Large mode volume integrated Brillouin lasers for scalable ultra-Low linewidth and high power

Daniel J. Blumenthal; Kaikai Liu; Karl D. Nelson; Ryan O. Behunin

REVIEW 3 major objections 5 minor 6 references

Large mode volume integrated Brillouin lasers for scalable ultra-Low linewidth and high power

T0 review · 3 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A 4-meter coil of silicon nitride waveguide on a chip lases via Brillouin scattering at a 31 mHz fundamental linewidth, 41 mW output power, and 73 dB mode suppression.

desk verdict The 31 mHz linewidth record is likely an upper bound set by the photodetector noise floor; the device work is solid, but the headline claim needs re-examination. read the letter →

arxiv 2502.01595 v2 pith:MVDAPZYC submitted 2025-02-03 physics.optics

Kaikai Liu , Karl D. Nelson , Ryan O. Behunin , Daniel J. Blumenthal This is my paper · ORCID

classification physics.optics

keywords Brillouinlasersiliconnitrideintegratedphotonicsultra-lowlinewidthhigh-powercoilresonatorstimulatedscatteringfrequencynoise

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a silicon-nitride chip with a 4-meter coiled waveguide can run as a stimulated Brillouin laser that is simultaneously far narrower and more powerful than prior integrated lasers. The demonstrated device puts out a single optical mode with a 31 mHz instantaneous fundamental linewidth, 41 mW on-chip power, and 73 dB suppression of all other modes, and it can be Vernier-tuned across 22.5 nm. The authors argue that the large cavity mode volume raises the power ceiling and lowers thermorefractive noise, while the Brillouin photon–phonon feedback itself picks one mode and starves all the others. If the scaling law they state holds, longer coils would push the fundamental linewidth toward 1 mHz and the output power past 1 W. Such a laser would matter because compact, portable, ultra-pure light sources are a bottleneck for atomic clocks, quantum sensing, and low-noise microwave generation.

What carries the argument

The load-bearing object is a 4-meter-long, ultra-low-loss (~0.16-0.2 dB/m) Si3N4 coil resonator coiled into a chip under a square centimeter, with 160 million intrinsic Q and a 48.1 MHz free spectral range. The mechanism that carries the argument is the stimulated Brillouin photon–phonon scattering process inside that cavity: below threshold the pump feeds several phonon gratings at different cavity resonances, and above threshold the mode with the largest Brillouin gain clamps the pump and suppresses all rival modes. A coupled-mode model, in which each Stokes optical mode couples to a single phonon, shows that this clamping leaves the other Stokes amplitudes at zero, which explains the high SMSR in a deliberately multimode cavity.

What would settle it

Look for additional Stokes modes or asymmetric beatnote sidebands as the pump power is raised toward the second-Stokes threshold, and repeat the measurement at temperatures or coil lengths where the 48.1 MHz FSR no longer clearly exceeds the ~30 MHz mechanical linewidth; appearance of secondary lasing modes would contradict the single-phonon-per-mode clamping picture.

Watch

Extended reading notes

Core claim

On its own terms, the paper reports the first integrated Brillouin laser whose cavity is a meter-scale, multi-FSR coil resonator rather than a small single-FSR ring, and shows that this geometry improves rather than degrades single-mode purity. Measured at the first-Stokes clamping point, the S1 laser has a 31 mHz fundamental linewidth (10 $mHz^{2}$/Hz white frequency noise at high offsets), 41 mW on-chip output at 242 mW pump (16.5% efficiency), and 73 dB sidemode suppression, with the frequency noise reaching the resonator thermorefractive floor between 10 kHz and 100 kHz. The central physical claim is that the 'winner take all' Brillouin dynamics—pump clamping once the highest-gain mode reaches threshold—prevents the other four cavity resonances inside the ~250 MHz Brillouin gain bandwidth from lasing, so single-mode operation is compatible with a large mode volume.

Load-bearing premise

The argument assumes that each optical cavity mode interacts with only one phonon mode, which requires the 48.1 MHz free spectral range to stay larger than the roughly 30 MHz mechanical linewidth; that margin is small, and if multiple phonon couplings or extra mode interactions appear, the winner-take-all single-mode explanation and the 73 dB suppression ratio could fail.

