{"id":"3ef5503c-8f2e-4771-92fc-aa76d0735c88","arxiv_id":"2507.10206","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Single-cycle near-infrared pulses drive sub-femtosecond electron transfer across an STM junction, enabling atomic-resolution imaging with attosecond-scale sensitivity.","lead":"Researchers combined near-infrared light pulses with a scanning tunnelling microscope to drive and detect electron motion on sub-femtosecond time scales while imaging individual atoms. The work is a step toward watching electrons move inside atoms and molecules in real space and time, a long-standing goal of ultrafast science.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sub-fs current duration is a TD-DFT prediction, not a measured observable; the central claim depends on unvalidated field/gap parameters.","rationale":"The reader's thermal concern is real but partially mitigated: a tau-independent thermal vector cannot produce the observed sub-cycle oscillations or phase steps, and the zero baseline at large |tau| in Fig. 2a constrains its magnitude. The more fundamental issue is that the sub-fs duration is not an observable; it is a simulation output whose inputs (field strength, gap, material) are only loosely calibrated to the experiment. This directly targets the strongest claim. The proposed field-variation test would settle whether the duration claim is robust or an artifact of the assumed 1.04 V/nm field. The reader's rationale already notes that the attosecond time resolution is inferred from a simulation matched with offsets and amplitude calibrations, which is partial agreement with my concern. I do not propose changing the CONDITIONAL verdict because the concern warrants additional checks and a clearer distinction between measured and simulated quantities, not rejection.","tokens_in":10239,"tokens_out":11889,"duration_ms":145812,"concrete_test":"Re-run the TD-DFT calculation for the same Na-cluster junction at peak fields of 0.69, 0.85 and 1.04 V/nm (the fields corresponding to 75, 115 and 171 pJ under the paper's square-root-power scaling), with all other parameters unchanged, and extract the FWHM of I(t)=dQ/dt for phi_CE=0 at tau=-0.6 fs. If the FWHM at 0.69 V/nm is greater than 1 fs, the sub-fs claim is not established for the 75 pJ condition of Fig. 3c and must be restricted to the highest pulse energy; if the FWHM remains below 1 fs across this range, the model-dependence concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim—that the lightwave-driven tunnelling current consists of an isolated electronic wave packet shorter than 1 fs—rests on the TD-DFT simulation in Fig. 3, not on a direct time-domain measurement. The experimental observable I_CEO(tau) is a lock-in demodulated current at the CEP modulation frequency, recorded as a function of the two-pulse delay tau; it is a nonlinear cross-correlation of the synthesized waveform with the junction response, not a measurement of I(t). The 985-as FWHM in Fig. 3a is an output of a simulation of two sodium clusters with a 16 Å gap and a peak field of 1.04 V/nm. The theory-experiment comparison in Fig. 3c is made by shifting the simulated curve in tau, applying a constant phase offset, and matching the amplitude to the 75 pJ experimental data. Yet the paper's own calibration in Fig. 4a assigns 1.04 V/nm to 171 pJ, so the simulation shown in Fig. 3 corresponds to the highest pulse energy, not the 75 pJ data it is compared to. If the local field at 75 pJ is about 0.69 V/nm (scaling as the square root of pulse energy), the simulated current duration may be longer than 1 fs, and the central claim would be restricted to a single, uncalibrated highest-energy condition. The absence of a direct measurement of the junction field or of the current temporal profile makes the sub-fs claim model-dependent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports an STM experiment in which two spectrally non-overlapping near-infrared pulses with variable delay tau are focused onto a low-temperature STM junction, and the carrier-envelope-phase-modulated component I_CEO(tau) of the tunnelling current is recorded by lock-in detection. The authors observe a CEP-dependent, waveform-sensitive current that oscillates on sub-femtosecond time scales in the delay scan, and they support this with time-dependent density functional theory (TD-DFT) simulations of charge transfer between two sodium clusters. The central claim is that single-cycle near-infrared pulses drive isolated electronic wave packets shorter than 1 fs, and that such attosecond lightwave-driven tunnelling currents can image a single copper adatom with atomic spatial resolution.","tokens_in":10568,"tokens_out":4386,"duration_ms":49339,"significance":"If the central claim holds, this is the first demonstration of attosecond temporal control combined with atomic spatial resolution in STM, which would open a new regime of lightwave-driven scanning probe microscopy. The experimental design is strong in several respects: the two-colour pulse synthesis with constant average power and a measured residual power modulation of 2e-5, the observation of sub-fs features in the delay-dependent CEP-modulated current, the phase-locking of the signal to the absolute CEP, and the atomic-resolution imaging capability. The TD-DFT calculation is a forward simulation driven by the measured optical spectrum rather than a fit to the current transient, which gives the theory-experiment comparison genuine predictive content. However, the sub-1-fs wave-packet duration is a