REVIEW 2 major objections 4 minor 4 references
Integrated Architecture for the Automated Generation and Coil Stabilization of a PZT-Enabled Microcomb
T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read A single stationary pump laser and a single PZT voltage sequence can generate and fully stabilize a soliton microcomb by locking two comb lines to a 16-meter coil resonator.
desk verdict A useful single-laser, single-actuator microcomb stabilization architecture with a clean noise model, but the headline repetition-rate phase noise number rests on an inconsistent and undocumented division factor. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing mechanism is dual-wavelength Pound-Drever-Hall locking of two comb lines to two modes of the same 16-meter silicon nitride coil resonator, whose 12 MHz free-spectral range forms a fine frequency grid. Because the comb has only two degrees of freedom, locking the pump and one line at mode index $r$ fixes both, and the remaining lines are constrained by $\nu_m=\nu_p+m f_{\mathrm{rep}}$. The PZT actuator is the second pillar: a single DC-to-70 MHz electrode both sweeps the resonator across the stationary pump to ignite the soliton and then modulates the repetition rate to steer any chosen line onto a coil resonance; the paper measures $-170$ MHz/V tuning strength and sub-nW DC bias power. The noise-propagation formula above quantifies how correlated noise on the two locks transfers to each comb line.
What would settle it
Measure the relative frequency noise between the two locked wavelengths by heterodyning the pump and soliton line and compare it with the sum of their individually locked noise spectra; if the relative noise equals the uncorrelated sum rather than falling below it, the common-mode thermorefractive noise rejection that carries the repetition-rate claim is not happening. Equivalently, a direct beat measurement of the 108 GHz repetition rate against a high-stability reference should show a noise floor above the predicted coil thermorefractive limit scaled by the division factor.
Extended reading notes
Core claim
The central claim is that dual-point locking to one integrated coil resonator removes the need for multiple lasers, fast pump tuning, and high-power actuators that current stabilized microcombs require. With the pump laser locked to one coil mode and a soliton line at 1560 nm locked to another coil mode 11 FSRs away, the two comb degrees of freedom are both constrained, so any comb line $m$ has frequency noise $S_m(f)=S_p(f)(1-m/r)^2+S_r(f)(m/r)^2+2C(f)\sqrt{S_p(f)S_r(f)}(1-m/r)(m/r)$, where $r=11$ is the locked line index. Measured noise follows this model, and the locked lines reach the coil's thermorefractive noise floor at mid-range offsets, yielding comb-line integral linewidths of 66 Hz and repetition-rate phase noise of $-118$ dBc/Hz at 10 kHz offset after division to 10 GHz. The paper argues this is the simplest stabilized comb architecture demonstrated to date and that all components are CMOS-compatible silicon nitride.
Load-bearing premise
The two PDH locks must share the same coil cavity with strongly correlated thermorefractive noise; if the noise on the pump lock and the soliton-line lock were not mostly common-mode, the repetition rate would inherit their uncorrelated difference and would not reach the coil's low noise floor.
Editorial extensions
If this is right
- Every comb line inherits the coil resonator's low thermorefractive noise, not just the two locked lines.
- The same single-point control sequence can ignite the soliton and stabilize it, so no fast-tunable pump laser is required.
- The 12 MHz coil FSR lets any soliton line serve as the second lock point, enabling flexible choice of division factor and locked line.
- Frequency noise across the full 35 nm comb spectrum drops by four orders of magnitude at 1 kHz offset, to below $10~\mathrm{Hz}^2/\mathrm{Hz}$.
- Because both microcomb and coil are CMOS-compatible silicon nitride, the architecture points to monolithic integration of a stabilized comb.
Reading between the lines
- Inference: Using a dispersive wave far from the pump would raise the integer $r$ and increase the optical division factor, potentially lowering the repetition-rate noise floor below the demonstrated $-118$ dBc/Hz.
- Inference: A direct measurement of the correlation factor $C(f)$ between the two PDH locks, by comparing the beat between locked lines with their individual noise, would quantify how much common-mode coil noise is actually rejected and predict the achievable $f_{\mathrm{rep}}$ floor for other line choices.
- Inference: Integrating a pump laser on the same silicon nitride chip, as sketched in the paper, would test whether the single-laser architecture survives the added noise of an on-chip laser source; flexible lock-point selection could then compensate for that laser's higher intrinsic noise.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript reports a Si3N4 dissipative Kerr soliton microcomb architecture that uses a single fixed-frequency pump laser and a single integrated PZT actuator, driven by a programmed voltage sequence, both to initiate a soliton and to lock the pump and one soliton line to a 16-m Si3N4 coil resonator via Pound-Drever-Hall locking. The authors claim that this dual lock transfers the coil's low thermorefractive noise to all comb lines, giving integral linewidths as low as 66 Hz, a 40 dB (four orders of magnitude) suppression of the 1-kHz frequency noise across the 35-nm comb spectrum, and a soliton repetition-rate phase noise equivalent to -118 dBc/Hz at 10 GHz when divided down from 108 GHz. The Supplement contains a two-degree-of-freedom noise-propagation model, a dual-wavelength locking characterization, and the repetition-rate beat-note measurement setup.
