{"id":"5d51ce57-0804-45b4-8015-2b3d6e9ac3cd","arxiv_id":"2502.09347","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":11,"one_line_summary":"Submolecular photocurrent maps of a single PTCDA anion reveal and separately image the two electronic configurations that make up its excited state.","lead":"This paper uses a scanning tunneling microscope with laser light to image tiny light-driven electrical currents on a single charged dye molecule at submolecular resolution. The images reveal two electronic configurations inside the molecule's excited state, and the authors show they can switch between them by changing the voltage.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'proof' that both configurations A and B appear in D1^- is undercut by the model's admitted non-uniqueness; the maps may be fit with different configuration weights, so the experimental data support but do not prove the two-configuration assignment.","rationale":"The paper's most valuable contribution is the demonstration that light-STM can produce submolecular photocurrent maps with a rich, bias-tunable contrast on a single open-shell molecule. The theoretical model is transparent and the simulations plausibly reproduce the maps. However, the headline claim uses 'prove' twice (abstract and concluding paragraph) to assert that the maps directly visualize two configurations of D1^-. The evidence is an indirect fit: experimental maps are compared with a simulation that has at least 11 parameters, several of which are tuned to match the same data. The authors themselves admit in the SI that different parametrizations, including different configuration weights, could produce similar results. That admission is a direct threat to the proof-level inference. Even if the Kohn-Sham approximation to the Dyson orbitals were perfect, the model's non-uniqueness would prevent a proof. The proposed refit test targets exactly this weakness: by trying to fit the maps with only one configuration, one can determine whether the second configuration is actually required by the data. This is a computationally feasible test using the existing model infrastructure. If the single-configuration fits fail, the 'prove' language becomes defensible; if they succeed, the language must be softened to 'consistent with' as the reader's conditional verdict already requests. The reader's weakest_assumption about Dyson orbitals is real and related, but the parameter degeneracy is the more load-bearing issue because it survives even perfect orbital fidelity.","tokens_in":24205,"tokens_out":9918,"duration_ms":94318,"concrete_test":"Refit the rate-equation model under the constraint that configuration A is absent: set the D1^- -> S0 transition rate involving orbital 4 to zero and set the configuration-A weight in D1^- to zero, then re-optimize the remaining free parameters (including gamma0_T, alpha, kappa_pl, eta0, and the state energies within their experimental uncertainties) to minimize a quantitative discrepancy metric (e.g., normalized cross-correlation or mean-square error) against the zero-bias and bias-dependent dI/dP maps in Figs. 2e and 3. Repeat with configuration B absent. If either single-configuration model achieves a fit statistically comparable to the published two-configuration fit, the maps do not prove the coexistence of A and B; if neither can reproduce the bidirectional maps, the proof is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Central claim: the positive and negative photocurrent features 'prove' the presence of both configurations A and B in D1^- (main text, final paragraph of the photocurrent section). This inference relies on a forward rate-equation model in which Dyson orbitals are approximated by single Kohn-Sham orbitals evaluated at z_T = 9.4 a.u., each normalized to its own maximum (SI Eq. 4), and in which the relative weights of the two configurations are inputs taken from TDDFT. The model has many free parameters (gamma0_T, gamma0_S, alpha, kappa0, kappa_pl, eta0, gamma_IC, E_R) fitted to the same experimental maps. The SI explicitly states that 'a different parametrization of the model could yield similar results even if different contributions of the two configurations were considered.' Therefore the maps do not uniquely determine the configuration weights. In particular, the negative current is also generated by orbital 3 of D2^- (main text), so the negative-lobe pattern could in principle be reproduced without any orbital-4 contribution from configuration A. No quantitative fit metric or uniqueness analysis is offered; the Fig. 2e comparison is qualitative. The data are consistent with the two-configuration model, but the 'prove' language exceeds what a non-unique forward model can establish. The Dyson-orbital approximation compounds this, but the parameter degeneracy alone is sufficient to undermine the proof-level claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports light-STM measurements of zero-bias and bias-dependent photon-induced currents on individual PTCDA anions on NaCl/Ag(111), combined with TEPL and dI/dV spectroscopy. The central claim is that submolecular photocurrent maps contain spatial fingerprints of two dominant electronic configurations, denoted A and B, of the multiconfigurational doublet excited state D1^-, and that applying a bias voltage switches the dominant recombination pathway between the two configurations. The interpretation is developed through a rate-equation model in which tip- and substrate-mediated tunneling rates are expressed in terms of Kohn-Sham orbital slices used as approximate Dyson orbitals, with many-body level energies extracted from the same experimental TEPL and dI/dV data. The authors conclude that the correspondence between experimental and simulated maps 'proves' the presence of both configurations in D1^-.","tokens_in":24561,"tokens_out":3860,"duration_ms":40209,"significance":"If the central claim were established, the work would be a substantial advance: it would provide direct real-space visualization of the individual electronic configurations of a multiconfigurational excited state in a single molecule, together with a way to select one recombination pathway over another by bias voltage. The experimental data are novel, the combination of submolecular photocurrent mapping with a rate-equation model is ambitious, and the paper explicitly addresses a difficult open problem in single-molecule spectroscopy. However, the proof-level claim is not supported by the present analysis: the forward model is admitted to be non-unique, the spatial fingerprints rest entirely on an unvalidated Dyson-orbital approximation, and several key parameters are fitted to the same data that the model then reproduces. With a quantitative robustness analysis and appropriately qualified conclusions, the paper could become a valuable contribution; in its current form, the evidentiary gap between 'consistent with' and 'proves' is too wide.","major_comments":[{"comment":"The statement that the signatures of orbital 1 in the positive current and orbital 4 in the negative current 'prove the presence of both configurations A and B in the state D1^-' is load-bearing and is not justified by the analysis. The SI explicitly concedes that 'a different parametrization of the model could yield similar results even if different contributions of the two configurations were considered,' and no quantitative fit metric or uniqueness study is provided for the comparison in Fig. 2e. Moreover, the negative current is also generated by orbital 3 of D2^-, so the negative-lobe pattern does not by itself single out orbital 4. The data are consistent with the two-configuration model, but they do not uniquely determine the configuration weights; the word 'prove' should be replaced by a weaker claim, or supplemented by a sensitivity analysis over the fitted parameters and configuration weights.","section":"Main text, final paragraph of the photocurrent section; SI, 'A different parametrization ...'"},{"comment":"The spatial fingerprints used to identify configurations A and B rest entirely on approximating each Dyson orbital by a single Kohn-Sham orbital evaluated on a constant-height plane and normalized to its own maximum. The manuscript provides no validation of this approximation and no test of robustness to the choice of plane height, the DFT functional, the fractional molecular charge (-0.5e) used in the Octopus calculation, or the expected difference between true Dyson orbitals and single Kohn-Sham orbitals. Since the identification of the negative-current feature with orbital 4 and configuration A depends on this approximation, a robustness check is required before the assignment can be regarded as established.","section":"SI, 'Finally, to obtain the spatial dependency ...', Eq. (4), z_T = 9.4 a.u."},{"comment":"The model re-uses the same experimental data from which its input energies and rates are derived, so the agreement in Fig. 2e and the bias-dependent maps is not an independent validation. The many-body energies in Table S1 are extracted from the TEPL and dI/dV spectra that the model then simulates, and eta0 is explicitly selected to match the ratio of photocurrent maxima to sequential-tunneling-current maxima. This does not invalidate the model as an interpretation tool, but it means that the correspondence between experiment and simulation cannot, by itself, carry the burden of proving the configuration assignment. The authors should clearly separate fitted parameters from predicted quantities and, where possible, test whether a single parameter set held fixed across all maps and bias voltages is sufficient.","section":"Table S1 and SI Eqs. (3)-(14); Fig. 2d,e"}],"minor_comments":[{"comment":"The caption uses the symbol omega2 for both the bias modulation frequency and the laser modulation frequency; one of