{"id":"cbf9b35e-c2a1-4d5e-9ecb-5f47e894cd5b","arxiv_id":"1908.07852","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"Mechanically exfoliated graphene and hBN membranes show perfect Nernst selectivity: protons carry all the ionic current and chloride ions are blocked.","lead":"Defect-free graphene and hexagonal boron nitride membranes pass only protons, blocking chloride ions completely. The result helps settle a dispute over whether protons really pierce pristine 2D crystals.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Defect-free status of the exact membranes used in ion transport is inferred only from sibling nanoballoon tests, and the SI describes leak tests for graphene only, so the 'perfect' proton-selectivity claim overreaches the defect characterization.","rationale":"The paper is careful and well controlled: bare-aperture controls, porous glass reference measurements, the multi-aperture graphene device, and agreement with bulk HCl transport numbers all support the methodology. The loophole is not the electrical measurement itself but the attribution of the observed selectivity to pristine, defect-free crystals. The reader's weakest assumption identifies this precisely, and I agree. The SI detail that the nanoballoon protocol is described for monolayer graphene only makes the hBN-based quantitative claim somewhat more exposed. This concern does not invalidate the main scientific message that exfoliated 2D crystals are radically more proton-selective than CVD graphene, but it means the words 'perfect' and 'defect-free' are conditional on an unverified premise about the exact membranes. The reader's CONDITIONAL verdict is appropriate, so no change is needed.","tokens_in":8711,"tokens_out":10451,"duration_ms":122776,"concrete_test":"Perform a sensitivity calibration: measure V0 on the same ion-transport device before and after deliberately introducing a small, known density of atomic-scale defects into that exact membrane (e.g., mild UV etching as in Supp. Fig. 4), and also run a gas-leak nanoballoon test on that identical membrane if geometry permits. If a few defects shift V0 by more than the ~2 mV experimental uncertainty, then the observed -58 mV on the unetched devices is meaningful evidence that those membranes were defect-free. If V0 remains at -58 mV after etching, the reversal-potential measurement is insensitive to small defects and the perfect-selectivity claim remains unsupported by the defect characterization.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that defect-free exfoliated graphene and hBN permit only proton transport. The load-bearing assumption is that the specific crystals mounted in the HCl transport cells had no atomic-scale defects. The nanoballoon gas-leak test, the only assay claimed to detect a single vacancy, is described in the SI as performed on monolayer graphene membranes made for that purpose; the devices whose reversal potentials are reported, especially monolayer hBN, which provides the quantitative tH = 0.99 result, were not individually leak-tested. The transport devices also underwent additional SU-8 washer transfer and 150 °C baking after the crystals were exfoliated, which could in principle introduce damage not captured by sibling tests. With tH = 0.99 ± 0.02, chloride transport numbers as large as 0.02 are not excluded, so 'perfect' and 'no detectable flow of counterions' are stronger than the membrane-potential precision justifies. If an actual device contained a small defect or crack, the reversal potential could still be consistent with the data, and the conclusion that intrinsic, pristine 2D crystals block chloride would not follow.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports ion-transport measurements through mechanically exfoliated monolayer graphene and hexagonal boron nitride (hBN) membranes separating hydrochloric acid solutions of different concentrations. The authors measure areal conductivities at various HCl concentrations, and for concentration ratios ΔC from 1 to 30 they record the zero-current (reversal) potential V0. For ΔC = 10 they find V0 ≈ −58 mV independent of the absolute concentrations, which matches the Nernst prediction for a membrane with proton transport number tH = 1. A fit to the data gives tH = 0.99 ± 0.02 for hBN, and graphene yields −55 ± 9 mV. Porous-glass control experiments give tH = 0.81 ± 0.04, in agreement with bulk HCl transport numbers. The paper concludes that exfoliated, defect-free 2D crystals exhibit 'perfect' Nernst selectivity, with protons accounting for all ionic current and chloride ions blocked, corroborating earlier claims of intrinsic proton permeation through such crystals.","tokens_in":8963,"tokens_out":6674,"duration_ms":62231,"significance":"If the strong claim holds, this is an important result: it would establish that mechanically exfoliated monolayer graphene and hBN can act as nearly ideal proton-selective membranes, with direct implications for theories of proton permeation through 2D crystals and for separation technologies. The experimental approach is clean: the Nernst prediction for tH = 1 is parameter-free, the porous glass control independently calibrates the setup, bare-aperture devices show that series resistance is negligible, and leakage currents are reported. The comparison with CVD graphene results is also valuable. However, the 'perfect' selectivity claim is stronger than the precision of the transport-number fit, and the inference that the specific