Editorial extensions

If this is right

If the scaling relation holds, lengthening the coil from 4 m to about 120 m would push the fundamental linewidth below 1 mHz and the first-Stokes output power to about 1.2 W.
The laser's frequency noise reaches the resonator's thermorefractive-noise floor between 10 kHz and 100 kHz, so the coil geometry directly lowers the noise floor that other integrated lasers hit.
Because Brillouin lasing here does not require acoustic guiding, the same coil resonator platform can be ported to visible and near-IR wavelengths and to other bands via two-point coupling.
The 22.5 nm Vernier tuning range, on a 48.1 MHz FSR grid with continuous thermal tuning in between, makes the laser usable across the C and L telecom bands.
A single 72 mW threshold and 41 mW output at 16.5% efficiency means the laser is practical as a pump-driven on-chip source rather than a laboratory curiosity.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the single-phonon assumption holds generally, then multi-FSR Brillouin cavities—long seen as unsuitable for single-mode lasing—should be able to combine large mode volume with high SMSR, freeing cavity designers to optimize for loss and power rather than for one-mode-per-gain-bandwidth.
The same winner-take-all mechanism suggests that cascaded Stokes emission might be suppressed by increasing the spacing between gain-selected modes, which could allow higher pump power before second-Stokes onset.
One testable extension is to measure the beatnote while sweeping cavity temperature across the point where the 48 MHz FSR approaches the ~30 MHz mechanical linewidth; the model predicts a sharp onset of multi-mode or multi-phonon behavior there.
A second test is at visible wavelengths, where Brillouin gain, FSR, and mechanical linewidth all scale differently; the coupled-mode model's prediction of single-mode lasing should be re-checked there.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The interesting thing here is the cavity: a 4-meter coil resonator used as the active SBS laser cavity rather than a cold reference. That is a real departure from prior SBS laser work, and the 41 mW output with 73 dB SMSR at reasonable pump power is a genuine advance. The tuning mechanism via the Vernier effect is also neat, and the paper is clearly written with an honest accounting of what was measured versus what was modeled. The coupled-mode model in Supplementary Note 2 is a reasonable first-principles explanation for the single-mode behavior, and the 73 dB SMSR is an independent measurement, so that part holds up.

The problem is the headline number. The text states that above 10 MHz the OFD frequency noise measurement is limited by photodetector noise, and the 10 mHz^2/Hz floor used to compute 31 mHz is taken at high frequency offsets. That is precisely where the detector floor would appear. The paper shows a light-blue dashed curve for the detector floor, but it never shows the measured laser noise curve against that floor in equivalent frequency noise units at the analysis offset. If the red curve sits on the floor, then 31 mHz is not the laser linewidth; it is the measurement floor. The authors themselves estimate the fundamental linewidth should be 13 mHz from scaling their previous 0.7 Hz result, and they attribute the excess to pump-transferred noise. That admission undercuts the claim that 31 mHz is the laser's fundamental linewidth. At best it is an upper bound—which would still be notable, but not "the lowest linewidth to date."

There is also a units issue: "10 mHz^2/Hz" is ambiguous. Read literally, it means (mHz)^2 per Hz, which would give 31 µHz, not 31 mHz. The authors clearly intend "milli-Hz^2/Hz" (10^-3 Hz^2/Hz), but the notation should be fixed. The single-phonon coupling assumption in the model also relies on the FSR exceeding the mechanical linewidth by only about 1.6x; the stress-test is right that this is a secondary concern because the SMSR measurement is independent. Data availability is limited to "upon reasonable request," which is disappointing for a record claim.

Recommendation: this deserves peer review, not a desk rejection. The engineering is substantial, and the power/SMSR/tuning results are solid. But the linewidth claim needs clarification before publication. The authors should plot the detector floor in the same units as the measured noise spectrum and explicitly state whether the 10 mHz^2/Hz floor lies above or coincides with the detector floor. If it coincides, they should rephrase the claim as an upper bound and adjust the abstract. I would send it to a serious referee with the expectation of major revision on the linewidth analysis.