the two should be labeled omega1 to avoid ambiguity.","section":"Fig. S5 caption"},{"comment":"The many-body energies are quoted without uncertainties; since they are derived from experimental spectra, an estimate of the error bars would help assess how well-determined the level alignment is.","section":"Table S1"},{"comment":"Normalizing the DOS by its maximum removes the absolute orbital amplitude from the tunneling rate; this choice should be stated explicitly and justified, because it affects the relative weights of the different tunneling channels.","section":"SI Eq. (4)"},{"comment":"Some references contain incomplete bibliographic data, for example reference [15] lacks a final page/article number and several Nature-family entries list only volume and starting page; these should be completed for publication.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The experimental effort and the combination of techniques are impressive, and the paper addresses a timely question in single-molecule spectroscopy. My main concern is the gap between the 'prove' claim and the admitted non-uniqueness of the forward model; in my view, the paper needs a robustness analysis and a softening of the central claim rather than a rejection of the underlying work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is applying light-STM photocurrent mapping to an open-shell anion with a multiconfigurational doublet state, and showing that bias voltage switches which configuration dominates the contrast. That is a useful capability for organic radical emitters and donor-acceptor interfaces. The experimental work is solid: zero-bias bidirectional photocurrent maps, TEPL with D1^- and D2^- emission, dI/dV with and without illumination, and a rate-equation model that is transparent enough to test. The SI deserves credit for listing all the free parameters and explicitly stating that a different parametrization could yield similar results even if the configuration weights changed. That honesty is why the paper is salvageable.\n\nThe soft spot is the central claim. The maps do not prove both configurations A and B are present in D1^-. The model has many fitted parameters, and several of them (the many-body state energies, the pumping rate eta0, the voltage-drop alpha) are extracted from the same experimental spectra the model then reproduces. The SI admits non-uniqueness. More specifically, the negative current is also generated by orbital 3 of D2^-, so the negative-lobe pattern could in principle be reproduced without any orbital-4 contribution from configuration A. The Dyson orbitals are approximated by single Kohn-Sham orbitals evaluated on a plane about 5 Å above the molecule, each normalized to its own maximum; if the true Dyson orbitals differ, the spatial fingerprint assignment would collapse. No quantitative fit metric or uniqueness analysis is offered; the Fig. 2e comparison is qualitative.\n\nThat said, the bias-switching observation is likely robust, because it tracks changes in the maps with bias rather than a single static assignment. The paper would be more convincing if the authors softened 'prove' to 'support,' added a sensitivity analysis of the fits, and either constrained eta0 and alpha independently or showed how the maps change with configuration weights.\n\nWho is this for? People working on single-molecule photophysics, light-STM, and radical emitters. It deserves a serious referee: the experiment is nontrivial and the model is openly documented. I would accept it after revisions, not desk-reject. For peer review, send to someone who knows STM optics and can judge the rate-equation model; the proof claim should be flagged for revision.","headline":"Bias-switchable photocurrent fingerprints on a single open-shell molecule are a real advance, but the 'prove' language overreaches a non-unique forward model.","tokens_in":25163,"tokens_out":2786,"would_cite":true,"duration_ms":28104,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["68.37.Ef","73.40.Gk"],"model":"deepseek-v4-flash","headline":"Submolecular photocurrent maps of a single PTCDA anion are claimed to directly image the two dominant electronic configurations of its multiconfigurational first excited doublet state, with bias choosing which configuration recombines.","keywords":["light-STM","photocurrent mapping","multiconfigurational excited state","PTCDA anion","Dyson orbitals","doublet radical","plasmonic nanocavity","rate-equation model"],"falsifier":"Compute true Dyson orbitals for the $D_1^- \\to S_0$ and $D_1^- \\to S_0^{2-}$ transitions with a correlated many-body method, including the NaCl substrate in the model, and compare their constant-height slices at $z_T = 9.4$ a.u. with the Kohn-Sham orbital 1 and orbital 3/4 slices used here; if the dominant Dyson densities differ appreciably from those slices, the claimed identification of the positive-current lobes with configuration B and the negative-current