{"id":"d8d1da45-1d89-4ada-94a9-2bb684b25606","arxiv_id":"1908.09636","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A feed-forward stabilized Er:Yb:glass laser achieves 3.5 mrad integrated phase noise, or 2.9 as timing jitter, from 1 Hz to 3 MHz, and holds frequency lock for eight hours with periodic manual pump adjustments.","lead":"Researchers stabilized the carrier-envelope phase of an Er:Yb:glass laser at 1.55 micrometers using a feed-forward control method, reporting 2.9 attosecond timing jitter and an eight-hour stability run. The result is a new performance mark for telecom-wavelength frequency combs used in precision metrology and timing networks.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2.9 as headline jitter rests on excluding 0.1–1 Hz noise as 'spurious' without a direct measurement; if that band is intrinsic CEP noise, the full-band jitter is 20 as and the 'single-digit attosecond' claim fails.","rationale":"The paper is a plausible feed-forward CEP stabilization demonstration with a conventional f-2f detection and an out-of-loop verification arm; the central construction (Eq. 3 and the AOFS chain) is internally coherent. The strongest claim, however, is the numerically small jitter number. The full measurement already contains an unexcluded band (0.1–1 Hz) that raises the integrated jitter from 2.9 as to 20 as. The authors' justification for dropping this band is a reference to prior flicker-noise work plus an assertion that the noise is 'spurious' and seen in all OOL systems. That justification is exactly the load-bearing step: if the low-frequency power is intrinsic to the laser (as 'quantum noise in the laser cavity' implies), the system does not deliver single-digit attosecond timing jitter over the full measurement band. The reader's conditional verdict identifies this same assumption, and I agree. The eight-hour stabilization claim is also weaker than the abstract suggests because the pump power is manually adjusted every half hour, but that affects a secondary claim; the band-selection issue directly controls the headline number. A cross-correlation or extended dead-time-free acquisition would settle whether the excluded band is a common-mode artifact or real phase noise.","tokens_in":8087,"tokens_out":12369,"duration_ms":132271,"concrete_test":"Run a second, fully independent OOL f-2f branch and compute the cross-spectral density of its phase noise against the existing OOL branch over 0.01 Hz–1 Hz. If the 0.1–1 Hz excess is correlated between branches, it is a real property of the stabilized laser and must be included in the headline number (~20 as); if it is uncorrelated, it is a characterization artifact and the 1 Hz–3 MHz 2.9 as figure can stand. A simpler first check is to acquire a dead-time-free phase record for several hours with the pump-adjustment loop disabled and recompute the integrated jitter down to 0.01 Hz; if the 0.1–1 Hz band remains populated, that noise cannot be excluded from the performance claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the out-of-loop CEP jitter of 3.5 mrad / 2.9 as over 1 Hz–3 MHz. The measurement shown in Fig. 3 yields 25 mrad / 20 as when integrated from 0.1 Hz. The paper excludes the 0.1–1 Hz band as '1/f spurious noise... attributed to quantum noise in the laser cavity [24]' and reports the narrower band as the system performance. The only support is a citation to prior work, not an in-manuscript measurement establishing that this band is an artifact of the OOL characterization rather than an intrinsic property of the stabilized laser. This matters because a quantum-noise origin makes the noise physically real and directly relevant to the stated applications in timing and spectroscopy; omitting it overstates the achieved stabilization by roughly an order of magnitude in jitter (20 as vs 2.9 as). The abstract's 'single-digit attosecond stabilization' and the record-low claim depend on this band-selection choice. A secondary but related concern is that the eight-hour stabilization is maintained by manual pump-power adjustments about every half hour with a stated large phase-noise penalty, so 'long-term stabilization' in the abstract should be read as frequency lock with human intervention, not autonomous phase stabilization.