simulation result, not a directly measured observable, and the field calibration used in the simulation is inconsistent with the pulse energy quoted for the main comparison. These issues affect the strength and scope of the headline claim.","major_comments":[{"comment":"","section":"Full quantum theory of NIR-induced tunnelling; Figs. 3a, 3c and 4a"},{"comment":"","section":"Full quantum theory of NIR-induced tunnelling; Fig. 3a"},{"comment":"","section":"Operating STM with NIR pulses; Figs. 1c and 1i"}],"minor_comments":[{"comment":"","section":"Full quantum theory of NIR-induced tunnelling; Fig. 3c caption"},{"comment":"","section":"Waveform synthesis for attosecond current control; inset of Fig. 1e"},{"comment":"","section":"Full quantum theory of NIR-induced tunnelling; Fig. 3h"}],"recommendation":"major_revision","confidential_remarks":"The paper is an impressive experimental advance, but the central quantitative claim of sub-femtosecond current duration is tied to a TD-DFT simulation that is compared with the wrong pulse-energy calibration in Fig. 3c. This is a load-bearing inconsistency that must be fixed before the claim can be accepted as stated. The issue is local and fixable by recalibrating the simulation or rephrasing the claim, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a real experimental step forward. The atom-resolved images with CEP-modulated NIR-driven tunnelling currents are convincing, and the two-colour synthesis for thermal stability is a smart solution to a problem that has plagued this kind of measurement. But the headline 'isolated electronic wave packets shorter than 1 fs' is a TD-DFT prediction, not a directly measured observable, and there is a calibration inconsistency between the simulation field and the pulse energy that needs to be addressed before that number is taken literally.\n\nWhat's new: CEP-modulated currents in STM have been reported (Garg-Kern 2020, Hu 2021), and attosecond field emission from tips is established. Here they combine the two at true atomic scale, image a single Cu adatom on Ag(100), and use a two-colour scheme to keep the thermal load stable while sweeping the CEP. The thermal control is genuinely careful: they show that roughly 10^-4 power modulations produce thermal currents that would swamp the lightwave signal, and their photodiode data show a tau-independent 2e-5 power modulation at the modulation frequency. The phase-locking of I_CEO to tau, including the sub-fs step structure, is clean evidence of sub-cycle field control.\n\nThe TD-DFT does a credible job of reproducing the tau dependence of both amplitude and phase. The picture of photon-assisted tunnelling through an effectively reduced barrier, with a non-instantaneous response set by the Keldysh time, is plausible and worth pursuing. They also provide source data, which is appreciated.\n\nSoft spots. First, the 985 as current duration is an output of the simulation, not a measurement. The experiment records I_CEO(tau) as a lock-in demodulated cross-correlation, which is not a time-domain trace of the current. That's a fine way to infer sub-cycle control, but it should be labeled as inference. Second, the field calibration looks inconsistent: Fig. 4a assigns 1.04 V/nm to 171 pJ, while Fig. 3c uses 1.04 V/nm for the 75 pJ data. If the local field scales as sqrt(pulse energy), the 75 pJ field would be roughly 0.69 V/nm, and the simulated charge transfer would likely be slower. The authors need to explain whether the simulation field was chosen to match the current amplitude or whether there is a different calibration. Third, there are no error bars on I_CEO or the phase, and while the tau-independent power modulation argues against a thermal vector, a constant thermal background could still bias the phase for small signals. A direct null with one of the two pulses blocked would settle this.\n\nWho is this for: anyone in ultrafast science, STM, or attosecond metrology. I would send it to referees and expect a conditional acceptance: fix the calibration, separate measured from simulated quantities in the abstract, and add error bars. The core experimental result—atomic-resolution imaging with sub-cycle lightwave-driven currents—is solid and will stand regardless.","headline":"A genuine advance in lightwave-driven STM with atomic resolution, but the sub-femtosecond wave-packet claim is a simulation output tied to a field calibration that needs checking.","tokens_in":805,"tokens_out":823,"would_cite":true,"duration_ms":54842,"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":"Scanning tunnelling microscopy can now image single atoms with attosecond-scale, sub-femtosecond charge-transfer currents driven by single-cycle near-infrared light pulses.","keywords":["attosecond","scanning tunnelling microscopy","lightwave-driven tunnelling","carrier-envelope phase","single-cycle pulses","time-dependent density functional theory","atomic resolution","petahertz electronics"],"falsifier":"Measure I_CEO(τ) for inter-pulse delays τ far outside the region of pulse overlap (e.g., |τ| > 10 fs), where no waveform asymmetry or sub-cycle tunnelling should occur; if a non-zero demodulated current with a phase tracking the power modulation persists, the signal is contaminated by a thermal artifact. Alternatively, replace one of the two pulses with a spectrally identical copy so that the superposition has no waveform asymmetry but identical average power; if a comparable I_CEO amplitude remains, the claim of attosecond charge transfer would be