measured crystals were defect-free rests on indirect evidence, so the significance as stated is not yet fully established.","major_comments":[{"comment":"The claim that the transport membranes are defect-free relies on nanoballoon gas-leak tests that were performed on sibling devices, not on the exact crystals used for the ion-transport measurements. The Supplementary leak-test section describes sealing microcavities with monolayer graphene only, yet the main text states that hBN and graphene membranes were tested; the monolayer hBN devices that yield the quantitative tH = 0.99 result are not individually leak-tested. In addition, the transport devices underwent SU-8 washer transfer and a 150 °C bake after exfoliation, a process not included in the nanoballoon protocol and one that could introduce damage. Because a small chloride-conducting defect would still allow a reversal potential close to −58 mV, the conclusion that pristine, defect-free crystals block chloride ions is not established for the specific measured membranes.","section":"Supplementary 'Leak tests using nanoballoons'; Device fabrication and electrical measurements"},{"comment":"The wording 'perfect Nernst selectivity' and 'no detectable flow of counterions' exceeds the measurement precision. The best fit tH = 0.99 ± 0.02 gives tCl ≤ 0.02 within uncertainty, and the experiment does not directly detect chloride flux; it only determines the reversal potential. The claim should be softened to 'near-perfect selectivity' or 'selectivity consistent with tH = 1 within experimental accuracy' in the title, abstract, and conclusion.","section":"Fig. 2b and Eq. (1)"},{"comment":"The Nernst potential is evaluated using concentration ratios ΔC rather than ion activities. For HCl concentrations up to 3 M, the mean ionic activity coefficient varies substantially with concentration, so the relation V0 = −(kBT/e)ln(ΔC) is an approximation. Please state whether activity corrections were applied, and if not, discuss how the extracted transport number and the presented comparison V0 ≈ −58 mV for 0.3 M | 3 M are affected.","section":"Fig. 2b, Nernst equation"}],"minor_comments":[{"comment":"The abstract says defect-free monolayers were 'previously shown' to be proton-permeable, while the main text describes the membranes as having 'few if any atomic-scale defects'; please make the defect-characterization language consistent.","section":"Abstract and introduction"},{"comment":"The Supplementary figure title reads 'Supplementary Figure 4I Leak tests using nanoballoons'; the 'I' appears to be a typo for the vertical bar separator.","section":"Supplementary Figure 4"},{"comment":"Please report the number of independent devices and measurements underlying each data point in Fig. 2b, and indicate whether the error bars represent device-to-device or measurement-to-measurement scatter.","section":"Fig. 2b"},{"comment":"The subtraction of the Ag/AgCl redox potential is described qualitatively; please provide the numerical value used for Vredox and the measured Vcell values for a representative device.","section":"Supplementary 'Membrane potential measurements'"}],"recommendation":"major_revision","confidential_remarks":"The data are clean and the control experiments are thoughtful. My main concern is the gap between the 'perfect selectivity from pristine crystals' claim and the actual evidence: the specific transport membranes, especially the hBN ones that provide the quantitative tH = 0.99 result, were not individually leak-tested after the SU-8/bake fabrication steps. I would encourage the editor to ask the authors to either provide leak-test evidence on the actual transport devices or on hBN devices processed through the same full procedure, or to moderate the claims accordingly. The paper would be publishable after such a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The first thing to know: this is the first ion-selectivity measurement on mechanically exfoliated 2D crystals, and it lands where it should. The reversal potential for hBN at a tenfold concentration gradient is -58 mV, within error of the Nernst value for proton-only transport. That directly challenges the defect-mediated picture drawn from CVD graphene, and the contrast in conductance (three orders of magnitude lower) and membrane potential (7x larger) is what actually makes the case for intrinsic proton transport. This is a significant within-subfield result and will be cited.\n\nThe controls here are the real strength. Bare-aperture devices conduct at least 1000x more than the membranes, leakage currents are at the 1 pA level, and the porous glass reference reproduces the bulk transport numbers (tH = 0.81 ± 0.04 against the known 0.83). The graphene devices, with their larger uncertainty, still give V0 = -55 ± 9 mV. The paper is also honest about the detection limit and the complications in the graphene measurements.\n\nSoft spots are there, but they don't flip the conclusion. The word \"perfect\" overreaches: a fitted tH of 0.99 ± 0.02 allows chloride transport numbers up to 0.02, so \"no detectable flow of counterions\" is stronger than the precision warrants. That is a wording issue, not a methodological one. A more substantive caveat is that the defect-free status of the exact hBN membranes used in the selectivity measurements is inferred from nanoballoon gas-leak tests done on separately fabricated devices, and the SI details the leak tests only for graphene. The transport devices also went through SU-8 washer transfer and a 150 °C bake, which could in principle introduce damage. So the paper relies on a sibling-device argument for its cleanest quantitative result. Still, this is a limitation, not a fatal flaw: dozens of tested crystals were leak-free, the hBN and graphene behavior are consistent, and if an individual device did have a small defect, the reversal potential would still be dominated by protons, so the intrinsic-vs-defect controversy is resolved even if \"perfect\" is not literally proven for every membrane.