Referee Report

3 major / 5 minor

Summary. The authors report an integrated silicon-nitride coil-resonator Brillouin laser with a 4-meter cavity, 160 million intrinsic Q, a 72 mW S1 threshold, 41 mW on-chip output power at 242 mW pump power, 73 dB sidemode suppression, and Vernier tuning across a 22.5 nm range. They claim a 31 mHz instantaneous linewidth derived from a white frequency-noise floor of 10 mHz^2/Hz, and they argue that increasing the coil length can scale the fundamental linewidth to 1 mHz and output power above 1 W. A coupled-mode model in Supplementary Note 2 is used to explain single-mode lasing through pump clamping in a multimode Brillouin gain bandwidth.

Significance. If the 31 mHz linewidth, 41 mW output power, and 73 dB SMSR are verified, this would be a record combination for an integrated laser and a meaningful advance for portable precision metrology, quantum sensing, and low-noise microwave generation. The paper has clear strengths: a large-mode-volume design that simultaneously addresses linewidth and power, direct spectral evidence of high sidemode suppression, a practical Vernier tuning scheme, and a first-principles coupled-mode framework for single-mode operation. The central linewidth claim, however, rests on the frequency-noise floor and its conversion, and the current presentation leaves a load-bearing ambiguity that must be resolved.

major comments (3)

[Fig. 3 and 'Coil SBS laser characterization'] The numerical relation between the quoted white-noise floor and the quoted linewidth is inconsistent as printed. The text reports '10 mHz2/Hz frequency noise' and a 'fundamental linewidth of 31 mHz' using the formula Δν_F = π S_w. Read literally, 10 mHz^2/Hz equals 10^-5 Hz, so π S_w would give 31 µHz, not 31 mHz. If the intended floor is 10 × 10^-3 Hz^2/Hz = 0.01 Hz^2/Hz, this must be stated explicitly and the linewidth recomputed from the corrected value. Because the 'lowest linewidth to date' claim depends on this conversion, this is not a purely cosmetic units issue.
[Fig. 3 and Materials and methods] The paper states that above 10 MHz the OFD frequency-noise measurement is limited by the photodetector noise, and the red curve used for the 31 mHz value is described as measured at high frequency offsets. The detector noise floor is not shown in equivalent frequency-noise units at the analysis offset, nor is the exact extraction offset for S_w specified. Please overlay the calibrated detector-noise floor on the laser frequency-noise spectrum, state the offset at which S_w is read, and show that the quoted flat segment lies above the floor by a clear margin. Without this, the 31 mHz number could be an upper bound set by the measurement system rather than the laser linewidth.
[Supplementary Note 2] The winner-take-all clamping argument assumes each Stokes optical mode couples to a single phonon, justified by the resonator FSR (48.1 MHz) exceeding the mechanical linewidth Γ/2π (~30 MHz). The margin is only a factor of about 1.6, and multi-phonon coupling or acoustic-mode overlap could modify the clamping condition and the resulting single-mode prediction. The 73 dB SMSR is independently measured, so this does not threaten the experimental single-mode claim, but the mechanistic explanation and the extrapolation to longer coils would be strengthened by a quantitative assessment of the single-phonon assumption's validity.

minor comments (5)

[Abstract] The abstract contains typos: 'Moreso' should be 'More so' or 'Moreover', and 'an large mode volume' should be 'a large mode volume'.
[Fig. 3 caption and main text] The detector-noise floor is referred to as 'light blue dashed curve' in the main text and 'light green dashed curve' in the figure caption; the color references should be made consistent.
[Methods and main text] The fiber MZI FSR is given as 1.03 MHz in the Methods section and 1.037 MHz elsewhere; please use a single consistent value and ensure the calibration is traceable to the measured FSR.
[Discussion] The claim of 'greater than 5 orders magnitude frequency noise reduction from the free running pump laser over a wide bandwidth' is not supported by a displayed measurement of the free-running pump frequency noise; either add the pump noise trace or qualify the statement.
[Supplementary Note 1, Fig. S5] The Brillouin gain rate μ is inferred from the measured 72 mW threshold rather than measured independently; please clarify this status and give the estimated uncertainty in μ, since the threshold projections use this value.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline 31 mHz linewidth is a measured frequency-noise value, not a fitted or self-referential output; the coupled-mode derivation and the scaling estimate are self-contained or use prior measurements as independent inputs.