Significance. If the central claims are correct, this is a significant architectural simplification for stabilized microcombs: one pump laser, one actuator, and a single control sequence replace the multi-laser, multi-actuator stacks used in earlier demonstrations. The Supplement III noise-propagation derivation is transparent, parameter-free, and uses measured locked-line noise as input, and the analytical curves in Fig. 4b reproduce the measured line-to-line scaling. The 66 Hz linewidth and the broadband 1-kHz suppression are strong results if confirmed. However, the repetition-rate phase-noise headline rests on a measurement conversion that is incompletely documented and internally inconsistent, and the common-mode correlation that justifies the thermorefractive-noise cancellation for frep is not directly measured. These gaps must be closed before the metrological claims can be accepted.
major comments (2)
- [Supplement IV / Methods ('Soliton line frep noise measurement')] The extraction of the soliton repetition-rate phase noise, on which the headline -118 dBc/Hz at 10 GHz rests, is incomplete and internally inconsistent. The Methods section states that the beat-note PSD is divided by the square of the soliton line mode number (r=11), while Supplement IV states that 'with our pump at 1550 nm and soliton line at 1560 nm, our division factor is .9872' and refers to an equation that is not present in the provided text. These two statements cannot both be correct: 0.9872 is neither r=11 nor the 108-GHz-to-10-GHz division ratio (10.8) nor the wavelength ratio 1550/1560 ≈ 0.9936. Using 0.9872 in place of 11 changes the inferred frep PSD by about 20*log10(11/0.9872) ≈ 21 dB, directly affecting the -118 dBc/Hz claim. Please supply the missing equation, define every factor in the conversion, explain how the fiber-comb repetition-rate noise term and the pump-noise transfer are removed, reconcile the factor with the measured optical frequency separation between the pump and soliton line, and re-state the affected headline numbers.
- [Results D / Supplement III Eq. (12)] The common-mode thermorefractive-noise cancellation that is central to the frep claim is not demonstrated by a measured correlation function. Supplement III provides Eq. (13) for extracting C(f), but the paper does not report C(f). Instead, Results D states that at 1 kHz the measured comb-line noise is lower than the C=0 model and attributes this to correlated environmental perturbations. For modes between the two locked lines (0 < m/r < 1), Eq. (12) predicts that a positive correlation raises the predicted line noise above the C=0 prediction; a lower-than-C=0 measurement would imply anti-correlated noise for those modes under that equation. Please report the measured C(f) obtained from Eq. (13) or an equivalent direct measurement, and reconcile the 1-kHz interpretation with the sign convention in Eq. (12). This is not a cosmetic point: the low-frequency frep floor and the statement that both locks share coil TRN depend on the sign and magnitude of C.
minor comments (4)
- [Results D] The phrase 'low reverse 1/p integral linewidths' should read '1/π integral linewidths'.
- [Fig. 4c caption and Results D] The caption states that the TRN limit is 'scaled by the optical division factor of 112'; this should be typeset as 11^2 or otherwise defined. The distinction between the comb mode-number factor r=11 used in the frep extraction and the 108-GHz-to-10-GHz division ratio (10.8) should also be made explicit, since the current wording conflates the two.
- [Methods (PZT actuation measurement; PZT soliton generation)] There are small textual errors: 'a laser is manually manually held' contains a duplicated word, and 'opposite ramp slopes ramps' is ungrammatical.
- [Results A, Eq. (1)] The inline equation for S_m(f) is garbled in the provided text; please ensure the printed equation shows the (1 - m/r)^2 and (m/r)^2 factors and the correlation term correctly.
Circularity Check
No circular derivation found; the noise model is an algebraic identity from the comb relation and the analytical curves use measured inputs. A non-circular but flagged gap: Supplement IV omits the frep conversion equation and states a division factor (0.9872) inconsistent with Methods' r=11.
full rationale
The claimed derivation chain is not circular in the sense of reducing a prediction to its inputs. The central noise-propagation formula (main text Section II A and Supplement III, Eq. 12) follows by algebra from the comb relation ν_m = ν_p + m f_rep (Supplement Eq. 1) after locking line r; it contains no fitted parameters. The dashed curves in Fig. 4b are generated from the measured frequency noise of the two locked points with C=0, so the agreement with independently measured comb-line noise is a consistency check, not a fit of the target result. PZT tuning strength, modulation bandwidth, and power consumption are direct measurements (Fig. 3, Methods). The repetition-rate phase-noise extraction is an indirect measurement, not a derivation: Methods state that 'Frequency and phase noise of frep are then obtained by dividing the PSD of the relative fluctuations by the square of the soliton line mode number (11 in this case),' which is an algebraic conversion from the same comb relation. That is not circular. I do flag a serious but non-circular support gap: Supplement IV says the beatnote 'gives the noise of the soliton's repetition rate given by the following equation' and then states 'our division factor is .9872,' while no equation is actually shown and Methods specify division by r^2 = 11^2; 0.9872 is neither 11 nor the 108-to-10 GHz division factor of 10.8, and if used would shift the headline -118 dBc/Hz figure by approximately 21 dB. This is an omitted equation and internal inconsistency in the measurement chain, but it does not make a prediction equivalent to an input. The common-mode TRN assumption for frep is a stated physical modeling assumption, not a restatement of the demonstrated result. Therefore the circularity score is low, with the flagged measurement-support issue noted separately.