\n\nI'd send this to referees. The request to the authors is straightforward: either temper the \"perfect\" language to what the uncertainty bars allow, or provide leak-test evidence for the specific hBN devices. The paper is otherwise carefully done, with clean controls and a clear narrative. A serious referee should engage with it, and the likely outcome is publication after minor revision.","headline":"Solid experimental resolution of a real controversy, with a well-targeted caveat: the 'perfect' selectivity claim slightly outruns the defect characterization, but the core result stands.","tokens_in":9493,"tokens_out":2118,"would_cite":true,"duration_ms":23009,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Defect-free monolayers of graphene and hexagonal boron nitride let protons through while blocking chloride ions completely, according to membrane-potential measurements that fit the ideal Nernst value.","keywords":["proton transport","Nernst selectivity","graphene","hexagonal boron nitride","ion exclusion","two-dimensional membranes","nanoballoon leak test","transport number"],"falsifier":"Deliberately etch a single known vacancy into one of these exfoliated membranes using the same ultraviolet treatment used in the nanoballoon controls and re-measure the reversal potential: the fitted $t_H$ should drop measurably below 1 and $V_0$ should move toward zero if defects are what control ion transport. Alternatively, place an isotopically labelled chloride tracer ($^{36}\\mathrm{Cl}^-$) in the high-concentration compartment and look for its appearance on the low-concentration side; any detected chloride through a pristine membrane would directly falsify perfect proton selectivity.","tokens_in":8547,"feed_emoji":"🧪","tokens_out":9453,"duration_ms":88827,"temperature":0.7,"pith_summary":"This paper claims that defect-free monolayers of graphene and hexagonal boron nitride, placed between two hydrochloric acid solutions, let protons cross but block chloride ions completely. The central measurement is the voltage at which the net ionic current vanishes under a tenfold concentration gradient; it comes out at $V_0 \\approx -58$ mV at $20^\\circ$C, exactly the ideal Nernst value for a membrane in which protons carry all the current. Fitting reversal potentials for concentration ratios from 1 to 30 gives a proton transport number $t_H = 0.99 \\pm 0.02$, meaning no detectable counter-ion flow. If this is right, proton permeation through pristine 2D crystals is an intrinsic property of the lattice, and these crystals behave as perfect proton-selective membranes.","feed_headline":"Protons only: pristine 2D crystals reject every other ion","feed_subtitle":"Reversal voltages match the ideal Nernst value, so chloride carries none of the current through defect-free graphene and hBN.","key_machinery":"The load-bearing object is the Nernst membrane-potential relation connecting the reversal voltage $V_0$ to the two ion transport numbers $t_H$ and $t_{Cl}$: $V_0 = (t_{Cl} - t_H)(k_BT/e)\\ln(\\Delta C)$. Measuring $V_0$ for known concentration ratios directly gives the fraction of current carried by protons, because any chloride contribution would shift $V_0$ toward zero. The experiments combine this electrochemical probe with nanoballoon gas-leak tests, which can detect even a single angstrom-sized vacancy by watching whether a sealed gas-filled cavity deflates; only membranes that showed no leakage in those tests were used for the ion-transport measurements.","core_discovery":"The central discovery is that mechanically exfoliated, essentially defect-free graphene and hBN membranes exhibit perfect Nernst selectivity for protons. When the two sides of the membrane are filled with HCl at different concentrations, the zero-current voltage follows $V_0 = -(2t_H - 1)(k_BT/e)\\ln(\\Delta C)$ with $t_H \\approx 1$; for $\\Delta C = 10$ the measured $V_0$ is $-58$ mV at $20^\\circ$C, and over $\\Delta C = 1$ to 30 the best fit is $t_H = 0.99 \\pm 0.02$. Protons therefore account for all ionic current through the membrane while chloride is blocked. The same behavior appears in graphene, with $V_0 = -55 \\pm 9$ mV. Because the crystals passed nanoballoon gas-leak tests sensitive to single vacancies, the authors conclude the selectivity is a property of the pristine lattice, not of defects, and that the much larger currents and weak selectivity reported for CVD graphene reflect defect-dominated transport.","pith_inferences":["If the pristine lattice truly blocks every ion except protons, then replacing HCl with other acids or salts should change the picture sharply: sulfuric and nitric acids should give the same ideal reversal potential, while salts without a proton gradient should produce no current.","The fact that chloride, with its small hydrated diameter, is rejected suggests the selectivity works on bare protons rather than hydrated-ion size, so the same membranes should also block larger alkali and halide