full rationale

The central claim rests on a direct OFD frequency-noise measurement (Fig. 3), with the linewidth computed by the standard relation Δν_F = πS_w; no parameter is fitted to make this number emerge. The coupled-mode model in Supplementary Note 2 is derived in the paper from Heisenberg-Langevin equations; the only assumption, that each Stokes optical mode couples to a single phonon, is stated explicitly and is not equivalent to the single-mode conclusion it is used to explain. The scaling estimate of 13 mHz uses a previously measured 0.7 Hz baseline from Refs 16/33 scaled by a 54× cavity-length increase; although those references include the present authors, the baseline is an independent measured quantity, and the estimate is explicitly labeled as lower than the measured 31 mHz and is not used to support the record claim. The Brillouin gain rate μ = 0.174 mHz is inferred from the measured 72 mW threshold and is used only for secondary wavelength-dependent threshold estimates, not for the headline linewidth. The admitted photodetector-noise floor above 10 MHz is a measurement-validity concern, not a circular-derivation concern. Therefore no load-bearing step reduces to its own input.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claims rest on standard SBS laser theory (Refs. 16, 22, 23) and the coupled-mode model in Supplementary Note 2. No new physical entities are introduced. The only fitted parameter is the Brillouin gain rate mu, which is used for secondary wavelength-dependent threshold estimates rather than the headline linewidth.

free parameters (1)

Brillouin gain rate per photon mu = 0.174 mHz
Inferred from the measured S1 threshold of 72 mW at 1570 nm (Supplementary Fig. S5). Used to estimate thresholds at other wavelengths, but not used to fit the headline linewidth.

assumptions (5)

domain assumption Each Stokes optical mode couples to a single phonon mode (valid when FSR > Gamma/(2*pi)).
Stated in Supplementary Note 2; FSR is 48.1 MHz and Gamma/(2*pi) is about 30 MHz, so the margin is small.
domain assumption The mode with the largest Brillouin gain reaches threshold first and clamps the pump, preventing other modes from lasing.
Derived from the coupled-mode model in Supplementary Note 2; underpins the 73 dB SMSR claim.
domain assumption The thermal occupation number of the optical modes is negligible at room temperature.
Stated in Table S1 footnote to simplify the linewidth formula.
domain assumption The SBS linewidth enhancement factor alpha is zero.
Implied in Table S1; if non-zero, linewidths would be multiplied by (1+alpha^2).
domain assumption Fundamental linewidth scales inversely with cavity length and S1 output power scales linearly with length, from prior theory.
Used to estimate 13 mHz for 4 m from 0.7 Hz at 74.3 mm, and to project 1 mHz at 120 m.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Large mode volume integrated Brillouin lasers for scalable ultra-Low linewidth and high power." pith.science (2026). https://pith.science/paper/MVDAPZYC

@misc{pith2026250201595,
  author       = {Pith},
  title        = {Pith review of: Large mode volume integrated Brillouin lasers for scalable ultra-Low linewidth and high power},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MVDAPZYC}},
  note         = {Machine review of arXiv:2502.01595}
}

read the original abstract

Generating ultra-low linewidths and high output power in an integrated single mode laser remains a critical challenge for future compact, portable, precision applications. Moreso, achieving these characteristics in a laser design that enables scaling to lower linewidths and higher power, and implementation in a wafer-scale integration platform that can operate from visible to near-IR and be integrated with other components. Such an advance could impact a wide array of applications including atomic and quantum sensing and computing, metrology, coherent fiber communications and sensing, and ultra-low-noise mmWave and RF generation. Yet, achieving these goals in an integrated laser has remained elusive. Here, we report a class of integrated laser that overcomes these limitations, with demonstration of a 31 mHz instantaneous linewidth, the lowest linewidth to date to the best of our knowledge, with 41 mW output power and 73 dB sidemode suppression ratio and can be tuned across a 22.5 nm range. This performance is possible due to Brillouin nonlinear dynamics that occurs within a large mode volume, nonlinear photon-phonon, MHz-scale-FSR, ultra-low loss silicon nitride resonator cavity. This laser design can scale to a new operating regime of mHz fundamental linewidth and Watt class lasers. Such lasers hold promise to unlock new sensitivity and fidelity for precision quantum experiments, portable precision applications, and atomic, molecular, and optical physics.

