Pith. sign in

REVIEW 3 major objections 6 minor 31 references

YIG/CoFeB bilayer magnonic diode

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A YIG/SiO2/CoFeB bilayer works as a spin-wave diode, transmitting magnetostatic surface waves over 10 micrometers in one direction while quenching them within about 3.8 micrometers in the other.

desk verdict A plausible and useful YIG-based dipolar magnonic diode, but the quantitative 3:1 decay-length claim needs error bars and a direct check of the operating wavevector before I'd trust it. read the letter →

arxiv 2412.08383 v1 pith:KSH44AI4 submitted 2024-12-11 physics.app-ph cond-mat.mes-hall

Noura Zenbaa , Khrystyna O. Levchenko , Jaganandha Panda , Kristýna Davídková , Moritz Ruhwedel , Sebastian Knauer , Morris Lindner , Carsten Dubs
show 4 more authors
Qi Wang Michal Urbánek Philipp Pirro Andrii V. Chumak
This is my paper · ORCID
classification physics.app-phcond-mat.mes-hall
keywords magnonicdiodespin-wavenon-reciprocityYIG/CoFeBbilayermagnetostaticsurfacespinwavesBrillouinlightscatteringdipolarcouplingmicromagneticsimulationmagnonics
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 reports a working magnonic diode built from a YIG/SiO2/CoFeB trilayer. The bilayer supports magnetostatic surface spin waves that travel much farther in one direction than the other: at ±200 mT the decay length is 10.92 micrometers versus 3.78 micrometers, roughly a factor of three. The non-reciprocity comes from dipolar coupling between the low-damping YIG layer and the high-anisotropy CoFeB layer, which breaks left-right symmetry of the dynamic stray-field energy. The authors confirm the effect with wavevector-resolved and micro-focused Brillouin light scattering and micromagnetic simulations, arguing that backscattered waves are strongly suppressed. If correct, this gives a practical route to directional, energy-efficient wave-based signal routing.

What carries the argument

The load-bearing element is the magnetically heterogeneous bilayer YIG(100 nm)/SiO2(5 nm)/CoFeB(40 nm), where the non-magnetic SiO2 spacer preserves dipolar coupling while keeping YIG damping near $10^{-4}$. CoFeB provides strong saturation magnetization and anisotropy, creating the dynamic stray-field interaction that breaks reciprocity; the energy density $\epsilon_d = -\frac{\mu_0}{2}\mathbf{m}\cdot\mathbf{h}_{\mathrm{stray}}$ is minimized when the dynamic magnetization and stray field are parallel. The central diagnostic is the decay-length asymmetry extracted from micro-focused BLS line scans fit to $I = I_0\exp(-2x/\lambda)$, backed by k-resolved BLS measurements of the non-reciprocal dispersion and by micromagnetic simulations that map the simulated color plot onto the measured points.

What would settle it

Measure the spin-wave decay length at several excitation frequencies around 7.23 GHz on a bare-YIG control sample with the identical antenna and field polarity; if similar asymmetry appears, the diode effect is not caused by the CoFeB bilayer. Additionally, wavevector-resolved BLS at the excitation frequency would reveal whether the excited population actually sits near $k = \pm10\,\mathrm{rad/\mu m}$, where the simulated dispersion predicts the strong non-reciprocity.

Watch

Extended reading notes

Core claim

The central claim is that a YIG(100 nm)/SiO2(5 nm)/CoFeB(40 nm) bilayer acts as a magnonic diode for magnetostatic surface spin waves in the Damon-Eshbach geometry. Because the two magnetic layers have different saturation magnetization and dynamic phase, the dipolar stray field of one layer acts on the other, making the interaction energy density $\epsilon_d = -\frac{\mu_0}{2}\mathbf{m}\cdot\mathbf{h}_{\mathrm{stray}}$ depend on propagation direction; the dispersion develops a flat plateau only for one sign of wavevector near $k = -10\,\mathrm{rad/\mu m}$. Experimentally, at an excitation frequency of 7.23 GHz and bias field $\pm200$ mT, the decay length is $\lambda_{+200\,\mathrm{mT}} = 10.92\,\mu\mathrm{m}$ versus $\lambda_{-200\,\mathrm{mT}} = 3.78\,\mu\mathrm{m}$, so forward-propagating waves survive over three times the distance of backward waves. The paper argues this asymmetry is the signature of unidirectional MSSW propagation and constitutes a functional magnonic diode.

