Pith. sign in

REVIEW 2 major objections 5 minor 24 references

Integrated phononic waveguide on thin-film lithium niobate on diamond

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A lithium-niobate-on-diamond waveguide converts microwaves to sound with >50% efficiency at 4 K

desk verdict Useful platform demonstration, but the headline >50% transducer efficiency is not supported because the measured peak is cavity-enhanced by standing waves between reflective IDTs. read the letter →

arxiv 2505.23100 v1 pith:TFZJ7MG3 submitted 2025-05-29 cond-mat.mes-hall physics.app-ph

Sultan Malik , Felix M. Mayor , Wentao Jiang , Hyunseok Oh , Carl Padgett , Viraj Dharod , Jayameenakshi Venkatraman , Ania C. Bleszynski Jayich
show 1 more author
Amir H. Safavi-Naeini
This is my paper
classification cond-mat.mes-hallphysics.app-ph
keywords phononicwaveguidelithiumniobateondiamondinterdigitaltransducerquasi-Lovemodecryogeniccharacterizationsilicon-vacancycentersspin-phononcouplingtransferprinting
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

The paper reports a new material platform for on-chip acoustic circuits: thin-film lithium niobate transfer-printed onto bulk diamond. The stack combines lithium niobate's strong piezoelectric response with diamond's high acoustic velocity, so a compact interdigital transducer can launch a guided shear-horizontal wave into a wavelength-scale rib waveguide. At 4 K the 100-µm-long delay line passes 26% of the input microwave power at 2.8 GHz, a two-port insertion loss of -5.8 dB that the authors attribute to better-than-50% efficiency per transducer. The intended payoff is a phonon-mediated interface to strain-sensitive color centers in diamond, such as silicon-vacancy spins, for hybrid quantum devices.

What carries the argument

The load-bearing element is the LN-on-diamond material stack together with a co-planar interdigital transducer (IDT). The IDT is oriented on X-cut lithium niobate so that the electric field drives the $d_{24}$ piezoelectric component, couples to the YZ strain, and excites a quasi-Love mode, a guided shear-horizontal wave confined near the surface; the large piezoelectric coupling $k^2_{\rm eff}\sim 21\%$ allows a small transducer (roughly 2 by 22 square micrometers) to match a 50-ohm transmission line. The diamond substrate's high acoustic velocity (above 12 km/s) confines the mechanical mode tightly to the lithium niobate rib, and a linear taper between IDT and the 1-µm-wide waveguide minimizes mode-mismatch loss. Transfer printing patterned lithium niobate from an LNOI source chip onto the diamond provides the integration route.

What would settle it

Fabricate two waveguides of different lengths (for example, the 100-µm and 145-µm devices already made by the authors) on the same chip and measure two-port insertion loss at 4 K; if the loss difference implies a propagation loss per unit length that is not small compared to the assumed 2.9 dB per transducer, or if the two IDTs show markedly asymmetric reflection spectra, the >50% transducer efficiency claim would be an overestimate.

Watch

Extended reading notes

Core claim

The paper's central claim is that a transfer-printed thin-film lithium niobate rib on bulk diamond supports efficient electro-acoustic transduction and guiding at gigahertz frequencies. With a 15-period interdigital transducer whose effective piezoelectric coupling is $k^2_{\rm eff}\sim 21\%$, the authors excite a quasi-Love mode in the 1-µm-wide waveguide and measure a total insertion loss of -5.8 dB at 4 K and 2.8 GHz, which they interpret as better than 50% transduction efficiency per IDT. From the simulated mode profile they estimate a zero-point strain at 100 nm depth in the diamond that yields a spin-phonon coupling rate of about $g_{\rm sp}/2\pi \sim 24$ kHz per phonon for a silicon-vacancy center, and they compute a single-spin cooperativity of about $3.2\times 10^{-2}$ for the straight waveguide, rising to roughly 2.6 if the same platform is used to fabricate a ring resonator with quality factor near $5\times 10^{4}$, in their estimate.

Load-bearing premise

The inference that -5.8 dB of two-port loss implies better-than-50% per-transducer efficiency assumes that acoustic propagation loss along the 100-µm waveguide is negligible and that the two identical transducers share the loss symmetrically; the paper states that low yield of longer structures prevented variable-length loss statistics, so the efficiency number is an upper-bound estimate rather than a directly measured quantity.

