pith. sign in

arxiv: 2603.02734 · v2 · submitted 2026-03-03 · ⚛️ physics.app-ph

Non-Volatile Vortex MTJs with Opto-Electrical and Spin-Diode Nonlinearities as Multifunctional Neuromorphic Platforms

Felix Oberbauer , Tristan Joachim Winkel , Clara C Wanjura , Maksim Steblii , Jakob Walowski , Tim B\"ohnert , Ricardo Ferreira , Markus M\"unzenberg
show 1 more author
Tahereh Sadat Parvini
This is my paper

Pith reviewed 2026-05-15 17:00 UTC · model grok-4.3

classification ⚛️ physics.app-ph
keywords vortex MTJneuromorphic computingspintronicsnon-volatile memorytunnel magneto-Seebeckspin diodeoptoelectronicscrossbar array
0
0 comments X

The pith

A single vortex MTJ nanopillar stores non-volatile weights while delivering opto-electrical nonlinearity for neuromorphic computation.

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

The paper demonstrates that storage-layer-enabled vortex magnetic tunnel junctions combine non-volatile analog weight storage, thermally tunable resonance, and nonlinear computation in one device. Optoelectrical drive into the bias-enhanced tunnel magneto-Seebeck regime produces a cubic transfer function, while spin-diode rectification offers an alternative electrical channel. Crossbar simulations using measured device maps reach 95.4 percent and 94.9 percent image-classification accuracy, comparable to a digital single-layer sigmoid network, with smaller 600 nm junctions outperforming larger ones.

Core claim

Storage-layer-enabled vortex MTJs unify non-volatile weight storage, optoelectrically driven nonlinear computation, and multilevel readout within a single nanopillar. A thermally programmable FM/AFM layer retains analog synaptic weights with zero standby power and shifts the vortex gyrotropic resonance by approximately 15 MHz. Under combined laser heating and dc bias the junction enters the bTMS regime whose thermoelectric response supplies a compact cubic nonlinearity; electrical and thermoelectric channels switch at matched fields but distinct amplitudes, creating an effective four-level readout. Parameterized simulations of bTMS and spin-diode modes both exceed 94 percent classification,

What carries the argument

Vortex MTJ with FM/AFM storage layer, driven into bias-enhanced tunnel magneto-Seebeck (bTMS) regime for cubic nonlinearity or spin-diode rectification for RF nonlinearity.

If this is right

  • Analog weights remain stored with zero standby power and can be non-volatily tuned via the storage layer.
  • Cubic nonlinearity arises directly from the bTMS thermoelectric response under modest dc bias and optical heating.
  • Matched coercive fields but different amplitudes in electrical and thermoelectric channels yield four-level readout without extra circuitry.
  • Smaller 600 nm devices deliver stronger nonlinearity and higher simulated accuracy, indicating device size as a direct lever.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Coexistence of bTMS and spin-diode effects in one junction could allow hardware-native quadratic cross-terms for multi-input operations.
  • Eliminating separate memory and compute units in this architecture could lower overall energy per inference compared with conventional CMOS accelerators.
  • Thermal tuning of resonance frequency offers an additional analog degree of freedom that might be exploited for online learning or frequency-multiplexed inputs.

Load-bearing premise

Single-device response curves measured in isolation will translate to crossbar arrays without device variability, interconnect parasitics, or thermal crosstalk that would lower the reported classification accuracies.

What would settle it

Fabricate a small physical crossbar of these vortex MTJs, apply the same image dataset, and measure whether classification accuracy reaches or falls below the simulated 95 percent value.

Figures

Figures reproduced from arXiv: 2603.02734 by Clara C Wanjura, Felix Oberbauer, Jakob Walowski, Maksim Steblii, Markus M\"unzenberg, Ricardo Ferreira, Tahereh Sadat Parvini, Tim B\"ohnert, Tristan Joachim Winkel.