Load-bearing premise

The measured decay-length asymmetry is credited to the bilayer's intrinsic non-reciprocity, but the measurements do not independently confirm that the excitation at 7.23 GHz couples to the wavevector band near $\pm10\,\mathrm{rad/\mu m}$ where the simulations predict the diode effect.

Editorial extensions

If this is right

  • The same structure can serve as an isolator, circulator, or phase shifter in magnonic circuits, since non-reciprocity is built into the material stack rather than requiring external biasing asymmetry beyond field sign.
  • Adjusting the CoFeB thickness tunes the flat dispersion plateau, allowing designers to place the diode band at a chosen wavevector and frequency.
  • Because YIG damping stays near $10^{-4}$, directional spin waves can be routed over 10 micrometers or more, long enough for on-chip interferometry or logic.
  • The diode suppresses backscattered waves, which should reduce spurious reflections in magnonic networks.
  • The bilayer's effect was confirmed at a single excitation frequency (7.23 GHz); further measurements could show whether the diode action persists across a usable bandwidth.

Reading between the lines

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

  • A direct test of the proposed mechanism would be to swap the layer order (CoFeB on the bottom versus on top), which should flip the preferred propagation direction if the stray-field asymmetry is the cause; the paper does not report this control.
  • The dipolar-coupling recipe could be extended to other high-magnetization ferromagnets or Heusler alloys, potentially moving the diode band to higher frequencies; this is speculative and not tested here.
  • The strong field-polarity dependence of the decay length suggests the structure could double as a sensitive magnetic-field sensor, although the paper does not pursue that use.
  • If the asymmetry persists over a wider frequency range than the single measured point, the bilayer could act as a broadband directional coupler for magnonic signal processing; this is an untested implication.
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

3 major / 6 minor

Summary. The manuscript reports a magnonic diode based on a YIG(100 nm)/SiO2(5 nm)/CoFeB(40 nm) bilayer, with non-reciprocity arising from dipolar coupling between the two magnetic layers. The authors use MuMax3 simulations to predict a non-reciprocal MSSW dispersion with a unidirectional window near k = ±10 rad/µm, wavevector-resolved BLS of thermal spin waves to confirm the dispersion, and micro-focused BLS line scans of coherent spin waves excited by a 7.23 GHz microstrip antenna to measure propagation. The central experimental result is an asymmetry in the intensity decay length: λ+200mT = 10.92 µm versus λ−200mT = 3.78 µm, which the authors interpret as unidirectional propagation and suppression of backward waves.

Significance. If the asymmetry is intrinsic to the bilayer, this is a valuable experimental demonstration of a dipolar-coupled YIG/CoFeB magnonic diode with a long propagation length in the forward direction. The work combines direct thermal BLS, coherent μ-BLS, and simulations that use literature material parameters rather than being fitted to the diode ratio, so the core non-reciprocity claim is not circular. However, the quantitative figure of merit—the 3:1 decay-length ratio—is not yet tied to the predicted unidirectional wavevector window, lacks uncertainty quantification, and the 'suppression of backward waves' statement goes beyond the presented data. These gaps currently limit the strength of the central claim.

