL models, Eqs. (4)-(6)"}],"minor_comments":[{"comment":"The title contains a typographical error: 'Polari zation' should be 'Polarization.'","section":"Title"},{"comment":"The sentence 'The horizontal white line divides the map into bonds pointing away from the group (termed H-outward) and toward the surface (H-inward)' uses 'group' where 'surface' or 'graphene' appears intended.","section":"Figure 1 caption"},{"comment":"Reference 23 is incomplete: 'The Journal of Chemical Physics 157 (2022)' lacks the article number or page range; please provide the full citation.","section":"Reference 23"},{"comment":"The notation phi_pzc,H2O^0 and the subscript/superscript formatting are occasionally inconsistent (e.g., in Eq. (4) and the surrounding text); standardizing the notation would improve readability.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The reader's report and the stress-test concern about transferability are well-founded and align with my reading of the manuscript. The central claim is interesting and potentially important, but the current evidence does not yet exclude the alternative explanation that ions restructure the interfacial water layer, in which case chi_H2O is not a clean linear response and the extended PB model's subtraction of phi_pzc,H2O^0 is unjustified. I would like the authors to address the transferability test and the circularity of the mirror-image argument before I can recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a careful MD study that makes a genuinely striking point. At uncharged graphene, the water contribution to the potential is ~1.7 V in NaBF4, swamping the ionic contribution of -1.45 V, and the total potential profile is nearly identical to pure water. If that holds up, the standard GCS picture is incomplete at these interfaces. The Maxwell decomposition into ion and water parts is straightforward and well executed, and the mirror-image relation between chi_H2O and phi_ion, with a ratio near epsilon_r, is a nice, self-consistent observation. The authors also cite the relevant prior work on dielectric overscreening and interfacial water structure; the novelty is the quantitative decomposition and the extension of PB.\n\nWhere I'd press: the decomposition phi_pzc,H2O = phi_pzc,H2O^0 + chi_H2O assumes the pure-water surface-induced polarization is unchanged when 0.8 M salt is added. The stress-test note is fair: they show similar total potentials and O-H orientation maps, but they don't report an uncertainty-aware comparison of water orientation/dipole distributions in pure water versus electrolyte. In NaBF4, BF4- sits at the interface at three times the Na+ density, so some restructuring of the first water layer is plausible. That doesn't kill the central claim, because phi_pzc,H2O^0 is large and the near-identity of total potentials suggests it's the dominant term, but it does blur the clean interpretation of chi_H2O as a purely ionic linear response, and it weakens the subtraction step in the extended PB model.\n\nThe second soft spot is the model's self-referentiality. The predicted PZC of 0.34 V vs MD 0.33 V is impressive, but the model consumes phi_H2O^0 and the ion PMFs from the same MD, so this is consistency, not prediction. The authors do show the model improves on PB and GCS at charged surfaces, which is a meaningful test of the framework, but it's still within the same simulation paradigm.\n\nMinor: no error bars anywhere, and no code or data provided. The non-polarizable force-field caveat is stated honestly and limits the scope to weakly polarizable surfaces.\n\nVerdict: worth a serious referee. The core observation is robust enough to merit scrutiny, and the model, while not fully predictive, is a useful synthesis. I'd send it out, with a request for a direct test of the transferability assumption and an uncertainty analysis.","headline":"Solid MD study with a provocative but plausible claim that interfacial water polarization, not ion screening, dominates the potential at graphene electrodes; the extended PB model is illustrative rather than independently predictive.","tokens_in":9990,"tokens_out":2816,"would_cite":true,"duration_ms":25347,"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":"Water at graphene electrodes is polarized by the carbon surface itself, not by the ions, and this surface-induced polarization sets the electric potential distribution, the potential of zero charge, and the double-layer capacitance.","keywords":["electrochemical interfaces","water dynamics","electric double layer model","graphene electrode","potential of zero charge","molecular dynamics","overscreening","Poisson-Boltzmann equation"],"falsifier":"Recompute the O–H orientational distribution of the first water layer in the graphene–NaBF$_4$ simulation, subtract the response that scales linearly with the ion density, and compare the residual with the pure graphene–water distribution: if the residual deviates beyond statistical error at any distance within about 5 Å of the surface, the subtracted $\\phi^0_{\\mathrm{pzc,H_2O}}$ is not transferable and the overscreening decomposition fails. The published simulation framework already contains the trajectories needed to run this check.","tokens_in":8878,"feed_emoji":"💧","tokens_out":17270,"duration_ms":111516,"temperature":0.7,"pith_summary":"This paper aims to overturn the textbook picture of water as a passive, structureless dielectric in electric double layers. Using molecular dynamics of graphene in pure water and in 0.8 M NaCl and NaBF$_4$, it shows that interfacial water is strongly polarized by the graphene surface itself, not by the ions, and that this surface-induced polarization dominates the electric potential distribution, the potential of zero charge, and the double-layer capacitance. The evidence includes near-identical potential oscillations (up to about 1.15 V peak-to-peak) with and without electrolyte, and a decomposition in which the water contribution at the potential of zero charge reaches $+1.71$ V in Gr–NaBF$_4$, overwhelming the ions' $-1.45$ V and flipping the sign of the total potential. The paper closes with an extended Poisson–Boltzmann model that subtracts the surface water term from the potential the ions feel, recovering the simulated potentials within a few hundredths of a volt.","feed_headline":"Water polarization, not ions, sets graphene electrode voltage","feed_subtitle":"Molecular dynamics show surface-oriented water overrides ionic screening of the double layer.","key_machinery":"The load-bearing device is a linear decomposition of the electric potential: $\\phi_{\\mathrm{pzc,total}} = \\phi_{\\mathrm{pzc,ion}} + \\phi^0_{\\mathrm{pzc,H_2O}} + \\chi_{\\mathrm{H_2O}}$, where $\\phi^0_{\\mathrm{pzc,H_2O}}$ is the water-polarization potential measured in the pure graphene–water system, assigned to water oriented by the bare surface, and $\\chi_{\\mathrm{H_2O}}$ is the residual water response to the ions. The paper shows that $\\chi_{\\mathrm{H_2O}}$ is a near mirror image of $\\phi_{\\mathrm{pzc,ion}}$, with the ratio $\\phi_{\\mathrm{pzc,ion}}/(\\phi_{\\mathrm{pzc,ion}}+\\chi_{\\mathrm{H_2O}})$ close to the simulated dielectric constant of water — the signature of ordinary linear screening. The same decomposition, extended to charged surfaces, yields the modified Poisson–Boltzmann equations (Eqs. (2)–(3) and (5)–(6)): the ion distribution obeys the classic equation in the reduced potential $\\phi_{\\mathrm{ele,total}} - \\phi^0_{\\mathrm{pzc,H_2O}} - \\Delta\\phi^0_{\\mathrm{H_2O,II}}$, and the surface water terms are added back afterward. A potential of mean force from the MD simulation supplies the non-electrostatic ion–surface interactions, with the surface water electrostatics removed to avoid double counting.","core_discovery":"The central claim is that a significant portion of water polarization at a graphene/water interface responds to the graphene surface itself rather than screening the ions, and this surface-induced polarization governs the electric potential distribution. The paper demonstrates an overscreening effect in which the water contribution $\\phi_{\\mathrm{pzc,H_2O}}$ exceeds the ionic contribution $\\phi_{\\mathrm{pzc,ion}}$ in magnitude: in the graphene–NaBF$_4$ system the water term is $+1.71$ V against the ions' $-1.45$ V, so the total potential of zero charge is positive ($+0.26$ V) even though the interface is anion-rich. The mechanism is isolated by decomposing the water potential into a surface-induced part taken from the pure water system, $\\phi^0_{\\mathrm{pzc,H_2O}}$, and a residual part $\\chi_{\\mathrm{H_2O}}$ that mirrors the ionic potential and acts as ordinary linear dielectric screening. Because $\\chi_{\\mathrm{H_2O}}$ nearly cancels $\\phi_{\\mathrm{pzc,ion}}$ (the ratio $\\phi_{\\mathrm{pzc,ion}}/(\\phi_{\\mathrm{pzc,ion}}+\\chi_{\\mathrm{H_2O}})$ is close to the simulated dielectric constant of water), the total potential reduces to $\\phi_{\\mathrm{pzc,total}} \\approx \\phi^0_{\\mathrm{pzc,H_2O}}$; at charged surfaces the interfacial water reorientation term $\\Delta\\phi^0_{\\mathrm{H_2O,II}}$ accounts for roughly 94–97% of the potential change. The extended Poisson–Boltzmann equations, which subtract $\\phi^0_{\\mathrm{pzc,H_2O}}$ (and $\\Delta\\phi^0_{\\mathrm{H_2O,II}}$) from the potential that the ions respond to, reproduce the MD surface potentials at Gr–NaCl within a few hundredths of a volt.","pith_inferences":["Testable extension: varying the surface (different hydrophobicity, lattice spacing, or surface polarizability) while keeping the electrolyte fixed should move the surface water term $\\phi^0_{\\mathrm{pzc,H_2O}}$, so the model predicts PZC values track the strength of surface-induced water structuring rather than ion identity.","The decomposition implies the sharper claim that on weakly polarizable, atomically flat electrodes the potential of zero charge is essentially a solvent-structure property; this would help explain why experimental PZC values at such surfaces cluster similarly across very