Editorial extensions

If this is right

  • Cryogenic phononic delay lines on diamond can reach a two-port insertion loss of -5.8 dB at 2.8 GHz, making efficient electrical-to-acoustic conversion practical in a substrate that can host color-center qubits.
  • The same stack, IDT design, and quasi-Love mode can be reused for ring resonators and more complex phononic circuits, since the guided mode is confined to wavelength scale.
  • If the ring-resonator cooperativity estimate holds, a quality factor around $5\times 10^{4}$ would raise single-spin cooperativity above 1, a regime needed for coherent spin-phonon control.
  • Cooling from room temperature to 4 K increases the transmitted power from about -21 dB to -5.8 dB, showing that acoustic and resistive losses in this platform drop sharply at cryogenic temperatures.

Reading between the lines

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

  • The efficiency figure is an upper bound: the paper lacks a variable-length propagation-loss measurement, so a dedicated two-length comparison would either confirm the 50% number or lower it once waveguide loss is separated from transducer loss.
  • Because the strain field in diamond is evanescent, the same waveguide design should couple to other strain-sensitive defects besides silicon vacancies, such as germanium-vacancy centers, extending the platform to different quantum memory species.
  • Adopting the direct-bonding fabrication route the authors mention would likely remove the yield bottleneck on long waveguides and ring resonators, and would make the variable-length loss measurement straightforward.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper demonstrates a phononic delay line consisting of a transfer-printed thin-film lithium niobate (LN) rib waveguide on bulk diamond, with aluminum interdigital transducers (IDTs). The authors simulate the quasi-Love guided mode near 2.67 GHz, fabricate a 100-μm-long device, and characterize two-port S-parameters at room temperature and at 4 K. At 4 K they observe a peak |S21|^2 of 26% (-5.8 dB) near 2.8 GHz, together with fringes of FSR ~12 MHz which they attribute to standing waves between the two IDTs. They interpret the -5.8 dB two-port loss as corresponding to a per-transducer efficiency >50% (insertion loss <3 dB), and they estimate a spin-phonon coupling rate for SiV centers in diamond. The paper's main quantitative claim is the cryogenic transducer efficiency.

Significance. If the >50% transducer efficiency claim were supported, this would be a significant advance: it would establish an efficient cryogenic phononic waveguide platform on diamond, combining LN's strong piezoelectricity with diamond's high acoustic velocity and color-center compatibility. The transfer-printing fabrication route is a genuine technical contribution, and the measured group velocity agrees well with simulation (2.8(3) km/s versus 2.84 km/s). The paper is clearly written and the FEM design work is substantial. However, the headline efficiency claim is currently not established because the device operates as a Fabry-Perot cavity whose on-resonance transmission can exceed the single-pass product of the two transducer efficiencies. This is a load-bearing issue for the abstract and conclusion. The platform demonstration itself remains valuable, and the manuscript is technically competent.