lobes with configuration A is falsified. A purely experimental second check: re-record the zero-bias photocurrent map at a substantially different tip height, since the orbital fingerprints should persist if they are genuine tunneling channels rather than features of the single chosen plane.","tokens_in":23949,"feed_emoji":"🔬","tokens_out":13769,"duration_ms":126377,"temperature":0.7,"pith_summary":"This paper claims that zero-bias photocurrent maps of a single PTCDA anion molecule, recorded with light-scanning tunneling microscopy in a plasmonic nanocavity, are direct real-space fingerprints of the two dominant electronic configurations of its multiconfigurational first excited doublet state, $D_1^-$. The positive part of the bidirectional current is traced to electron injection into orbital 1 (configuration B), while the negative part is traced mostly to electron extraction from orbital 4 (configuration A); the two orbital signatures appearing in the same map are taken as proof that $D_1^-$ is a superposition of both configurations. A rate-equation model fed by TD-DFT transition densities and Kohn-Sham orbital slices reproduces the contrast and predicts that the bias voltage smoothly selects which recombination pathway dominates, which the experiments confirm by converting the map from an orbital-1 pattern to an orbital-3/4 pattern as the bias crosses zero. If correct, this gives experimental access, at the single-molecule level, to the correlated composition of excited states that previously could only be inferred from ensemble-averaged spectra.","feed_headline":"Photocurrent maps expose a molecule's dual excited-state character","feed_subtitle":"Zero-bias light-STM images on a single PTCDA anion reveal both configurations of its excited doublet state.","key_machinery":"The central object is a set of Dyson-like orbital maps: for every tunneling transition between many-body states, the tip-mediated rate is taken proportional to the squared wave function of the Kohn-Sham orbital that approximates the transition's Dyson orbital, evaluated on a constant-height plane about 5 Å above the molecule ($z_T = 9.4$ a.u.). Each of the five many-body states ($S_0$, $D_0^-$, $D_1^-$, $D_2^-$, $S_0^{2-}$) is assigned a definite frontier-orbital occupation, so a charge transfer between states is tied to one specific orbital. Those two-dimensional densities are multiplied by a bias-dependent threshold function derived from a Franck-Condon model of the NaCl substrate reorganization ($E_R = 850$ meV), and the steady-state solution of the resulting rate equations gives the position- and bias-dependent net current that is compared, pixel by pixel, with the measured $\\mathrm{d}I/\\mathrm{d}P$ maps.","core_discovery":"In a single PTCDA anion on a NaCl/Ag(111) surface inside a plasmonic nanocavity, illumination at 785 nm produces a tunneling current at zero bias whose direction flips between positive and negative over distances of a few bond lengths. The paper explains this bidirectional photocurrent with a five-state model: absorption populates the $D_1^-$ and, through internal conversion, the $D_2^-$ states, and the excited molecule decays back to $D_0^-$ by two competing sequences of single-electron transfers, one passing through the neutral $S_0$ and the other through the doubly negative $S_0^{2-}$. Because $D_1^-$ is a superposition of two configurations (A, with an electron in orbital 4, and B, with a hole in orbital 1), each decay path carries a distinct spatial fingerprint given by the tunneling probability of the orbital involved in its rate-limiting step. The positive current lobes match the tunneling pattern of orbital 1 and the negative lobes match orbitals 3 and 4, so the maps are read as proving that both configurations contribute to the same excited state; applying a few hundred millivolts of bias progressively suppresses one path and enhances the other, turning the visualization into a controlled switch. The authors note that the relative weights of the two configurations are not fixed by this scheme, since a different choice of weights can reproduce the contrast; what is established is the coexistence of both occupations, not their coefficients.","pith_inferences":["The paper does not exploit the quantitative side of the fingerprints: if the maps are proportional to Dyson-orbital densities, the relative intensity of positive and negative lobes should encode the configuration weights (TD-DFT here gives 74% for the $1\\to 2$ transition and 26% for $2\\to 4$), turning the coexistence proof into a measurement of superposition coefficients.","Because the rate equations already contain substrate reorganization, the same mapping should be able to track how configuration weights change when the molecule sits on different decoupling layers or couples to different metallic environments.","The bias-controlled selection of one decay path over the other is, in miniature, the