major comments (3)
  1. [Section III, Fig. 3(a) and fitting paragraph] The central quantitative claim is the decay-length asymmetry λ+200mT = 10.92 µm versus λ−200mT = 3.78 µm, but no error bars or confidence intervals are reported for the fitted values or for the data points in Fig. 3(a). In addition, the fit excludes the first three points and, for the −200 mT scan, all points beyond 10 µm, with the justifications that these points are influenced by the antenna near field or have reached the thermal level. Because the 3:1 ratio is the load-bearing result and the exclusion choices directly affect it, the authors should report fit uncertainties and a sensitivity analysis (e.g., varying the fit window, including or excluding the near-antenna points, and subtracting a thermal background) to demonstrate that the ratio is robust rather than an artifact of the chosen fit range.
  2. [Section II-B and Section III] The measured coherent spin waves at 7.23 GHz are not connected to the simulated unidirectional wavevector region near k = ±10 rad/µm. The k-resolved BLS data in Fig. 1(b) are thermal spectra; they show that the dispersion is non-reciprocal, but they do not show which wavevectors the antenna excites coherently. The μ-BLS objective integrates over wavevectors up to 12 rad/µm, so a single-exponential fit to the line scan cannot exclude antenna near-field contributions, multimode propagation, or a CoFeB-localized higher-damping mode. The authors should either measure the k-spectrum of the coherently excited waves at 7.23 GHz, mark the operating point on the simulated dispersion, or otherwise justify that the chosen frequency populates the predicted unidirectional modes.
  3. [Abstract and Section IV] The abstract and conclusion claim that backward waves are 'significantly suppressed' or that the device demonstrates 'suppression of backward waves,' but no measurement of backscattered waves is reported. The line scans were taken on one side of the antenna at each field polarity, so they show that propagation in one direction decays more slowly than propagation in the opposite direction under field reversal; they do not directly demonstrate the absence of waves propagating backward on the same side. A direct test would be to measure the BLS intensity on both sides of the antenna at a fixed field polarity, or to detect a reflected signal. The wording should be relaxed to 'strongly asymmetric propagation' unless such data are added.
minor comments (6)
  1. [Section III, fitting equation] The fit formula I = I0 exp(−2x/λ) defines λ as an amplitude decay length if I is the spin-wave intensity, but the text refers to it simply as 'decay length' without specifying amplitude versus intensity; this ambiguity should be clarified to avoid a factor-of-two confusion.
  2. [Section II-A] The VNA-FMR results give α = (4.4 ± 0.02) × 10−4 for the YIG layer in the YIG/SiO2/CoFeB stack, whereas the simulations use α = 2 × 10−4 for YIG; the authors should justify this choice or comment on the sensitivity of the simulated dispersion and decay lengths to the YIG damping value.
  3. [Section II-C] The statement that the objective with NA = 0.85 and λL = 457 nm 'enables the detection of wavevectors up to 12 rad/µm' would benefit from the explicit relation kmax = (4π/λL) sin θ or a reference to the formula, since the numerical aperture and wavelength alone do not directly give the wavevector cutoff without the incidence-angle dependence.
  4. [Fig. 1(b)] The experimental error bars are mentioned but their estimation method is not described, and the 'gaps' attributed to phonon modes are not identified or subtracted; a brief description of the error analysis and phonon handling would improve reproducibility.
  5. [Author affiliations] There are two affiliations numbered 7 in the author list—one for Huazhong University of Science and Technology and one for RPTU Kaiserslautern-Landau; the numbering should be corrected.
  6. [Fig. 2] The schematic in Fig. 2 would be easier to follow if the coordinate axes and the direction of the line scan relative to the antenna were labeled explicitly, since the text refers to the x-direction and the y-direction without a clear visual reference.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: decay lengths and dispersion are measured directly; simulations use literature parameters and are not fitted to the diode ratio.

full rationale

The central quantitative claim, the 3:1 decay-length asymmetry (lambda_+200 mT = 10.92 um vs lambda_-200 mT = 3.78 um), is obtained directly from micro-focused BLS line scans fit to I = I0 exp(-2x/lambda). No parameter of this fit is derived from the simulations, and the simulations are not fitted to this ratio. The MuMax3 simulations in Fig. 1(b) use independently characterized or literature material parameters (Ms, Aex, alpha for YIG, SiO2, and CoFeB) and are compared with wavevector-resolved BLS thermal spectra, which independently confirm a non-reciprocal dispersion. The small frequency shift between simulation and k-BLS data is attributed to possible thickness inaccuracy, but this explanation does not enter the extraction of the decay-length ratio or the definition of non-reciprocity. The selection of 7.23 GHz as the excitation frequency because it gave the strongest coherent signal, and the exclusion of near-antenna or thermal-level data points, are experimental analysis choices that affect evidentiary strength, not circular reasoning. There is no load-bearing self-citation chain: the cited prior works (e.g., Refs. 19 and 20) are used for context or analogy, not to define the measured asymmetry. The claim that backward waves are suppressed goes somewhat beyond the presented backscattering measurement, but that is an overstatement of evidence rather than a circular derivation. Overall, the experimental result and the simulation support each other without either being constructed from the other.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim relies on standard micromagnetics and on the assumption that the two magnetic layers are uniformly magnetized and interact only through dipolar fields. The paper provides no free parameters in the classical sense: material parameters are taken from literature or measured by VNA-FMR, and the thickness adjustment used to explain the frequency shift is not quantified as a fit.