major comments (2)
  1. [Cryogenic characterization at 4 K (Fig. 4), Abstract, and Conclusion] The conversion of the measured two-port peak |S21|^2 = 26% (-5.8 dB) into a per-transducer efficiency >50% is not justified, because the device is operated as a Fabry-Perot cavity. The text attributes the FSR ~12 MHz fringes to standing waves between the two IDTs; with reflective IDTs, the on-resonance transmission can be much larger than the single-pass product η1·η2·α_L. The measured peak therefore does not bound the single-transducer efficiency from below, and the claims of ">50% transducer efficiency" in the abstract and "transducer with insertion loss less than 3 dB" in the conclusion are unsupported as stated. I note that the reader's concern about unmeasured propagation loss would, under the single-pass model, make the inferred per-transducer efficiency larger rather than smaller; the standing-wave/cavity effect is the decisive issue. The impulse response in Fig. 4(c) should be time-gated to isolate the first-arrival acoustic pulse (at τ_d ≈ 36 ns) and obtain the true single-pass insertion loss, or the full frequency response should be fit with a Fabry-Perot model that includes IDT reflectivity, propagation loss, and transduction efficiency.
  2. [Cryogenic characterization at 4 K (Fig. 4)] The paper states that "the lower acoustic loss at cryogenic temperatures allows these standing waves to be more prominent" but does not quantify the IDT reflectivity or cavity finesse. Without such quantification, the 26% peak value cannot be assigned to transduction rather than resonant enhancement. The filtered response in Fig. 4(b) removes only microwave crosstalk, not the acoustic standing-wave enhancement. The authors should either use the impulse response to separate the first-arrival pulse from echoes (the VNA bandwidth appears sufficient to resolve τ_d) or report a cavity-model extraction of η1, η2, and α_L. The room-temperature data, where standing waves are less prominent, could additionally serve as a cross-check of the single-pass interpretation.
minor comments (5)
  1. [Abstract and Conclusion] The term "transducer efficiency" should be defined explicitly (electrical-to-acoustic conversion efficiency of one IDT), and the statement that the -5.8 dB two-port insertion loss "corresponds to" a per-transducer efficiency should be replaced or qualified with the single-pass assumption once the standing-wave issue is addressed.
  2. [References [15] and [23]] References [15] and [23] appear to be the same paper (Sukachev et al., Phys. Rev. Lett. 119, 223603 (2017)); please consolidate to avoid duplicate citations.
  3. [Equation (1)] The symbol c0 is introduced as "capacitance per unit area" while C0 is the capacitance per unit cell from the unit-cell simulation; the relation between c0 and C0 (e.g., c0 = C0/(a_IDT × w_IDT)) should be stated explicitly.
  4. [Cryogenic characterization at 4 K] The time-domain gate used to filter microwave crosstalk is not specified; please state the gate limits and confirm that the gate does not remove the first acoustic arrival at τ_d ≈ 36 ns.
  5. [Figure 4(c) caption] The impulse-response units "dB Hz" are ambiguous; please use units that clearly indicate a spectral density, such as dB/Hz or dB·Hz^{-1}.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the central result is a direct transmission measurement, and the design and spin-phonon estimates use external formulas, simulations, and literature parameters rather than quantities fitted to the device data.

full rationale

The paper's load-bearing claims are (i) the measured two-port |S21|^2 peak of 26% (−5.8 dB) at 4 K and (ii) the inferred per-transducer efficiency >50% under a symmetric two-transducer, negligible-propagation-loss model. Neither claim reduces to a fitted input or to a self-citation. The transducer design uses standard IDT formulas (Eq. 1) from Refs. [18,19] with assumed G0 and bandwidth; the subsequent full simulation and measured admittance are independent checks, not inputs used to construct the transmission peak. The transmission peak is a direct VNA measurement with the cable response removed and microwave crosstalk filtered through the impulse response. The spin-phonon coupling estimate g_sp/2π = d × ε_zpf uses the literature SiV strain susceptibility d and the simulated waveguide strain field; the cooperativity estimate uses assumed spin and phonon decay rates, all external inputs. The FSR consistency check is a post-hoc comparison, not a fitted prediction. The paper's own limitation statement—that low yield prevented variable-length propagation-loss measurements—and its attribution of the 12 MHz fringes to standing waves between the IDTs is a validity caveat on the efficiency extraction, not a circular derivation; the extraction is an assumption-based inference, not a definitional equivalence. Self-citations to prior work (Refs. [9,18,19,21]) supply formulas, simulation methods, and fabrication procedures that are independently documented and are not used to define the target quantity. No step in the paper equates its conclusion to its input by construction.

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

The central result is an experimental demonstration, so the ledger is mostly about modeling and inference assumptions rather than fitted constants. No parameter was fitted to the reported insertion-loss data to produce the headline number. The main burden is the unmeasured propagation loss that supports the per-transducer efficiency inference, plus the assumption that the transfer-printed interface and the simulated strain field behave as modeled.