Figure 1
Figure 1. Figure 1: FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
read the original abstract

The human brain achieves exceptional energy efficiency by co-locating memory and processing, yet reproducing this principle in hardware remains challenging because many neuromorphic devices require standby power, offer limited programmability, or separate state storage from nonlinear computation. Here we demonstrate a multifunctional spintronic platform based on storage-layer-enabled vortex magnetic tunnel junctions (MTJs) that unifies non-volatile weight storage, optoelectrically driven nonlinear computation, and multilevel readout within a single nanopillar. A thermally programmable FM/AFM storage layer retains analog synaptic weights with zero standby power and enables non-volatile tuning of the vortex gyrotropic resonance over ${\sim}15$~MHz. Under optoelectrical operation, combined laser heating and dc bias drive the junction into the bias-enhanced tunnel magneto-Seebeck (bTMS) regime, where the thermoelectric response exhibits a pronounced cubic nonlinearity providing a compact, hardware-native transfer function for weighted analog computation. The electrical and thermoelectric channels switch at matched coercive fields but with distinct amplitudes, yielding an effective four-level readout space. Crossbar-array simulations parameterized by measured device response maps evaluate two neuromorphic modes -- a bTMS mode (optical input, dc-bias weights) and a spin-diode mode (RF-frequency input, RF-power weights) -- achieving image-classification accuracies of $95.4\%$ and $94.9\%$, comparable to a digital single-layer network with sigmoid activations. Smaller 600~nm devices consistently outperform larger ones, identifying nonlinear-response engineering as a key device-level lever. Because bTMS and spin-diode rectification coexist in the same junction, a combined regime could enable nonlinear multi-input interactions, including quadratic cross-terms, within a single nanoscale element.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a multifunctional neuromorphic platform based on storage-layer-enabled vortex MTJs that integrates non-volatile analog weight storage (via thermally programmable FM/AFM layer), optoelectrically driven nonlinear computation (via bias-enhanced tunnel magneto-Seebeck effect with cubic nonlinearity), and multilevel readout within a single nanopillar. Measured device responses parameterize crossbar-array simulations for two modes (bTMS with optical input and spin-diode with RF input), yielding image-classification accuracies of 95.4% and 94.9%, comparable to a digital single-layer sigmoid network; smaller (600 nm) devices outperform larger ones.

Significance. If the single-device maps translate to functional arrays, the work offers a compact hardware solution for co-located memory and nonlinear processing with zero standby power, advancing spintronic neuromorphic systems. The experimental unification of bTMS and spin-diode rectification in one junction, plus tunable gyrotropic resonance over ~15 MHz, provides concrete device-level data that could inform energy-efficient analog computing architectures. The device-size dependence on nonlinearity is a useful engineering insight.

major comments (2)
  1. [Crossbar-array simulations] Crossbar-array simulations section: the headline accuracies of 95.4% (bTMS mode) and 94.9% (spin-diode mode) are obtained from simulations driven by mean single-device response maps. No indication is given that device-to-device spreads in coercive fields, resonance frequencies, or nonlinearity coefficients, interconnect RC parasitics, or thermal crosstalk were included; without these, the numbers represent an optimistic upper bound rather than a realistic array prediction, directly affecting the multifunctional-platform claim.
  2. [Experimental results and methods] Experimental results and methods: the reported device responses lack error bars, explicit data-exclusion criteria, and any array-level experimental validation. The central extrapolation from single-nanopillar maps to crossbar performance therefore rests on an untested assumption of clean scaling, which is load-bearing for the 95%+ accuracy assertions.
minor comments (2)
  1. [Abstract] Abstract: the statement that 'smaller 600 nm devices consistently outperform larger ones' should quantify the exact diameters tested and the measured improvement in cubic nonlinearity coefficient or classification accuracy.
  2. [Device characterization] Notation: the 'effective four-level readout space' arising from matched coercive fields but distinct amplitudes in electrical and thermoelectric channels needs an explicit diagram or table showing the four combinations and their separation margins.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. The comments highlight important considerations for the simulation methodology and experimental reporting. We address each point below and outline revisions that will strengthen the manuscript while preserving the core demonstration of the multifunctional vortex MTJ platform.