falsified.","tokens_in":10073,"feed_emoji":"⚛️","tokens_out":3464,"duration_ms":39030,"temperature":0.7,"pith_summary":"This paper claims to have achieved the first combination of attosecond temporal resolution with atomic spatial resolution in scanning tunnelling microscopy (STM). By synthesizing phase-controlled single-cycle near-infrared pulses and keeping the tip-sample junction thermally stable, the authors show that tunnelling currents are modulated on sub-cycle time scales and that these currents can image a single copper adatom on a silver surface. The central claim is that single-cycle near-infrared waveforms drive isolated electronic wave packets shorter than 1 fs, and that the resulting lightwave-driven current retains angstrom-scale spatial resolution. The paper supports this with a joint experiment and time-dependent density functional theory (TD-DFT) simulation, in which the simulated current flows within a window of about 985 attoseconds.","feed_headline":"Sub-femtosecond currents now resolve single atoms","feed_subtitle":"Phase-controlled single-cycle light pulses drive attosecond tunnelling currents in an STM and image a copper atom.","key_machinery":"The central mechanism is the CEP-modulated lightwave-driven tunnelling current I_CEO, detected by lock-in demodulation of the total tunnelling current at the carrier-envelope-offset frequency f_CEO. The waveform itself is the key object: two spectrally non-overlapping near-infrared pulses (centre frequencies 164 THz and 249 THz) are superposed with an attosecond-precise delay τ, forming single-cycle transients with a 5.2 fs intensity envelope and peak fields up to 7.6 MV/cm. The CEP of this synthesized waveform is modulated linearly in time, and the component of the tunnelling current that follows this modulation is I_CEO, which serves as a direct fingerprint of sub-cycle charge transfer. The theory side uses time-dependent density functional theory (TD-DFT) to simulate charge transfer between two sodium clusters at 16 Å separation, computing Q(t) and the resulting CEP-modulated current, which reproduces the experimental I_CEO and phase.","core_discovery":"The core discovery is that single-cycle near-infrared pulses, synthesized from two spectrally non-overlapping pulses with controlled carrier-envelope phase (CEP), can drive a CEP-dependent tunnelling current that is confined to a sub-femtosecond time window and yet retains atomic spatial resolution. The authors measure the CEP-modulated current I_CEO as a function of inter-pulse delay τ, finding oscillations on sub-femtosecond scales and phase steps that are locked to the absolute waveform. TD-DFT simulations of a sodium-cluster junction reproduce the measured I_CEO and phase, showing that the transferred charge Q(t) undergoes a sharp step near t = 0 and that the current transient has a full width at half maximum of 985 as. The same simulations show a 0.5 fs retardation between the field maximum and the current maximum, indicating that the electron response is no longer instantaneous on this timescale. Finally, the authors demonstrate atomic-resolution imaging of a single Cu adatom on Ag(100) using the lightwave-driven current, with a decay length of about 8.7 Å for the highest pulse energy.","pith_inferences":["If the 0.5 fs retardation is a genuine Keldysh-time effect, then this STM approach could be used to measure tunnelling delays in real space as a function of barrier shape and material, potentially resolving long-standing debates about tunnelling time.","The reduced decay length (~8.7 Å) suggests that excited electronic states with extended wavefunctions participate in the tunnelling process; tuning the pulse parameters might selectively enhance or suppress specific electronic states, offering chemically sensitive imaging contrast.","The decrease of I_CEO above the Cu adatom, tentatively assigned to a local work-function reduction, could be exploited as a new contrast mechanism for mapping work-function variations at surfaces with sub-femtosecond time resolution.","Extending the method to two-pulse pump-probe schemes with variable τ should allow time-resolved imaging of electron dynamics in single molecules, including charge transfer and diabatic transitions, with combined attosecond and atomic resolution."],"forward_implications":["Attosecond temporal resolution and atomic spatial resolution can coexist in a single STM instrument, enabling direct videography of electron motion at atomic length scales.","The observed sub-femtosecond current transient and the 0.5 fs retardation between field and current maxima make the intrinsic response time of electrons in solids accessible in a tunnelling geometry.","The waveform-dependent tunnelling current provides a new experimental probe of the crossover regime between multi-photon and field-driven tunnelling (Keldysh parameter κ ≈ 1), where the effective barrier is dynamically reduced by excited states.","Atomic-resolution imaging with attosecond currents works on metal surfaces and can be extended to single molecules and defects, as demonstrated on a single Cu adatom on Ag(100).","The technique establishes a path to atom-scale petahertz electronics, where electronic switching is controlled by the optical field on sub-femtosecond timescales."],"supporting_citations":[{"why":"Establishes the terahertz lightwave-driven STM approach that this paper extends from THz to near-infrared frequencies.","marker":"[22]"},{"why":"Demonstrates atomic-scale femtosecond imaging of a single molecule by THz-STM, providing