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports carrier-envelope phase (CEP) stabilization of a SESAM mode-locked Er:Yb:glass laser at 1.55 µm using the feed-forward technique. The in-loop f-2f beat note drives an acousto-optic frequency shifter in the output beamline, and a separate out-of-loop f-2f interferometer is used to verify the stabilized signal. The authors report an integrated phase noise of 3.5 mrad (1 Hz–3 MHz), corresponding to 2.9 as rms timing jitter at 1.55 µm, and an eight-hour record of the stabilized frequency with 0.16 Hz rms deviation. The full 0.1 Hz–3 MHz integration gives 25 mrad (20 as), which the authors attribute to low-frequency 'spurious' noise and exclude from the headline claim.","tokens_in":8343,"tokens_out":5260,"duration_ms":46464,"significance":"If the headline result is accepted, this is a useful demonstration of feed-forward CEP stabilization on a telecom-wavelength solid-state laser, with an out-of-loop verification that is considerably stronger than an in-loop lock alone. The authors honestly report the full-band 0.1 Hz–3 MHz value in the body of the paper, and the experimental description is detailed and reproducible in structure. The out-of-loop measurement methodology and the use of a rubidium-derived reference are strengths. However, the central performance claim (2.9 as, single-digit attosecond stabilization) depends on excluding the 0.1–1 Hz band without a direct measurement in this work, and the 'eight-hour stabilization' statement in the abstract overstates what was actually demonstrated, as the text explains that the system required manual pump-power adjustments about every half hour. These are load-bearing issues for the claims as stated, though they are addressable by additional measurements or by reframing the claims.","major_comments":[{"comment":"The headline result, 3.5 mrad / 2.9 as, is obtained by integrating the phase noise only from 1 Hz to 3 MHz, excluding the 0.1–1 Hz band as 'spurious noise ... attributed to quantum noise in the laser cavity [24].' The manuscript offers no measurement that supports this exclusion; the only support is the citation to Ref. [24], and the cited origin (quantum noise in the laser cavity) describes a physically real, intrinsic noise source rather than a measurement artifact. If that band is intrinsic to the stabilized laser, the full-band jitter is 20 as, an order of magnitude larger, and the 'single-digit attosecond' claim fails. The authors should either provide a direct diagnostic (for example, a second out-of-loop detection arm, a coherence test between two OOL measurements, or a comparison of the disputed band with the in-loop error signal) or report the 0.1 Hz–3 MHz value as the headline and present the 1 Hz–3 MHz value as a conditional, band-limited performance metric.","section":"Fig. 3 and the paragraph beginning 'The stabilized OOL signal...'"},{"comment":"The abstract's claim of 'long-term stabilization over a duration of eight hours' is not supported as stated. The body text explains that long-term stability is maintained by manual adjustment of the oscillator pump power approximately every half hour, and that each adjustment 'introduces large amounts of phase noise.' Figure 4 therefore documents an eight-hour frequency lock relative to f_LO with human intervention, not an autonomous CEP phase stabilization. Please revise the abstract and the interpretation to distinguish frequency locking from CEP phase stabilization, and state explicitly that the eight-hour record involved periodic manual pump adjustments.","section":"Abstract and the paragraph beginning 'Because of the natural slow drift...'"},{"comment":"The conclusion that the reported 3.5 mrad (1 Hz–3 MHz) is 'a significant improvement' over the 120 mrad (0.01 Hz–1 MHz) of Ref. [16] is not supported because the integration intervals are different, and the disputed low-frequency band is exactly where the two measurements differ most. To make the comparison meaningful, the authors should report phase noise integrated over the same frequency range as Ref. [16], or explicitly state that the comparison is band-limited and therefore not a like-for-like improvement.","section":"Conclusion and comparison with Ref. [16]"}],"minor_comments":[{"comment":"Affiliation 1 contains a typo: 'SLAC National Acceleratory Laboratory' should be 'SLAC National Accelerator Laboratory.'","section":"Affiliations"},{"comment":"The sentence 'may lead to to excess phase-noise' contains a duplicated word ('to to'); please correct it.","section":"Introduction, second paragraph"},{"comment":"Reference 11 lists the author as 'T. W. Hksch'; this should be 'T. W. Hänsch.'","section":"Reference 11"},{"comment":"The reference list appears both in a compact form and in a full form in the submitted text; please consolidate to a single list.","section":"References (general)"},{"comment":"The abbreviation 'IPND' is used inconsistently (for example, 'IPND of 25 mrad' versus 'IPN at 3.5 mrad'); please define and use one abbreviation consistently.","section":"Fig. 3 and Eq. (4)"},{"comment":"The phrase '0.16 Hz rms variance' is not a standard statistical expression; please specify whether 0.16 Hz is the standard deviation of the frequency record over the eight-hour period.","section":"Fig. 4 caption"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a plausible candidate for publication after revision. The out-of-loop verification and the honest reporting of the full 0.1 Hz–3 MHz result are positive features. The main reservations are the band-selection for the headline number and the overstatement of 'eight-hour stabilization' given the manual interventions. In the editor's consideration, note that the specific attribution of the 0.1–1 Hz noise to quantum noise rests solely on a citation to Ref. [24], a paper co-authored by a co-author of this manuscript; given the load-bearing role of that exclusion, direct measurement or analysis in this paper would be preferable. Also, the abstract should be revised so that 'long-term stabilization' does not imply autonomous phase stabilization."