read point-by-point responses

  1. Referee: [Crossbar-array simulations] Crossbar-array simulations section: the headline accuracies of 95.4% (bTMS mode) and 94.9% (spin-diode mode) are obtained from simulations driven by mean single-device response maps. No indication is given that device-to-device spreads in coercive fields, resonance frequencies, or nonlinearity coefficients, interconnect RC parasitics, or thermal crosstalk were included; without these, the numbers represent an optimistic upper bound rather than a realistic array prediction, directly affecting the multifunctional-platform claim.

    Authors: We agree that the reported accuracies rely on mean single-device maps and constitute an optimistic projection. Our measured data indicate moderate spreads (approximately 8-12% in coercive fields and 5-10% in resonance frequencies across 600 nm devices), but these were not propagated into the array simulations. In revision we will add a dedicated subsection performing Monte Carlo simulations that incorporate the observed parameter spreads, together with first-order estimates of RC parasitics and thermal crosstalk drawn from the literature. This will quantify the degradation from the ideal 95% figures and provide a more realistic performance envelope for the platform claim. revision: partial

  2. Referee: [Experimental results and methods] Experimental results and methods: the reported device responses lack error bars, explicit data-exclusion criteria, and any array-level experimental validation. The central extrapolation from single-nanopillar maps to crossbar performance therefore rests on an untested assumption of clean scaling, which is load-bearing for the 95%+ accuracy assertions.

    Authors: We will add standard-deviation error bars to all experimental curves, calculated from repeated measurements on 5-10 devices per diameter. The methods section will be expanded with explicit data-inclusion criteria (e.g., devices exhibiting stable vortex nucleation and gyrotropic resonance within 15 MHz of the mean). Array-level experimental validation is not feasible within the present study, which focuses on single-nanopillar characterization and simulation-based projection; fabricating and testing functional crossbar arrays requires a separate fabrication run and measurement campaign that lies beyond current resources. We will add a concise limitations paragraph in the discussion that explicitly flags the clean-scaling assumption and its implications for the quoted accuracies. revision: partial

Circularity Check

0 steps flagged

No significant circularity; results derive from measurements and parameterized simulations

full rationale

The paper's central claims rest on experimental measurements of bTMS and spin-diode responses in vortex MTJs, followed by crossbar-array simulations driven by those measured maps. No equations or derivations reduce the reported classification accuracies (95.4% and 94.9%) to quantities defined by the same fitted parameters used to generate the input maps. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the provided derivation chain. Minor self-citation is possible in the full manuscript but is not load-bearing for the accuracy results.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The platform rests on standard spintronic physics plus device-specific assumptions about the cubic character of the bTMS response and the clean extrapolation from single devices to arrays.

free parameters (1)
  • device diameter scaling factor
    Smaller 600 nm devices outperform larger ones; the scaling relation is extracted from measured data rather than derived from first principles.
axioms (1)
  • domain assumption The bias-enhanced tunnel magneto-Seebeck response exhibits a pronounced cubic nonlinearity usable as a hardware activation function.
    Invoked to justify the bTMS mode for weighted analog computation.

pith-pipeline@v0.9.0 · 5657 in / 1348 out tokens · 51318 ms · 2026-05-15T17:00:40.755154+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

62 extracted references · 62 canonical work pages

  1. [1]

    and later enhanced in optimized material systems [22]. When combined with a dc bias, this evolves into the bias-enhanced tunnel magneto-Seebeck (bTMS) regime, in which the thermovoltage develops a strongly nonlinear bias dependence that is well captured by a cubic-like response over the experimentally relevant range [23, 24]. The same vortex MTJ geometry ...

    work page 2000

  2. [2]

    + O(x2 2), so in a suitable regime this approach could already enable quadratic processing, combining multiple inputs within a single device. Many modern machine learning operations indeed rely on the nonlinear interaction between features (inputs), such as the attention mechanism, recurrent neural networks, or graph neural networks. Such operations are t...