assumptions (4)
  • domain assumption The Landau-Lifshitz-Gilbert equation as implemented in MuMax3 is an accurate model of the magnetization dynamics in the YIG/SiO2/CoFeB stack.
    Invoked in Sec. II-B; the simulations are used to predict the dispersion and unidirectional window, so the entire interpretation depends on the LLG model's validity for these layers.
  • domain assumption The CoFeB layer remains uniformly magnetized along the bias field direction, collinear with the YIG magnetization, so the dynamic dipolar coupling mechanism applies.
    Assumed throughout the modeling and in the interpretation of the field-polarity asymmetry; no magnetic imaging of the CoFeB state is provided.
  • domain assumption The 5 nm SiO2 spacer fully suppresses exchange coupling between YIG and CoFeB, leaving only dipolar coupling.
    Assumed from the presence of the spacer (Sec. II-A); the paper contrasts this with the exchanged-coupled case (Ref. 27) but does not independently verify the absence of interlayer exchange.
  • domain assumption The observed non-reciprocity is caused by the dynamic dipolar field between the two layers, with no significant contribution from two-magnon scattering, edge effects, or Dzyaloshinskii-Moriya interaction.
    Discussed qualitatively in Sec. I and III; alternative mechanisms are not experimentally excluded.

how reviews work

0 comments
Cite this review

Pith. "Pith review of YIG/CoFeB bilayer magnonic diode." pith.science (2026). https://pith.science/paper/KSH44AI4

@misc{pith2026241208383,
  author       = {Pith},
  title        = {Pith review of: YIG/CoFeB bilayer magnonic diode},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KSH44AI4}},
  note         = {Machine review of arXiv:2412.08383}
}
abstract

We demonstrate a magnonic diode based on a bilayer structure of Yttrium Iron Garnet (YIG) and Cobalt Iron Boron (CoFeB). The bilayer exhibits pronounced non-reciprocal spin-wave propagation, enabled by dipolar coupling and the magnetic properties of the two layers. The YIG layer provides low damping and efficient spin-wave propagation, while the CoFeB layer introduces strong magnetic anisotropy, critical for achieving diode functionality. Experimental results, supported by numerical simulations, show unidirectional propagation of Magnetostatic Surface Spin Waves (MSSW), significantly suppressing backscattered waves. This behavior was confirmed through wavevector-resolved and micro-focused Brillouin Light Scattering measurements and is supported by numerical simulations. The proposed YIG/SiO$_2$/CoFeB bilayer magnonic diode demonstrates the feasibility of leveraging non-reciprocal spin-wave dynamics for functional magnonic devices, paving the way for energy-efficient, wave-based signal processing technologies.

Figures

Figures reproduced from arXiv: 2412.08383 by the authors.

Figure 1
Figure 1. (a) Schematics of the nano-thick magnetic bilayer of [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. (a) Normalized BLS intensity as a function of laser position in the x-direction with regards to the antenna position dependent on the applied [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 2
Figure 2. We used the microstrip microwave antenna on the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 30 canonical work pages

  1. [1]

    Wu, et al

    Y. Wu, et al. , Advanced Materials 29, 1603031 (2017)

    work page 2017

  2. [2]

    Barman, et al

    A. Barman, et al. , Journal of Physics: Condensed Matter 33, 413001 (2021)

    work page 2021

  3. [3]

    A. V . Chumak, et al. , IEEE Transactions on Magnetics 58, 1 (2022)

    work page 2022

  4. [4]

    Hertel, SPIN 03, 1340009 (2013)

    R. Hertel, SPIN 03, 1340009 (2013)

    work page 2013

  5. [5]

    Pylypovskyi, et al

    O. Pylypovskyi, et al. , Physical Review Letters 114, 197204 (2015)

    work page 2015

  6. [6]