the benchmark for spatial resolution that the present work improves in time resolution.","marker":"[3]"},{"why":"Reports mid-infrared lightwave-driven STM with 30 fs time resolution, the state of the art that this paper surpasses to the sub-femtosecond regime.","marker":"[24]"},{"why":"Supplies the strong-field theory of attosecond tunnelling microscopy that motivates the interpretation of the observed CEP-dependent currents.","marker":"[40]"},{"why":"Provides the time-dependent DFT methodology for CEP-modulated currents in STM junctions, which the present simulations build upon.","marker":"[42]"},{"why":"Demonstrates the synthesis of single-cycle light pulses using erbium-fibre technology, the enabling laser technology for the waveform-controlled NIR pulses used here.","marker":"[41]"},{"why":"Shows attosecond control of electrons emitted from a nanoscale metal tip, a related nanoscale strong-field platform that the STM junction now extends to atomic resolution.","marker":"[34]"},{"why":"Reports tracing attosecond electron emission from a nanometric metal tip, providing a reference for sub-cycle field-driven electron dynamics in nanoscale junctions.","marker":"[37]"},{"why":"Demonstrates attosecond coherent manipulation of electrons in tunnelling microscopy, establishing the feasibility of CEP-sensitive tunnelling currents.","marker":"[26]"},{"why":"Defines the Keldysh parameter and tunnelling adiabaticity that classify the crossover regime (κ ≈ 1) explored in this paper.","marker":"[28]"}],"fun_headline_variants":["Attosecond currents image single atoms in STM","Sub-femtosecond tunnelling resolves atomic details","Lightwave-driven STM clocks electron motion","Atomic-resolution STM reaches attosecond timescale","Single-cycle pulses drive attosecond currents"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The residual 2×10⁻⁵ power modulation at the CEP-modulation frequency produces a negligible thermal current, so that the demodulated signal I_CEO(τ) is purely electronic and reflects the waveform-dependent tunnelling charge transfer rather than a thermal artifact.","fun_headline_variants_meta":{"raw":{"variants":["Attosecond currents image single atoms in STM","Sub-femtosecond tunnelling resolves atomic details","Lightwave-driven STM clocks electron motion","Atomic-resolution STM reaches attosecond timescale","Single-cycle pulses drive attosecond currents"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000514,"raw_usage":{"total_tokens":2523,"prompt_tokens":1000,"completion_tokens":1523,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":616,"completion_tokens_details":{"reasoning_tokens":1455}},"tokens_in":616,"tokens_out":1523,"duration_ms":10823,"temperature":1.0,"reasoning_tokens":1455,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:37:15.265820+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure I_CEO(τ) for inter-pulse delays τ far outside the region of pulse overlap (e.g., |τ| > 10 fs), where no waveform asymmetry or sub-cycle tunnelling should occur; if a non-zero demodulated current with a phase tracking the power modulation persists, the signal is contaminated by a thermal artifact. Alternatively, replace one of the two pulses with a spectrally identical copy so that the superposition has no waveform asymmetry but identical average power; if a comparable I_CEO amplitude remains, the claim of attosecond charge transfer would be falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the terahertz lightwave-driven STM approach that this paper extends from THz to near-infrared frequencies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates atomic-scale femtosecond imaging of a single molecule by THz-STM, providing the benchmark for spatial resolution that the present work improves in time resolution."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports mid-infrared lightwave-driven STM with 30 fs time resolution, the state of the art that this paper surpasses to the sub-femtosecond regime."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the strong-field theory of attosecond tunnelling microscopy that motivates the interpretation of the observed CEP-dependent currents."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the time-dependent DFT methodology for CEP-modulated currents in STM junctions, which the present simulations build upon."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates the synthesis of single-cycle light pulses using erbium-fibre technology, the enabling laser technology for the waveform-controlled NIR pulses used here."},{"cited_title":"Attosecond control of electrons emitted from a nanoscale metal tip","cited_arxiv_id":null,"evidence_quote":"Shows attosecond control of electrons emitted from a nanoscale metal tip, a related nanoscale strong-field platform that the STM junction now extends to atomic resolution."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports tracing attosecond electron emission from a nanometric metal tip, providing a reference for sub-cycle field-driven electron dynamics in nanoscale junctions."},{"cited_title":"Attosecond coherent manipulation of electrons in tunneling microscopy","cited_arxiv_id":null,"evidence_quote":"Demonstrates attosecond coherent manipulation of electrons in tunnelling microscopy, establishing the feasibility of CEP-sensitive tunnelling currents."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Keldysh parameter and tunnelling adiabaticity that classify the crossover regime (κ ≈ 1) explored in this paper."}],"review_version":1}