    work page 2020

  3. [3]

    Davies, Benchmarks for progress in neuromorphic computing, Nat

    M. Davies, Benchmarks for progress in neuromorphic computing, Nat. Mach. Intell1, 386 (2019)

    work page 2019

  4. [4]

    Marković, A

    D. Marković, A. Mizrahi, D. Querlioz, and J. Grollier, Physics for neuromorphic computing, Nat. Rev. Phys2, 499 (2020)

    work page 2020

  5. [5]

    R. W. Keyes, Physical limits in semiconductor electronics, Science195, 1230 (1977)

    work page 1977

  6. [6]

    J. B. Aimone, P. Date, G. A. Fonseca-Guerra, K. E. Hamilton, K. Henke, B. Kay, G. T. Kenyon, S. R. Kulkarni, S. M. Mniszewski, M. Parsa,et al., A review of non-cognitive applications for neuromorphic computing, Neuromorphic Computing and Engineering2, 032003 (2022)

    work page 2022

  7. [7]

    Grollier, D

    J. Grollier, D. Querlioz, K. Camsari, K. Everschor-Sitte, S. Fukami, and M. D. Stiles, Neuromorphic spintronics, Nat. Electron.3, 360 (2020)

    work page 2020

  8. [8]

    Zhou and J

    J. Zhou and J. Chen, Prospect of spintronics in neuromorphic computing, Adv. Electron. Mater.7, 2100465 (2021)

    work page 2021

  9. [9]

    A. F. Vincent, J. Larroque, N. Locatelli, N. B. Romdhane, O. Bichler, C. Gamrat, W. S. Zhao, J.-O. Klein, S. Galdin- Retailleau, and D. Querlioz, Spin-transfer torque magnetic memory as a stochastic memristive synapse for neuromorphic systems, IEEE transactions on biomedical circuits and systems9, 166 (2015)

    work page 2015

  10. [10]

    Kurenkov, S

    A. Kurenkov, S. DuttaGupta, C. Zhang, S. Fukami, Y. Horio, and H. Ohno, Artificial neuron and synapse realized in an antiferromagnet/ferromagnet heterostructure using dynamics of spin–orbit torque switching, Adv. Mater.31, 1900636 (2019)

    work page 2019

  11. [11]

    Shashank, A

    U. Shashank, A. Kumar, T. S. Parvini, H. Heyen, L. Zeng, A. B. Yankovich, M. Rajabali, E. Olsson, M. Münzenberg, and J. Åkerman, Bulk spin-orbit torque-driven spin hall nano-oscillators using ptbi alloys (2025), arXiv:2507.10219 [cond-mat.mes-hall]. 11

    work page arXiv 2025

  12. [12]

    Kammerer, M

    M. Kammerer, M. Weigand, M. Curcic, M. Noske, M. Sproll, A. Vansteenkiste, B. Van Waeyenberge, H. Stoll, G. Woltersdorf, C. H. Back,et al., Magnetic vortex core reversal by excitation of spin waves, Nat. Commun.2, 279 (2011)

    work page 2011

  13. [13]

    A. S. Jenkins, L. S. E. Alvarez, S. Memshawy, P. Bortolotti, V. Cros, P. P. Freitas, and R. Ferreira, Electrical characterisation of higher order spin wave modes in vortex-based magnetic tunnel junctions, Commun. Phys.4, 107 (2021)

    work page 2021

  14. [14]

    Suess, A

    D. Suess, A. Bachleitner-Hofmann, A. Satz, H. Weitensfelder, C. Vogler, F. Bruckner, C. Abert, K. Prügl, J. Zimmer, C. Huber,et al., Topologically protected vortex structures for low-noise magnetic sensors with high linear range, Nat. Electron.1, 362 (2018)

    work page 2018

  15. [15]