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this is a solid, incremental experimental result—lowest reported CEP jitter for a feed-forward stabilized Er:Yb:glass laser, with an honest out-of-loop measurement. The headline 2.9 as depends on excluding the 0.1–1 Hz band, and the eight-hour “stabilization” is really a frequency lock with manual pump tweaks every half hour. Both caveats are in the paper, but the abstract oversells.\n\nWhat’s new: combining feed-forward CEP stabilization with an amplified monolithic f-2f interferometer on a SESAM mode-locked Er:Yb:glass oscillator. The measurement is reproducible in principle: they report 3.5 mrad integrated phase noise from 1 Hz to 3 MHz, corresponding to 2.9 as at 1.55 µm. That beats Kundermann et al.’s 120 mrad, though note the integration bands differ (their 120 mrad is 0.01 Hz–1 MHz), so the comparison is not perfectly apples-to-apples. The out-of-loop verification with a second f-2f interferometer is a real strength, and the paper is transparent that integrating from 0.1 Hz gives 25 mrad / 20 as.\n\nSoft spots, in order of weight:\n\n1. The 0.1–1 Hz exclusion. The paper attributes that band to “1/f spurious noise” from quantum noise in the cavity, citing prior work [24], but doesn’t measure or demonstrate that this band is an artifact rather than intrinsic CEP noise in their system. If it’s intrinsic, the full-band jitter is 20 as and the “single-digit attosecond” claim collapses. Given the applications they advertise (spectroscopy, timing networks), low-frequency noise is not irrelevant. They should either measure this band under controlled conditions or at least report the full-band number in the abstract alongside the narrowband number.\n\n2. The long-term claim. Eight hours of lock is maintained by manual pump-power adjustment about every half hour, with a large phase-noise penalty each time. That’s fine to report as a proof-of-principle frequency lock, but “long-term stabilization” in the abstract should be qualified. The paper itself says a slow PID would fix this, which is reasonable future work.\n\n3. Minor: no error bars or repeated measurements on the phase noise. Given that the headline is a record, one would like to see run-to-run variation. Also the 31 dB SNR in the out-of-loop beat is modest; a bit more detail on how the measurement noise floor affects the integrated jitter would help.\n\nThe citation pattern is fine—Steinmeyer’s own work on flicker noise is the relevant literature, and the paper doesn’t hide the caveat. Nothing looks like overfitting or circular reasoning; it’s a measurement against a rubidium reference.\n\nWho this is for: anyone working on CEP stabilization or frequency combs at telecom wavelengths. It’s a useful data point and a clear engineering advance, not a paradigm shift. A serious referee can fix the framing issues; the measurement appears sound. I’d accept it for peer review and recommend conditional acceptance after the abstract is adjusted and the low-frequency exclusion is either justified with measurement or reported as a full-band number.","headline":"Solid incremental record for feed-forward CEP stabilization, but the 2.9 as headline excludes the 0.1–1 Hz band and the long-term lock is manual — the paper is honest about both, the abstract is not.","tokens_in":8890,"tokens_out":1857,"would_cite":false,"duration_ms":17019,"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":"Feed-forward control stabilizes a 1.55 µm Er:Yb:glass laser's carrier-envelope phase to 3.5 mrad, or 2.9 as of timing jitter.","keywords":["carrier-envelope phase stabilization","feed-forward technique","acousto-optic frequency shifter","f-2f interferometry","Er:Yb:glass laser","SESAM mode locking","timing jitter","optical frequency comb"],"falsifier":"Take two independent out-of-loop f-2f interferometers on the same stabilized beam and cross-correlate their phase-noise spectra; coherent 0.1–1 Hz noise would prove the low-frequency flicker is intrinsic and force the 20 as full-band number, while incoherent noise would justify the 3.5 mrad claim.","tokens_in":7879,"feed_emoji":"🔒","tokens_out":13036,"duration_ms":110329,"temperature":0.7,"pith_summary":"Few-cycle lasers need shot-to-shot consistency of the carrier-envelope phase (CEP) for frequency metrology, attosecond pulse generation, and timing distribution over fiber networks. This