    D. D. Sheka, V . P . Kravchuk, K. V . Yershov, Y. Gaididei,Physical Review B 92, 054417 (2015)

    work page 2015

  7. [7]

    O. V . Pylypovskyi, et al. , Scientific Reports 6, 23316 (2016)

    work page 2016

  8. [8]

    J. A. Otálora, M. Yan, H. Schultheiss, R. Hertel, A. KÃ ˛ akay, Physical Review Letters 117, 227203 (2016)

    work page 2016

Show all 31 references
  1. [9]

    Dzyaloshinsky, Journal of Physics and Chemistry of Solids 4, 241 (1958)

    I. Dzyaloshinsky, Journal of Physics and Chemistry of Solids 4, 241 (1958)

    1958

  2. [10]

    Moriya, Physical Review 120, 91 (1960)

    T . Moriya, Physical Review 120, 91 (1960)

    1960

  3. [11]

    H. T . Nembach, J. M. Shaw, M. Weiler, E. Jué, T . J. Silva, Nature Physics 11, 825 (2015)

    2015

  4. [12]

    Tacchi, et al

    S. Tacchi, et al. , Physical Review Letters 118, 147201 (2017)

    2017

  5. [13]

    F . J. d. Santos, M. d. S. Dias, S. Lounis, Physical Review B 102, 104401 (2020). ArXiv: 2003.11649

    2020 arXiv

  6. [14]

    P . I. Gerevenkov, et al. , Nanoscale 15, 6785 (2023)

    2023

  7. [15]

    R. A. Gallardo, et al. , New Journal of Physics 21, 033026 (2019)

    2019

  8. [16]

    Grünberg, R

    P . Grünberg, R. Schreiber, Y. Pang, M. B. Brodsky, H. Sowers, Physical Review Letters 57, 2442 (1986)

    1986

  9. [17]

    Gallardo, et al

    R. Gallardo, et al. , Physical Review Applied 12, 034012 (2019)

    2019

  10. [18]

    Verba, V

    R. Verba, V . Tiberkevich, A. Slavin, Physical Review Applied 12, 054061 (2019)

    2019

  11. [19]

    Wojewoda, et al

    O. Wojewoda, et al. , Applied Physics Letters 125, 132401 (2024)

    2024

  12. [20]

    Grassi, et al

    M. Grassi, et al. , Physical Review Applied 14, 024047 (2020)

    2020

  13. [21]

    Verba, et al

    R. Verba, et al. , Applied Physics Letters 103, 082407 (2013)

    2013

  14. [22]

    Yu, et al

    G. Yu, et al. , Nano Letters 16, 1981 (2016)

    2016

  15. [23]

    Gladii, M

    O. Gladii, M. Haidar, Y. Henry, M. Kostylev, M. Bailleul, Physical Review B 93, 054430 (2016)

    2016

  16. [24]

    Dubs, et al

    C. Dubs, et al. , Journal of Physics D: Applied Physics 50, 204005 (2017)

    2017

  17. [25]

    Dubs, et al

    C. Dubs, et al. , Physical Review Materials 4, 024416 (2020)

    2020

  18. [26]

    I. S. Maksymov, M. Kostylev, Physica E: Low-dimensional Systems and Nanostructures 69, 253 (2015)

    2015

  19. [27]

    H. Qin, S. J. Hämäläinen, S. van Dijken, Scientific Reports 8, 5755 (2018)

    2018

  20. [28]

    Vansteenkiste, et al

    A. Vansteenkiste, et al. , AIP Advances 4, 107133 (2014)

    2014

  21. [29]

    Sebastian, K

    T . Sebastian, K. Schultheiss, B. Obry, B. Hillebrands, H. Schultheiss, Frontiers in Physics 3 (2015)

    2015

  22. [30]

    R. Mock, B. Hillebrands, R. Sandercock, Journal of Physics E: Scien- tific Instruments 20, 656 (1987)

    1987

  23. [31]

    Wojewoda, Modeling of the micro-focused Brillouin light scatter- ing spectra (2024)

    O. Wojewoda, Modeling of the micro-focused Brillouin light scatter- ing spectra (2024). ArXiv:2409.05141. 5

    2024 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.