Figures

Figures reproduced from arXiv: 2502.01595 by the authors.

**Figure 1.** Large mode volume SBS laser design and operation principles. a 6 𝜇m by 80 nm Si3N4 waveguide design supports the fundamental TE mode with a minimum bending radius of ~1 mm and propagation loss of ~0.2 dB/m in C and L band. b Large mode volume SBS laser implemented in a 4-meter-coil resonator. c Coil resonator with 50 MHz FSR overlaps with the broad spontaneous acoustic mode continuum of the Brillouin gain bandwidth … view at source ↗

**Figure 5.** Scaling the integrated coil Brillouin laser to 1 mHz fundamental linewidth and greater than 1 Watt output power . The green line indicates what is possible with the coil SBS lasers, and by increasing coil lengths the SBS laser output power at the S1 clamping point increases linearly with the coil lengths while the fundamental linewidth decreases linearly with the coil length, assuming that the coil resonator Qs stay… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

[1]

O., Otterstrom, N

Behunin, R. O., Otterstrom, N. T., Rakich, P. T., Gundavarapu, S. & Blumenthal, D. J. Fundamental noise dynamics in cascaded-order Brillouin lasers. Phys. Rev. A 98, 023832 (2018)

work page 2018
[2]

& Vahala, K

Yuan, Z., Wang, H., Wu, L., Gao, M. & Vahala, K. Linewidth enhancement factor in a microcavity Brillouin laser. Optica 7, 1150–1153 (2020)

work page 2020
[3]

& Vahala, K

Li, J., Lee, H., Chen, T. & Vahala, K. J. Characterization of a high coherence, Brillouin microcavity laser on silicon. Opt. Express 20, 20170 (2012)

work page 2012
[4]

Liu, K. et al. Integrated photonic molecule Brillouin laser with a high-power sub-100-mHz fundamental linewidth. Opt. Lett. 49, 45–48 (2024)

work page 2024
[5]

Puckett, M. W. et al. 422 Million intrinsic quality factor planar integrated all-waveguide resonator with sub-MHz linewidth. Nat. Commun. 12, 934 (2021)

work page 2021
[6]

T., Bauters, J

Spencer, D. T., Bauters, J. F., Heck, M. J. R. & Bowers, J. E. Integrated waveguide coupled Si_3N_4 resonators in the ultrahigh-Q regime. Optica 1, 153 (2014)

work page 2014

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

O., Otterstrom, N

Behunin, R. O., Otterstrom, N. T., Rakich, P. T., Gundavarapu, S. & Blumenthal, D. J. Fundamental noise dynamics in cascaded-order Brillouin lasers. Phys. Rev. A 98, 023832 (2018)

work page 2018

[2] [2]

& Vahala, K

Yuan, Z., Wang, H., Wu, L., Gao, M. & Vahala, K. Linewidth enhancement factor in a microcavity Brillouin laser. Optica 7, 1150–1153 (2020)

work page 2020

[3] [3]

& Vahala, K

Li, J., Lee, H., Chen, T. & Vahala, K. J. Characterization of a high coherence, Brillouin microcavity laser on silicon. Opt. Express 20, 20170 (2012)

work page 2012

[4] [4]

Liu, K. et al. Integrated photonic molecule Brillouin laser with a high-power sub-100-mHz fundamental linewidth. Opt. Lett. 49, 45–48 (2024)

work page 2024

[5] [5]

Puckett, M. W. et al. 422 Million intrinsic quality factor planar integrated all-waveguide resonator with sub-MHz linewidth. Nat. Commun. 12, 934 (2021)

work page 2021

[6] [6]

T., Bauters, J

Spencer, D. T., Bauters, J. F., Heck, M. J. R. & Bowers, J. E. Integrated waveguide coupled Si_3N_4 resonators in the ultrahigh-Q regime. Optica 1, 153 (2014)

work page 2014