assumptions (5)
  • domain assumption Finite-element simulations of the waveguide cross section and IDT unit cell correctly predict the operational frequency, mode shape, and coupling strength.
    Design section uses FEM dispersion and unit-cell simulations to choose the quasi-Love mode and IDT period; the measured center frequency is 2.79 GHz versus 2.67 GHz simulated, a 4% mismatch, so the assumption is mostly but not exactly validated.
  • domain assumption Propagation loss in the 100-um waveguide is negligible compared with transducer conversion loss when interpreting the two-port measurement.
    The >50% transducer efficiency claim relies on this; no variable-length loss measurement is reported, and the authors state yield of longer waveguides was low.
  • domain assumption The transfer-printed LN-diamond bond is acoustically low-loss and does not introduce significant scattering.
    Fabrication section describes transfer printing and annealing, but no bond quality or interface loss measurement is provided.
  • domain assumption An SiV color center at about 100 nm depth experiences the strain field calculated for the bare waveguide, and can be incorporated without degrading the device.
    The spin-phonon rate estimate in the last section uses this; no color-center experiment is reported.
  • standard math The IDT equivalent-circuit formula from prior work [18,19] applies to the LNOD stack.
    Used to size the IDT; the measured admittance (G0=9.2 mS, gamma/2pi=12.5 MHz) is in reasonable agreement with the model, so this is a standard assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Integrated phononic waveguide on thin-film lithium niobate on diamond." pith.science (2026). https://pith.science/paper/TFZJ7MG3

@misc{pith2026250523100,
  author       = {Pith},
  title        = {Pith review of: Integrated phononic waveguide on thin-film lithium niobate on diamond},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TFZJ7MG3}},
  note         = {Machine review of arXiv:2505.23100}
}
read the original abstract

We demonstrate wavelength-scale phononic waveguides formed by transfer-printed thin-film lithium niobate (LN) on bulk diamond (LNOD), a material stack that combines the strong piezoelectricity of LN with the high acoustic velocity and color-center compatibility of diamond. We characterize a delay line based on a 100 micron long phononic waveguide at room and cryogenic temperatures. The total insertion loss through the device at 4 kelvin is -5.8 dB, corresponding to a >50% transducer efficiency, at a frequency of 2.8 gigahertz. Our work represents a step towards phonon-mediated hybrid quantum systems consisting of strain-sensitive color centers in diamond.

Figures

Figures reproduced from arXiv: 2505.23100 by the authors.

Figure 1
Figure 1. (e). We identify the strongly-coupled mode at 2.67GHz by noting its large frequency shift (highlighted as thicker lines) with the changing electric circuit boundary condition from open to closed. From this shift, we estimate [18] the piezoelectric coupling to be k 2 eff ∼ 21%. Solving the model to obtain the admittance at low frequencies, we obtain the dielec￾tric capacitanceC0 = 4.3fF per unit cell. This allows us … view at source ↗
Figure 2
Figure 2. FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

24 extracted references · 23 canonical work pages

  1. [1]

    R. Lu, Y . Yang, M.-H. Li, M. Breen, and S. Gong, IEEE Trans- actions on Microwave Theory and Techniques68, 573 (2019)

    work page 2019

  2. [2]

    Zhang, R

    S. Zhang, R. Lu, H. Zhou, S. Link, Y . Yang, Z. Li, K. Huang, X. Ou, and S. Gong, IEEE Transactions on Microwave Theory and Techniques 68, 3653 (2020)

    work page 2020

  3. [3]

    Z. Xu, S. W. Ding, E. Cornell, S. Mohideen, M. Yeh, K. Ku- ruma, L. Magalhaes, A. Shams-Ansari, B. Pingault, and M. Loncar, arXiv preprint arXiv:2505.08895 (2025)

    arXiv 2025

  4. [4]

    L. Shao, S. Maity, L. Zheng, L. Wu, A. Shams-Ansari, Y .-I. Sohn, E. Puma, M. Gadalla, M. Zhang, C. Wang,et al., Physical Review Applied 12, 014022 (2019)

    work page 2019

  5. [5]

    F. Z. Bi and B. P. Barber, IEEE microwave magazine 9, 65 (2008)

    work page 2008

  6. [6]

    Y . Chu, P. Kharel, W. H. Renninger, L. D. Burkhart, L. Frunzio, P. T. Rakich, and R. J. Schoelkopf, Science 358, 199 (2017)

    work page 2017

  7. [7]

    Thomson, A

    D. Thomson, A. Zilkie, J. E. Bowers, T. Komljenovic, G. T. Reed, L. Vivien, D. Marris-Morini, E. Cassan, L. Virot, J.-M. Fédéli, et al., Journal of Optics 18, 073003 (2016)

    work page 2016

  8. [8]