ions in electrochemical applications.","A practical extension the authors leave implicit: a large-area, defect-free hBN or graphene layer could serve as a proton-selective barrier that prevents anion crossover in fuel cells or electrolysers, if the required crystal quality can be scaled.","Reversal-potential measurements could become a quantitative in-situ probe for atomic-scale damage in 2D membranes, sensitive enough to complement nanoballoon gas tests."],"forward_implications":["A pristine monolayer of hBN or graphene can act as a proton-only membrane, excluding every other ion, which is the defining behavior of an ideal proton-exchange membrane.","The result corroborates the earlier claim that thermal protons pierce defect-free 2D crystals and removes the vacancy-based alternative explanation for those experiments.","The conductance of exfoliated crystals is nearly three orders of magnitude below that reported for CVD graphene, so CVD membranes are not representative of intrinsic 2D-crystal ion transport.","In nanoporous 2D separation membranes, the proton permeability of the bulk crystal itself contributes to ion current and must be included when designing for selectivity.","The ideal Nernst reversal potential doubles as a quality assay: a membrane that gives $V_0 = -58$ mV at a tenfold gradient is behaving as a defect-free proton conductor."],"supporting_citations":[{"why":"Showed thermal protons permeate mechanically exfoliated graphene and hBN with activation barriers, the claim this work corroborates.","marker":"[1]"},{"why":"Demonstrated quantum isotope effects in proton transport through 2D crystals, supporting intrinsic rather than defect-mediated permeation.","marker":"[2]"},{"why":"Attributed proton currents through single-layer graphene to atomic-scale defects, the competing view this work challenges.","marker":"[10]"},{"why":"Reported weak proton selectivity for CVD graphene; its contrasting current densities and membrane potentials anchor the comparison.","marker":"[11]"},{"why":"Introduced suspended 2D atomic membranes and established their impermeability to gases, forming the device platform.","marker":"[15]"},{"why":"Established nanoballoon gas-leak detection for atomic-scale pores, used here to certify defect-free crystals.","marker":"[16]"},{"why":"Showed a single angstrom-sized vacancy makes a sealed cavity deflate in seconds, defining the sensitivity of the leak tests.","marker":"[17]"},{"why":"Supplies the Nernst equation connecting reversal potential to transport numbers, the theoretical basis for extracting $t_H$.","marker":"[21]"}],"fun_headline_variants":["Perfect proton sieving: 2D crystals block all other ions","Only protons get through pristine graphene and hBN","Perfect Nernst selectivity: only protons cross 2D crystals","Ideal proton filter: 2D crystals pass only H+, block Cl-","Defect-free 2D membranes: perfect proton-only transport"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion rests on the assumption that the exact crystals used in the ion-transport measurements were as defect-free as the separately made nanoballoon devices; if any measured membrane contained an unseen vacancy or crack, chloride could pass through it and the perfect selectivity would not reflect the pristine lattice.","fun_headline_variants_meta":{"raw":{"variants":["Perfect proton sieving: 2D crystals block all other ions","Only protons get through pristine graphene and hBN","Perfect Nernst selectivity: only protons cross 2D crystals","Ideal proton filter: 2D crystals pass only H+, block Cl-","Defect-free 2D membranes: perfect proton-only transport"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000458,"raw_usage":{"total_tokens":2284,"prompt_tokens":921,"completion_tokens":1363,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":537,"completion_tokens_details":{"reasoning_tokens":1274}},"tokens_in":537,"tokens_out":1363,"duration_ms":19539,"temperature":1.0,"reasoning_tokens":1274,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:54:32.205715+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deliberately etch a single known vacancy into one of these exfoliated membranes using the same ultraviolet treatment used in the nanoballoon controls and re-measure the reversal potential: the fitted $t_H$ should drop measurably below 1 and $V_0$ should move toward zero if defects are what control ion transport. Alternatively, place an isotopically labelled chloride tracer ($^{36}\\mathrm{Cl}^-$) in the high-concentration compartment and look for its appearance on the low-concentration side; any detected chloride through a pristine membrane would directly falsify perfect proton selectivity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Attributed proton currents through single-layer graphene to atomic-scale defects, the competing view this work challenges."},{"cited_title":"I., Braeuninger-Weimer, P., Weatherup, R","cited_arxiv_id":null,"evidence_quote":"Reported weak proton selectivity for CVD graphene; its contrasting current densities and membrane potentials anchor the comparison."},{"cited_title":"P., Wang, L., Pellegrino, J","cited_arxiv_id":null,"evidence_quote":"Established nanoballoon gas-leak detection for atomic-scale pores, used here to certify defect-free crystals."},{"cited_title":"Ion Exchange","cited_arxiv_id":null,"evidence_quote":"Supplies the Nernst equation connecting reversal potential to transport numbers, the theoretical basis for extracting $t_H$."}],"review_version":1}