Assumptions & free parameters
assumptions (4)
- domain assumption The soliton comb satisfies the energy-conservation relation ν_m = ν_p + m f_rep with exactly two fluctuating degrees of freedom.
- domain assumption The two selected coil modes experience common-mode thermorefractive noise, so cavity noise cancels in the repetition-rate signal.
- domain assumption The PZT actuator preserves the resonator Q and provides linear, broadband tuning without disturbing the soliton state.
- domain assumption The self-referenced fiber comb measurement chain isolates soliton frep noise and is quieter than the soliton at the measured offsets.
Cite this review
Pith. "Pith review of Integrated Architecture for the Automated Generation and Coil Stabilization of a PZT-Enabled Microcomb." pith.science (2026). https://pith.science/paper/5SUUDADO
@misc{pith2026250716979,
author = {Pith},
title = {Pith review of: Integrated Architecture for the Automated Generation and Coil Stabilization of a PZT-Enabled Microcomb},
year = {2026},
howpublished = {\url{https://pith.science/paper/5SUUDADO}},
note = {Machine review of arXiv:2507.16979}
}
read the original abstract
Silicon nitride Dissipative Kerr Soliton (DKS) microcombs have emerged as a future solution to bring metrological optical frequency comb capabilities into a photonic integrated platform with mass-scale fabrication benefits. Precision applications demand low comb line phase noise as well as high repetition rate stability, but current approaches to achieve this involve complex architectures, multiple lasers, and high-power components, which are challenging to integrate to the chip scale. To achieve this goal, new architectures are needed to simplify the comb generation, actuation, and pump laser requirements, while enabling chip-integrated solutions. Here we demonstrate a greatly simplified stabilized DKS comb architecture with a single laser and a single point electronic control of both the microcomb generation and its stabilization to a coil-resonator reference. The silicon nitride microcomb is integrated with a low power, broadband PZT actuator that is driven by a simple electronic control sequence that generates a soliton and stabilizes it to the 16-meter silicon nitride coil resonator. PZT-enabled control brings flexibility and simplicity to the soliton generation and stabilization using a single CW pump laser, resulting in significantly reduced electronic and optical infrastructure. We demonstrate coil-resonator locking which suppresses the 1 kHz frequency noise by 40 dB over the 35 nm wide comb spectrum, with comb line linewidths as low as 66 Hz and 108 GHz soliton repetition rate phase noise equivalent to -118 dBc/Hz when divided down to 10 GHz. The low power PZT actuator consumes nW bias power and the coil resonator allows flexible dual locking using arbitrary comb lines. These results show a clear path towards full chip integration of stabilized soliton microcombs with simplicity and versatility absent in other schemes.
Reference graph
Works this paper leans on
-
[1]
(𝑓) *1−#$-(+𝑆'#(𝑓)*#$-(+2𝐶(𝑓)0𝑆'
INTRODUCTION Dissipative Kerr Soliton (DKS) microcombs1–4, have emerged as a promising candidate for integrated multiwavelength optical sources for a wide variety of precision applications including ultra-low phase noise microwave generators5–10 atomic clocks11, quantum sensing12,13 and precision spectroscopy14–16. Commercial stabilized fiber frequency co...
work page 2014
-
[7]
Zhao, Y . et al. All-optical frequency division on-chip using a single laser. Nature 627, 546–552 (2024). 8. Sun, S. et al. Microcavity Kerr optical frequency division with integrated SiN photonics. Nat. Photon. 1–6 (2025) doi:10.1038/s41566-025-01668-3. 9. Ji, Q.-X. et al. Dispersive-wave-agile optical frequency division. Nat. Photon. 1–6 (2025) doi:10.1...
-
[19]
Brasch, V ., Geiselmann, M., Pfeiffer, M. H. P. & Kippenberg, T. J. Bringing short-lived dissipative Kerr soliton states in microresonators into a steady state. Opt. Express, OE 24, 29312–29320 (2016). 20. Carmon, T., Yang, L. & Vahala, K. J. Dynamical thermal behavior and thermal self-stability of microcavities. Opt. Express, OE 12, 4742–4750 (2004). 21....
work page Pith review arXiv doi:10.48550/arxiv.2405.01238 2016
-
[32]
High Extinction Ratio Widely Tunable Low-Loss Integrated Si3N4 Third-Order Filter
John, D. D. et al. Multilayer Platform for Ultra-Low-Loss Waveguide Applications. IEEE Photonics Technology Letters 24, 876–878 (2012). 33. Liu, K. et al. Tunable broadband two-point-coupled ultra-high-Q visible and near-infrared photonic integrated resonators. Photon. Res., PRJ 12, 1890–1898 (2024). 34. Heim, D. A. S., Bose, D., Liu, K., Isichenko, A. & ...
work page Pith review arXiv 2012
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.