paper reports that a SESAM (semiconductor saturable absorber mirror) mode-locked Er:Yb:glass laser at 1.55 µm can hold its CEP fixed with feed-forward control: the laser's carrier-envelope offset beat is measured by f-2f interferometry and used to drive an external acousto-optic frequency shifter, rather than feeding back onto the cavity. The authors measure an out-of-loop integrated phase noise of 3.5 mrad from 1 Hz to 3 MHz, corresponding to 2.9 attoseconds of rms timing jitter at 1.55 µm, and a lock sustained over eight hours with 0.16 Hz rms frequency drift. They attribute the additional noise seen down to 0.1 Hz (20 as full-band) to known low-frequency flicker, and present the 1 Hz–3 MHz number as the performance of the stabilization, an improvement over the previous 120 mrad feed-forward result for erbium lasers.","feed_headline":"Feed-forward locks a 1.55 µm laser's phase to 2.9-attosecond jitter","feed_subtitle":"Feed-forward control of a SESAM Er:Yb:glass laser achieves 3.5 mrad integrated phase noise from 1 Hz to 3 MHz.","key_machinery":"The load-bearing element is the feed-forward chain built around the acousto-optic frequency shifter (AOFS). The measured carrier-envelope offset frequency $f_{\\mathrm{CEO}}$ is not used to correct the cavity; it is added to a local-oscillator frequency $f_{\\mathrm{LO}}$ and the sum drives the AOFS, which subtracts this drive from every comb line outside the cavity. The identity $f_{\\mathrm{OOL}}=f_n-f_{\\mathrm{AOFS}}=n f_{\\mathrm{REP}}-f_{\\mathrm{LO}}$ is what converts CEO stabilization into the stability of an RF reference. Supporting pieces are the f-2f self-referencing detection with amplifiers before the interferometer to raise the signal-to-noise ratio, and the SESAM mode-locked Er:Yb:glass oscillator, whose intrinsically low timing jitter the authors credit for the low phase noise.","core_discovery":"The paper's central claim is that the carrier-envelope phase of a soliton mode-locked Er:Yb:glass laser can be stabilized by feed-forward rather than by cavity feedback, and that this yields attosecond-level timing jitter at 1.55 µm. In the setup, the oscillator output is split: one arm is amplified, spectrally broadened to more than an octave in a highly nonlinear fiber, and sent into an f-2f interferometer whose carrier-envelope offset beat $f_{\\mathrm{CEO}}$ is mixed with a reference comb derived from a 10 MHz rubidium clock, producing an 80 MHz drive $f_{\\mathrm{AOFS}}=f_{\\mathrm{CEO}}+f_{\\mathrm{LO}}$. The other arm passes through an acousto-optic frequency shifter driven by this signal, so every comb line is shifted down and the out-of-loop offset becomes $f_{\\mathrm{OOL}}=n f_{\\mathrm{REP}}-f_{\\mathrm{LO}}$: the CEP is then linked to the phase of a stable radio-frequency local oscillator instead of to the cavity. Out-of-loop phase-noise measurements give 3.5 mrad integrated phase noise (1 Hz–3 MHz), or 2.9 as rms timing jitter at 1.55 µm; including the 0.1–1 Hz band raises the integral to 20 as, which the authors attribute to known low-frequency flicker from quantum noise in the cavity rather than to a failure of the stabilization. An eight-hour record shows the out-of-loop beat staying locked with 0.16 Hz rms variance, with occasional manual pump-power adjustments to keep the acousto-optic shifter near its optimum.","pith_inferences":["Beyond the paper's claims, the feed-forward architecture is not specific to Er:Yb:glass: any SESAM mode-locked oscillator with a detectable CEO beat could use the same AOFS chain, so the demonstrated 3.5 mrad figure is a benchmark for the method rather than for this laser alone.","Adding a slow pump-power feedback loop, which the authors name as future work, would likely suppress the 0.1–1 Hz drift and could bring the full-band integral down to the 3.5 mrad value; the present paper only relies on manual adjustments.","For fiber timing networks, the stabilized comb's phase is set by an RF reference rather than by the cavity, so distributing the CEO-stabilized pulse train over long fiber links should preserve sub-10-attosecond timing at 1.55 µm; the paper does not test this propagation."],"forward_implications":["The CEO frequency of an Er:Yb:glass laser can be stabilized by feed-forward with out-of-loop integrated phase noise of 3.5 mrad from 1 Hz to 3 MHz.","This corresponds to 2.9 as rms timing jitter at 1.55 µm, which the authors present as a substantial improvement over the prior 120 mrad feed-forward result for erbium lasers.","Stabilization holds for at least eight hours, with 0.16 Hz rms drift of the out-of-loop beat relative to the rubidium-derived reference.","Because the AOFS acts outside the cavity, short-term phase stabilization and long-term drift are decoupled; residual long-term drift is currently managed by occasional pump-power adjustments and could be automated with a slow feedback loop.","The high signal-to-noise ratio from