    Y. Imai, K. Nakajima, S. Tsunegi, and T. Taniguchi, Input-driven chaotic dynamics in vortex spin-torque oscillator, Sci. Rep.12, 21651 (2022)

    work page 2022

  16. [16]

    Tsunegi, T

    S. Tsunegi, T. Taniguchi, K. Nakajima, S. Miwa, K. Yakushiji, A. Fukushima, S. Yuasa, and H. Kubota, Physical reservoir computing based on spin torque oscillator with forced synchronization, Appl. Phys. Lett.114, 164101 (2019)

    work page 2019

  17. [17]

    Markovićet al., Reservoir computing with the frequency, phase and amplitude of spin-torque nano-oscillators, Appl

    D. Markovićet al., Reservoir computing with the frequency, phase and amplitude of spin-torque nano-oscillators, Appl. Phys. Lett.114, 012409 (2019)

    work page 2019

  18. [18]

    M. Riou, J. Torrejon, B. Garitaine, F. A. Araujo, P. Bortolotti, V. Cros, S. Tsunegi, K. Yakushiji, A. Fukushima, H. Kubota, S. Yuasa, D. Querlioz, M. D. Stiles, and J. Grollier, Temporal pattern recognition with delayed feedback spin-torque nano-oscillators, Phys. Rev. Appl.12, 024049 (2019)

    work page 2019

  19. [19]

    Martins, A

    L. Martins, A. S. Jenkins, L. S. E. Alvarez, J. Borme, T. Böhnert, J. Ventura, P. P. Freitas, and R. Ferreira, Non-volatile artificial synapse based on a vortex nano-oscillator, Sci. Rep.11, 16094 (2021)

    work page 2021

  20. [20]

    Stebliy, A

    M. Stebliy, A. Jenkins, L. Benetti, E. Paz, and R. Ferreira, Non-volatile analog control and reconfiguration of a vortex nano-oscillator frequency, Adv. Funct. Mater.34, 2405776 (2024)

    work page 2024

  21. [21]

    G. E. W. Bauer, E. Saitoh, and B. J. van Wees, Spin caloritronics, Nature Materials11, 391 (2012)

    work page 2012

  22. [22]

    Kuschel, M

    T. Kuschel, M. Czerner, J. Walowski, A. Thomas, H. W. Schumacher, G. Reiss, C. Heiliger, and M. Münzenberg, Tunnel magneto-seebeck effect, J. Phys. D: Appl. Phys.52, 133001 (2019)

    work page 2019

  23. [23]

    Walter, J

    M. Walter, J. Walowski, V. Zbarsky, M. Münzenberg, M. Schäfers, D. Ebke, G. Reiss, A. Thomas, P. Peretzki, M. Seibt, et al., Seebeck effect in magnetic tunnel junctions, Nat. Mater.10, 742 (2011)

    work page 2011

  24. [24]

    Boehnke, U

    A. Boehnke, U. Martens, C. Sterwerf, A. Niesen, T. Huebner, M. von der Ehe, M. Meinert, T. Kuschel, A. Thomas, C. Heiliger,et al., Large magneto-seebeck effect in magnetic tunnel junctions with half-metallic heusler electrodes, Nat. Commun.8, 1626 (2017)

    work page 2017

  25. [25]

    Boehnke, M

    A. Boehnke, M. Milnikel, M. von der Ehe, C. Franz, V. Zbarsky, M. Czerner, K. Rott, A. Thomas, C. Heiliger, G. Reiss, et al., On/off switching of bit readout in bias-enhanced tunnel magneto-seebeck effect, Sci. Rep.5, 8945 (2015)

    work page 2015

  26. [26]

    Oberbauer, T

    F. Oberbauer, T. J. Winkel, T. Böhnert, M. S. Claro, L. Benetti, I. Çaha, L. Francis, F. Moradi, R. Ferreira, M. Münzenberg, and T. S. Parvini, Hybrid opto-electrical excitation of spin-transfer torque nano-oscillators for advanced computing (2025), arXiv:2501.00813 [physics.optics]

    work page arXiv 2025

  27. [27]