same competition that governs charge separation at donor-acceptor interfaces in organic photovoltaics, so the technique offers a single-molecule test bed for how configuration mixing steers the direction of photoinduced charge transfer."],"forward_implications":["A zero-bias photocurrent map on a single open-shell molecule can serve as a real-space readout of which orbitals participate in an excited state's decay, and not merely as a measure of total photocurrent.","Positive bias above about 150 mV pushes the decay through the $S_0^{2-}$ path, and the map collapses onto the orbital 1 pattern, isolating configuration B; negative bias below about $-150$ mV favors the $S_0$ path and exposes the orbital 3/4 pattern, isolating configuration A.","Under illumination the $\\mathrm{d}I/\\mathrm{d}V$ maps develop structured, location-dependent features inside the transport gap, showing that the optically populated states open tunneling channels that are absent in the dark.","The level scheme extracted from $\\mathrm{d}I/\\mathrm{d}V$ onsets and TEPL spectra, combined with the reorganization threshold function, reproduces the bias values at which the photocurrent maps switch character."],"supporting_citations":[{"why":"the orbital-resolved single-molecule photocurrent channel method that this work extends from closed-shell molecules to an open-shell multiconfigurational anion","marker":"[18]"},{"why":"supplies the tip-position-dependent absorption mechanism used to set the position-dependent excitation rate of the model","marker":"[19]"},{"why":"prior PTCDA-on-NaCl study that provides the orbital framework, level assignments, and TEPL patterns the present model builds on","marker":"[4]"},{"why":"the sub-nanometre-resolution photoluminescence imaging work that grounds the position-dependent nanocavity coupling used in the maps","marker":"[22]"},{"why":"single-electron charging experiments whose level energies are consistent with the five-state energy scheme of the model","marker":"[24]"},{"why":"supports the charge-state energetics and resonance assignments of molecules on NaCl films used to fix state energies","marker":"[25]"},{"why":"formal justification for describing tunneling transitions by Dyson-orbital densities, which the model approximates with orbital slices","marker":"[33]"},{"why":"supplies the transition-density / plasmon-potential coupling scheme used to compute position-dependent absorption and emission rates","marker":"[37]"}],"fun_headline_variants":["Light-STM exposes both configurations of a molecule's excited doublet","Bias switches which configuration of a molecule's excited state you see","Photocurrent maps reveal two faces of a single molecule's doublet","Bidirectional photocurrent exposes a doublet's dual configuration"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire spatial assignment rests on approximating each tunneling transition's Dyson orbital by a single Kohn-Sham orbital evaluated on a flat plane about 5 Å above the molecule, so if the true Dyson orbitals differ substantially from those orbital slices, the matching of map features to configurations A and B would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Light-STM exposes both configurations of a molecule's excited doublet","Bias switches which configuration of a molecule's excited state you see","Photocurrent maps reveal two faces of a single molecule's doublet","Bidirectional photocurrent exposes a doublet's dual configuration"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001115,"raw_usage":{"total_tokens":4691,"prompt_tokens":1044,"completion_tokens":3647,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":660,"completion_tokens_details":{"reasoning_tokens":3572}},"tokens_in":660,"tokens_out":3647,"duration_ms":24745,"temperature":1.0,"reasoning_tokens":3572,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T21:46:24.413407+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute true Dyson orbitals for the $D_1^- \\to S_0$ and $D_1^- \\to S_0^{2-}$ transitions with a correlated many-body method, including the NaCl substrate in the model, and compare their constant-height slices at $z_T = 9.4$ a.u. with the Kohn-Sham orbital 1 and orbital 3/4 slices used here; if the dominant Dyson densities differ appreciably from those slices, the claimed identification of the positive-current lobes with configuration B and the negative-current lobes with configuration A is falsified. A purely experimental second check: re-record the zero-bias photocurrent map at a substantially different tip height, since the orbital fingerprints should persist if they are genuine tunneling channels rather than features of the single chosen plane.","supporting_citations":[{"cited_title":"Sellies, J","cited_arxiv_id":null,"evidence_quote":"prior PTCDA-on-NaCl study that provides the orbital framework, level assignments, and TEPL patterns the present model builds on"}],"review_version":1}