    W. Fu, Z. Shen, Y . Xu, C.-L. Zou, R. Cheng, X. Han, and H. X. Tang, Nature Communications 10, 2743 (2019)

    work page 2019

Show all 24 references
  1. [9]

    F. M. Mayor, W. Jiang, C. J. Sarabalis, T. P. McKenna, J. D. Witmer, and A. H. Safavi-Naeini, Physical Review Applied15, 014039 (2021)

    2021

  2. [10]

    S. W. Ding, B. Pingault, L. Shao, N. Sinclair, B. Machielse, C. Chia, S. Maity, and M. Lon ˇcar, Physical Review Applied 21, 014034 (2024)

    2024

  3. [11]

    Y . Deng, D. Anderson, X. Du, W. Roberts, M. Miller, B. Smith, L. Hackett, T. Olsson, and M. Eichenfield, arXiv preprint arXiv:2503.18113 (2025)

    2025 arXiv

  4. [12]

    Atatüre, D

    M. Atatüre, D. Englund, N. Vamivakas, S.-Y . Lee, and J. Wrachtrup, Nature Reviews Materials 3, 38 (2018)

    2018

  5. [13]

    Meesala, Y .-I

    S. Meesala, Y .-I. Sohn, B. Pingault, L. Shao, H. A. Atikian, J. Holzgrafe, M. Gündo˘gan, C. Stavrakas, A. Sipahigil, C. Chia, et al., Physical Review B 97, 205444 (2018)

    2018

  6. [14]

    Maity, L

    S. Maity, L. Shao, S. Bogdanovi ´c, S. Meesala, Y .-I. Sohn, N. Sinclair, B. Pingault, M. Chalupnik, C. Chia, L. Zheng,et al., Nature communications 11, 193 (2020)

    2020

  7. [15]

    D. D. Sukachev, A. Sipahigil, C. T. Nguyen, M. K. Bhaskar, R. E. Evans, F. Jelezko, and M. D. Lukin, Physical review let- ters 119, 223602 (2017)

    2017

  8. [16]

    Lemonde, S

    M.-A. Lemonde, S. Meesala, A. Sipahigil, M. Schuetz, M. Lukin, M. Loncar, and P. Rabl, Physical review letters 120, 213603 (2018). 6

    2018

  9. [17]

    Yamamoto, H

    M. Yamamoto, H. Kurokawa, S. Fujii, T. Makino, H. Kato, and H. Kosaka, Journal of Applied Physics 134 (2023)

    2023

  10. [18]

    Y . D. Dahmani, C. J. Sarabalis, W. Jiang, F. M. Mayor, and A. H. Safavi-Naeini, Physical Review Applied 13, 024069 (2020)

    2020

  11. [19]

    C. J. Sarabalis, Y . D. Dahmani, A. Y . Cleland, and A. H. Safavi- Naeini, Journal of Applied Physics 127, 054501 (2020)

    2020

  12. [20]

    M. A. Meitl, Z.-T. Zhu, V . Kumar, K. J. Lee, X. Feng, Y . Y . Huang, I. Adesida, R. G. Nuzzo, and J. A. Rogers, Nature Ma- terials 5, 33 (2006)

    2006

  13. [21]

    Jiang, F

    W. Jiang, F. M. Mayor, S. Malik, R. Van Laer, T. P. McKenna, R. N. Patel, J. D. Witmer, and A. H. Safavi-Naeini, Nature Physics 19, 1423 (2023)

    2023

  14. [22]

    The impulse response for a different data set at 4 K is shown in Fig

    From an inverse Fourier transform of S21, we obtain the impulse re- sponse h21 and use it to filter out the microwave crosstalk. The impulse response for a different data set at 4 K is shown in Fig. 4(c). This microwave crosstalk happens on a much faster time scale than the de...

  15. [23]

    C. Hepp, T. Müller, V . Waselowski, J. N. Becker, B. Pin- gault, H. Sternschulte, D. Steinmüller-Nethl, A. Gali, J. R. Maze, M. Atatüre, and C. Becher, Physical Review Letters112, 036405 (2014)

    2014

  16. [24]

    Sukachev, A

    D. Sukachev, A. Sipahigil, C. Nguyen, M. Bhaskar, R. Evans, F. Jelezko, and M. Lukin, Physical Review Letters119, 223602 (2017)

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.