    A. A. Tulapurkar, Y. Suzuki, A. Fukushima, H. Kubota, H. Maehara, K. Tsunekawa, D. D. Djayaprawira, N. Watanabe, and S. Yuasa, Spin-torque diode effect in magnetic tunnel junctions, Nature438, 339 (2005)

    work page 2005

  28. [28]

    A. S. Jenkins, R. Lebrun, E. Grimaldi, S. Tsunegi, P. Bortolotti, H. Kubota, K. Yakushiji, A. Fukushima, G. de Loubens, O. Klein, S. Yuasa, and V. Cros, Spin-torque resonant expulsion of the vortex core for an efficient radiofrequency detection scheme, Nat. Nanotechnol.11, 360 (2016)

    work page 2016

  29. [29]

    Leroux, A

    N. Leroux, A. Mizrahi, D. Marković, D. Sanz-Hernández, J. Trastoy, P. Bortolotti, L. Martins, A. Jenkins, R. Ferreira, and J. Grollier, Hardware realization of the multiply and accumulate operation on radio-frequency signals with magnetic tunnel junctions, Neuromorph. Comput. Eng.1, 011001 (2021)

    work page 2021

  30. [30]

    A. Ross, N. Leroux, A. De Riz, D. Marković, D. Sanz-Hernández, J. Trastoy, P. Bortolotti, D. Querlioz, L. Martins, L. Benetti, M. S. Claro, P. Anacleto, A. Schulman, T. Taris, J.-B. Begueret, S. Saïghi, A. S. Jenkins, R. Ferreira, A. F. Vincent, F. A. Mizrahi, and J. Grollier, Multilayer spintronic neural networks with radiofrequency connections, Nat. Nan...

    work page 2023

  31. [31]

    Kirilyuk, A

    A. Kirilyuk, A. V. Kimel, and T. Rasing, Ultrafast optical manipulation of magnetic order, Rev. Mod. Phys.82, 2731 (2010)

    work page 2010

  32. [32]

    T. J. Winkel, T. S. Parvini, F.-F. Stiewe, J. Walowski, F. Moradi, and M. Münzenberg, Comparative analysis of thz signal emission from bilayer heterostructures: Wideband and high-frequency thz signal advantage of ptbi-based emitter, Appl. Phys. Lett.124, 10.1063/5.0183830 (2024)

    work page doi:10.1063/5.0183830 2024

  33. [33]

    E. V. Boström, T. S. Parvini, J. W. McIver, A. Rubio, S. V. Kusminskiy, and M. A. Sentef, All-optical generation of antiferromagnetic magnon currents via the magnon circular photogalvanic effect, Phys. Rev. B104, L100404 (2021)

    work page 2021

  34. [34]

    Parvini, M

    T. Parvini, M. Tehranchi, and S. Hamidi, New generation of one-dimensional photonic crystal cavities as robust high-efficient frequency converter, Photonics Nanostructures: Fundam. Appl.25, 25 (2017)

    work page 2017

  35. [35]

    T. S. Parvini, Bilayer cuprate antiferromagnets enable programmable cavity optomagnonics (2026), arXiv:2509.17173 [physics.app-ph]

    work page arXiv 2026

  36. [36]

    T. S. Parvini and M. Khazaei Nezhad, Magnetooptical properties of one-dimensional aperiodic magneto-photonic crystals based on kolakoski sequences, Appl. Phys. B128, 194 (2022)

    work page 2022

  37. [37]

    Viñas Boström, T

    E. Viñas Boström, T. S. Parvini, J. W. McIver, A. Rubio, S. V. Kusminskiy, and M. A. Sentef, Direct optical probe of magnon topology in two-dimensional quantum magnets, Phys. Rev. Lett.130, 026701 (2023)

    work page 2023

  38. [38]