different electrolytes.","Surface-specific vibrational spectroscopy comparing the O–H orientation distribution at the graphene interface in pure water and in 0.8 M NaBF$_4$ at the PZC would directly test the transferability of $\\phi^0_{\\mathrm{pzc,H_2O}}$.","The authors themselves restrict the claim to weakly polarizable surfaces and note that non-polarizable force fields omit induced electronic polarization; running the same decomposition with polarizable force fields or on metallic electrodes is the natural stress test of where the effect ends."],"forward_implications":["Classic Poisson–Boltzmann and Gouy–Chapman–Stern models omit the surface-induced water term, so at graphene/aqueous interfaces they under-predict the potential of zero charge and surface potentials by about an order of magnitude.","The potential of zero charge at uncharged graphene can be positive even when the interface is dominated by adsorbed anions, because the water orientation term, not the ion distribution, sets the sign.","Subtracting the surface water polarization from the electrolyte potential leaves a reduced potential in which the ions follow the classic Poisson–Boltzmann equation, which is why the classic models still work semi-quantitatively in many electrochemical systems.","Because the interfacial water reorientation supplies roughly 94–97% of the potential change at charged graphene surfaces, the double-layer capacitance of graphene electrodes is predominantly a property of the water's orientational response, not of ion packing.","The extended model reproduces the MD surface potentials at Gr–NaCl — 0.34 V versus 0.33 V at the PZC, and 0.99 V versus 0.98 V at the positively charged surface — matching simulation within a few hundredths of a volt."],"supporting_citations":[{"why":"It supplies the experimental suspended-graphene–water measurements against which the simulated PZC values are checked.","marker":"[13]"},{"why":"It provides the classic dielectric-overscreening expectation — water's potential opposite in sign and smaller than the ions' — that the measured overscreening overturns.","marker":"[22]"},{"why":"It is the ab initio molecular dynamics reference for the water density and O–H orientation profiles near graphene that the force-field simulations reproduce.","marker":"[17]"},{"why":"It provides experimental electric-double-layer data on suspended graphene that motivate the water-dominated PZC interpretation.","marker":"[12]"},{"why":"It is the source of the approximation that the ion distribution is insensitive to the static water field, used to drop the surface water electrostatics from the ion Boltzmann factor.","marker":"[23]"},{"why":"It supplies the prior simulation methodology for graphene-electrode double-layer capacitance that the present force-field setup extends.","marker":"[20]"},{"why":"It documents the experimental asymmetric response of interfacial water to applied electric fields that the charged-surface decomposition reflects.","marker":"[14]"}],"fun_headline_variants":["Water overrides ions at graphene electrode","Water, not ions, sets graphene voltage","Interfacial water beats ions in graphene double layer","Surface water flips graphene electrode polarity","Graphene voltage controlled by water, not ions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the water polarization induced by the bare graphene surface is exactly the same when 0.8 M salt is present, so the pure-water profile $\\phi^0_{\\mathrm{pzc,H_2O}}$ can be cleanly subtracted from the electrolyte potential; the paper infers this from matching total potential oscillations but does not test the interfacial water orientation distribution directly in the salt solution.","fun_headline_variants_meta":{"raw":{"variants":["Water overrides ions at graphene electrode","Water, not ions, sets graphene voltage","Interfacial water beats ions in graphene double layer","Surface water flips graphene electrode polarity","Graphene voltage controlled by water, not ions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000352,"raw_usage":{"total_tokens":1985,"prompt_tokens":1077,"completion_tokens":908,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":693,"completion_tokens_details":{"reasoning_tokens":840}},"tokens_in":693,"tokens_out":908,"duration_ms":867865,"temperature":1.0,"reasoning_tokens":840,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:56:20.948173+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the O–H orientational distribution of the first water layer in the graphene–NaBF$_4$ simulation, subtract the response that scales linearly with the ion density, and compare the residual with the pure graphene–water distribution: if the residual deviates beyond statistical error at any distance within about 5 Å of the surface, the subtracted $\\phi^0_{\\mathrm{pzc,H_2O}}$ is not transferable and the overscreening decomposition fails. The published simulation framework already contains the trajectories needed to run this check.","supporting_citations":[],"review_version":1}