amplifying the signal before the f-2f interferometer is identified as a key enabler, together with the intrinsically low timing jitter of Er:Yb:glass lasers."],"supporting_citations":[{"why":"Defines the f-2f self-referencing scheme used to detect the carrier-envelope offset beat in both in-loop and out-of-loop beamlines.","marker":"[6]"},{"why":"Introduces the feed-forward technique of shifting the comb with an AOFS driven by the CEO beat.","marker":"[12]"},{"why":"Demonstrates long-term feed-forward CEP stabilization and compensation of the AOFS angular chirp.","marker":"[13]"},{"why":"Reports the previous lowest feed-forward CEP jitter (120 mrad) for erbium lasers, the baseline this work improves on.","marker":"[16]"},{"why":"Establishes the intrinsically low timing jitter of Er:Yb:glass lasers that the authors credit for the low phase noise.","marker":"[18]"},{"why":"Shows that amplifying the signal before the f-2f interferometer raises the signal-to-noise ratio and reduces shot-noise contributions.","marker":"[21]"},{"why":"Attributes low-frequency flicker noise in CEP stabilization to quantum cavity noise, supporting the decision to exclude the 0.1–1 Hz band from the headline figure.","marker":"[24]"}],"fun_headline_variants":["Feed-forward locks Er:Yb:glass laser to 2.9-as timing jitter","Er:Yb:glass laser achieves 2.9-as jitter via feed-forward","Single-digit attosecond stabilization at telecom wavelengths","Attosecond CEP control in a 1.55-µm fiber laser","Feed-forward stabilizes carrier-envelope phase to 2.9 as"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline 2.9 as figure rests on assuming that the added phase noise seen when integrating down to 0.1 Hz is spurious low-frequency flicker rather than real carrier-envelope noise; if that assumption is wrong, the full-band jitter is 20 as.","fun_headline_variants_meta":{"raw":{"variants":["Feed-forward locks Er:Yb:glass laser to 2.9-as timing jitter","Er:Yb:glass laser achieves 2.9-as jitter via feed-forward","Single-digit attosecond stabilization at telecom wavelengths","Attosecond CEP control in a 1.55-µm fiber laser","Feed-forward stabilizes carrier-envelope phase to 2.9 as"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000643,"raw_usage":{"total_tokens":3009,"prompt_tokens":1047,"completion_tokens":1962,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":663,"completion_tokens_details":{"reasoning_tokens":1862}},"tokens_in":663,"tokens_out":1962,"duration_ms":13385,"temperature":1.0,"reasoning_tokens":1862,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:33:05.292750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take two independent out-of-loop f-2f interferometers on the same stabilized beam and cross-correlate their phase-noise spectra; coherent 0.1–1 Hz noise would prove the low-frequency flicker is intrinsic and force the 20 as full-band number, while incoherent noise would justify the 3.5 mrad claim.","supporting_citations":[{"cited_title":"Carrier-envelope offset phase control: A novel concept for absolute optical frequency measurement and ultrashort pulse generation,","cited_arxiv_id":null,"evidence_quote":"Defines the f-2f self-referencing scheme used to detect the carrier-envelope offset beat in both in-loop and out-of-loop beamlines."},{"cited_title":"Direct frequency comb synthesis with arbitrary offset and shot-noise-limited phase noise,","cited_arxiv_id":null,"evidence_quote":"Introduces the feed-forward technique of shifting the comb with an AOFS driven by the CEO beat."},{"cited_title":"Long-term carrier-envelope-phase-stable few-cycle pulses by use of the feed-forward method,","cited_arxiv_id":null,"evidence_quote":"Demonstrates long-term feed-forward CEP stabilization and compensation of the AOFS angular chirp."},{"cited_title":"Comparison of different carrier-envelope frequency stabilization methods for a high performance DPSSL frequency comb,","cited_arxiv_id":null,"evidence_quote":"Reports the previous lowest feed-forward CEP jitter (120 mrad) for erbium lasers, the baseline this work improves on."},{"cited_title":"Nearly quantum-noise-limited timing jitter from miniature Er : Yb : glass lasers,","cited_arxiv_id":null,"evidence_quote":"Establishes the intrinsically low timing jitter of Er:Yb:glass lasers that the authors credit for the low phase noise."},{"cited_title":"Active f-to-2f interferometer for record-low jitter carrier- envelope phase locking,","cited_arxiv_id":null,"evidence_quote":"Shows that amplifying the signal before the f-2f interferometer raises the signal-to-noise ratio and reduces shot-noise contributions."},{"cited_title":"On the origin of flicker noise in carrier-envelope phase stabilization,","cited_arxiv_id":null,"evidence_quote":"Attributes low-frequency flicker noise in CEP stabilization to quantum cavity noise, supporting the decision to exclude the 0.1–1 Hz band from the headline figure."}],"review_version":1}