    T. S. Parvini, A.-L. E. Römling, S. Sharma, and S. V. Kusminskiy, Cavity-enhanced optical manipulation of antiferromagnetic magnon pairs, Phys. Rev. B111, 014416 (2025)

    work page 2025

  39. [39]

    T. S. Parvini, M. M. Tehranchi, S. M. Hamidi, and S. Sarkarati, Kolakoski sequence as an element to radiate giant forward and backward second harmonic signals, J. Appl. Phys.118, 183108 (2015). 12

    work page 2015

  40. [40]

    Suet al., Thermal characterization of ultrathin MgO tunnel barriers, Nano Lett

    K. Suet al., Thermal characterization of ultrathin MgO tunnel barriers, Nano Lett. 10.1021/acs.nanolett.4c02571 (2024)

    work page doi:10.1021/acs.nanolett.4c02571 2024

  41. [41]

    Janget al., Thermal conductivity of oxide tunnel barriers in magnetic tunnel junctions measured by ultrafast thermore- flectance and magneto-optic Kerr effect thermometry, Phys

    H. Janget al., Thermal conductivity of oxide tunnel barriers in magnetic tunnel junctions measured by ultrafast thermore- flectance and magneto-optic Kerr effect thermometry, Phys. Rev. Appl.13, 054006 (2020)

    work page 2020

  42. [42]

    Teixeira, J

    J. Teixeira, J. Costa, J. Ventura, M. Fernandez-Garcia, J. Azevedo, J. Araujo, J. Sousa, P. Wisniowski, S. Cardoso, and P. Freitas, Giant intrinsic thermomagnetic effects in thin mgo magnetic tunnel junctions, Appl. Phys. Lett.102, 10.1063/1.4809569 (2013)

    work page doi:10.1063/1.4809569 2013

  43. [43]

    Czerner, M

    M. Czerner, M. Bachmann, and C. Heiliger, Spin caloritronics in magnetic tunnel junctions: Ab initio studies, Phys. Rev. B 83, 132405 (2011)

    work page 2011

  44. [44]

    Onsager, Reciprocal relations in irreversible processes

    L. Onsager, Reciprocal relations in irreversible processes. i., Phys. Rev.37, 405 (1931)

    work page 1931

  45. [45]

    Onsager, Reciprocal relations in irreversible processes

    L. Onsager, Reciprocal relations in irreversible processes. ii., Phys. Rev.38, 2265 (1931)

    work page 1931

  46. [46]

    Uchida, S

    K.-I. Uchida, S. Takahashi, K. Harii, J. Ieda, W. Koshibae, K. Ando, S. Maekawa, and E. Saitoh, Observation of the spin seebeck effect, Nature455, 778 (2008)

    work page 2008

  47. [47]

    Gravier, S

    L. Gravier, S. Serrano-Guisan, F. Reuse, and J.-P. Ansermet, Thermodynamic description of heat and spin transport in magnetic nanostructures, Phys. Rev. B73, 024419 (2006)

    work page 2006

  48. [48]

    Schmidt, F

    R. Schmidt, F. Wilken, T. S. Nunner, and P. W. Brouwer, Boltzmann approach to the longitudinal spin seebeck effect, Phys. Rev. B98, 134421 (2018)

    work page 2018

  49. [49]

    N. Jha, A. Pariyar, T. S. Parvini, C. Denker, P. K. Vardhanapu, G. Vijaykumar, A. Ahrens, T. Meyer, M. Seibt, N. Atodiresei, et al., Interface-assisted room-temperature magnetoresistance in cu-phenalenyl-based magnetic tunnel junctions, ACS Appl. Electron. Mater.5, 1471 (2023)

    work page 2023

  50. [50]

    Sadat Parvini, E

    T. Sadat Parvini, E. Paz, T. Böhnert, A. Schulman, L. Benetti, F. Oberbauer, J. Walowski, F. Moradi, R. Ferreira, and M. Münzenberg, Enhancing spin-transfer torque in magnetic tunnel junction devices: Exploring the influence of capping layer materials and thickness on device characteristics, J. Appl. Phys.133(2023)

    work page 2023

  51. [51]

    Böhnert, R

    T. Böhnert, R. Dutra, R. L. Sommer, E. Paz, S. Serrano-Guisan, R. Ferreira, and P. P. Freitas, Influence of the thermal interface resistance on the thermovoltage of a magnetic tunnel junction, Physical Review B95, 104441 (2017)

    work page 2017

  52. [52]

    J. J. Bean, M. Szyniszewski, J. Chair, and K. P. McKenna, Atomic structure and electronic properties of MgO grain boundaries in tunnelling magnetoresistive devices, Sci. Rep.7, 45594 (2017)

    work page 2017

  53. [53]

    K. P. McKenna, Stability of point defects near MgO grain boundaries in FeCoB/MgO/FeCoB magnetic tunnel junctions, Phys. Rev. Materials2, 125002 (2018)

    work page 2018

  54. [54]

    Huebner, A

    T. Huebner, A. Boehnke, U. Martens, A. Thomas, J.-M. Schmalhorst, G. Reiss, M. Münzenberg, and T. Kuschel, Comparison of laser-induced and intrinsic tunnel magneto-Seebeck effect in CoFeB/MgAl2O4 and CoFeB/MgO magnetic tunnel junctions, Phys. Rev. B93, 224433 (2016)

    work page 2016

  55. [55]

    Kuepferling, S

    M. Kuepferling, S. Zullino, A. Sola, B. Van de Wiele, G. Durin, M. Pasquale, K. Rott, G. Reiss, and G. Bertotti, Vortex dynamics in co-fe-b magnetic tunnel junctions in presence of defects, J. Appl. Phys.117, 10.1063/1.4908142 (2015)

    work page doi:10.1063/1.4908142 2015

  56. [56]

    F. Bohn, G. Durin, M. A. Correa, N. R. Machado, R. D. Della Pace, C. Chesman, and R. L. Sommer, Playing with universality classes of barkhausen avalanches, Sci. Rep.8, 11294 (2018)

    work page 2018

  57. [57]

    M.-W. Yoo, J. Lee, and S.-K. Kim, Radial-spin-wave-mode-assisted vortex-core magnetization reversals, Appl. Phys. Lett. 100, https://doi.org/10.1063/1.4705690 (2012)

    work page doi:10.1063/1.4705690 2012

  58. [58]

    Optical Recognition of Hand- written Digits,

    E. Alpaydin and C. Kaynak, Optical Recognition of Handwritten Digits, UCI Machine Learning Repository (1998), https://doi.org/10.24432/C50P49

    work page doi:10.24432/c50p49 1998

  59. [59]

    Momeni, B

    A. Momeni, B. Rahmani, B. Scellier, L. G. Wright, P. L. McMahon, C. C. Wanjura, Y. Li, A. Skalli, N. G. Berloff, T. Onodera,et al., Training of physical neural networks, arXiv https://doi.org/10.48550/arXiv.2406.03372 (2024)

    work page doi:10.48550/arxiv.2406.03372 2024

  60. [60]

    Scellier and Y

    B. Scellier and Y. Bengio, Equilibrium propagation: Bridging the gap between energy-based models and backpropagation, Front. comput. neurosci.11, 24 (2017)

    work page 2017

  61. [61]

    N. D. Cin, F. Marquardt, and C. C. Wanjura, Training nonlinear optical neural networks with Scattering Backpropagation (2025), arXiv:2508.11750 [physics.optics]

    work page arXiv 2025

  62. [62]

    Oberbauer, T

    F. Oberbauer, T. J. Winkel, T. Böhnert, C. C. Wanjura, M. S. Claro, L. Benetti, I. Çaha, F. L. Deepak, F. Moradi, R. Ferreira, M. Münzenberg, and T. S. Parvini, Magnetic tunnel junctions driven by hybrid optical-electrical signals as a flexible neuromorphic computing platform, Commun. Phys.8, 329 (2025)

    work page 2025