pith. sign in

arxiv: 2604.22574 · v1 · pith:V7JKU4KXnew · submitted 2026-04-24 · ❄️ cond-mat.mes-hall · physics.app-ph

Pulse Shaping to Mitigate the Impact of Device Imperfections in Field-Free Switching Using Combined Spin-Orbit and Spin-Transfer Torques

Kuldeep Ray , J\'er\'emie Vigier , Sylvain Martin , Chlo\'e Bouard , Nicolas Lefoulon , Marc Drouard , Gilles Gaudin This is my paper

Pith reviewed 2026-05-08 10:18 UTC · model grok-4.3

classification ❄️ cond-mat.mes-hall physics.app-ph
keywords SOT-MRAMfield-free switchingspin-orbit torquespin-transfer torquepulse shapingwrite error ratebackhoppingmacrospin model
0
0 comments X

The pith

Shaping the STT pulse reduces write error rates and improves robustness in field-free SOT-STT switching of top-pinned devices.

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

The paper examines combined spin-orbit and spin-transfer torque switching for field-free operation in SOT-MRAM, noting that this approach brings backhopping and pronounced asymmetry between AP-to-P and P-to-AP transitions in top-pinned stacks. Experiments on non-optimized devices reveal these reliability issues, while a macrospin model built from two coupled Landau-Lifshitz-Gilbert equations reproduces the asymmetry and identifies an intermediate loss-of-determinism regime. The authors then demonstrate that shaping the STT pulse component lowers the write error rate and makes switching more tolerant of device imperfections.

Core claim

In top-pinned SOT-MRAM devices, the combination of SOT and STT for field-free switching exhibits pronounced asymmetry between AP-to-P and P-to-AP transitions due to STT-induced backhopping. A macrospin model with two coupled LLG equations reproduces this asymmetry and uncovers an intermediate loss-of-determinism regime. Pulse shaping of the STT component reduces the write error rate and enhances switching robustness against device imperfections.

What carries the argument

Macrospin model of two coupled Landau-Lifshitz-Gilbert equations for the free and reference layers, which simulates the dynamics and guides the identification of effective STT pulse shapes.

If this is right

  • Lower write error rates can be achieved without optimizing the stack for pure STT switching.
  • Switching becomes more robust to variations in device parameters such as pinning strength or layer thickness.
  • The loss-of-determinism regime can be avoided by appropriate pulse timing and amplitude.
  • Practical mitigation strategies become available for industrial deployment of SOT-MRAM.

Where Pith is reading between the lines

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

  • Similar pulse shaping approaches could extend to other torque combinations or bottom-pinned stacks.
  • Circuit-level integration of pulse generators might improve overall memory yield and energy efficiency.
  • Quantitative mapping of the loss-of-determinism boundaries could inform design rules for scaled devices.

Load-bearing premise

The macrospin approximation with two coupled equations sufficiently captures the essential dynamics of real multilayer devices to guide effective pulse shaping.

What would settle it

Performing the pulse shaping experiment on devices and finding no statistically significant reduction in write error rate compared to unshaped pulses would falsify the mitigation claim.

Figures

Figures reproduced from arXiv: 2604.22574 by Chlo\'e Bouard, Gilles Gaudin, J\'er\'emie Vigier, Kuldeep Ray, Marc Drouard, Nicolas Lefoulon, Sylvain Martin.

Figure 1
Figure 1. Figure 1: (a) Schematic of the material stack of SOT-MRAM devices measured in this study. (b) Representative TMR hysteresis loop obtained by sweeping the out-of-plane field (HZ), showing a TMR = 63% at HZ=0 and an offset field Hoff = –11.94 kA/m. B. Experimental Setup The measurement setup is shown in view at source ↗
Figure 5
Figure 5. Figure 5: Simulated WER versus SOT current density using different view at source ↗
Figure 6
Figure 6. Figure 6: (a) WER versus VSTT for standard rectangular and modified pulse shapes. (b) WER versus VSOT for a modified two-step SOT pulse. Despite the presence of imperfections, the simulations in view at source ↗
Figure 7
Figure 7. Figure 7: WER versus VSOT using modified SOT and STT pulses (shown in inset) (a) for different VSTT, and (b) for different overlaps between SOT and STT pulses with VSTT = 0.77 V. A negative overlap corresponds to the separation between the end of the SOT and the beginning of the STT pulses. VI. CONCLUSION WER measurements provide substantial evidence of device imperfections in our top-pinned SOT-MRAM. Through macros… view at source ↗
read the original abstract

Combining spin-orbit (SOT) and spin-transfer torques (STT) provides a practical approach for field-free switching in spin-orbit torque magnetic random-access memory (SOT-MRAM), a prerequisite for industrial deployment, but can compromise reliability through phenomena such as backhopping, especially in top-pinned stacks commonly used for SOT-MRAM. We investigate the write error rate (WER) of combined SOT + STT switching in top-pinned devices that are not optimized for STT switching. Experiments reveal clear indications of STT-induced backhopping and a pronounced field-free SOT switching asymmetry between AP-to-P and P-to-AP transitions. Our macrospin model, using two coupled Landau Lifshitz Gilbert equations for the free and the reference layers, qualitatively reproduces this asymmetry and reveals an intermediate loss-of-determinism regime in addition to the well-known backhopping region. Based on these simulations, we propose mitigation strategies and experimentally demonstrate that STT pulse shaping reduces WER and improves switching robustness in the presence of device imperfections.

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 examines field-free switching in top-pinned SOT-MRAM devices using combined spin-orbit torque (SOT) and spin-transfer torque (STT). Experiments reveal STT-induced backhopping and a pronounced asymmetry between AP-to-P and P-to-AP transitions. A macrospin model based on two coupled Landau-Lifshitz-Gilbert equations qualitatively reproduces the asymmetry, identifies an intermediate loss-of-determinism regime, and is used to propose STT pulse-shaping mitigation strategies, which are then shown experimentally to reduce write error rate (WER) and improve robustness against device imperfections.

Significance. If the central claim holds, the work is significant for SOT-MRAM reliability, a prerequisite for industrial use. The experimental demonstration that pulse shaping mitigates observed errors, informed by the model, offers a practical route to handling imperfections in combined-torque switching. The macrospin approach and qualitative agreement provide a starting point, but the absence of quantitative WER matching or micromagnetic checks limits the strength of the guidance provided by the model.

major comments (2)
  1. [Modeling and simulation section] The macrospin model (two coupled LLG equations for free and reference layers) is invoked to identify the loss-of-determinism regime and to guide the STT pulse-shaping strategy. Because the manuscript explicitly attributes the backhopping and asymmetry to device imperfections in top-pinned stacks (which commonly produce spatial variations in anisotropy or pinning that nucleate non-uniform modes), the uniform-magnetization assumption requires explicit justification or cross-validation; without micromagnetic simulations or quantitative WER matching, it is unclear whether the model's regime corresponds to the experimental error mechanisms.
  2. [Experimental results and discussion] The experimental demonstration of reduced WER via pulse shaping is presented as qualitative agreement with the model. Specific numerical WER values (before/after shaping), error bars, exact pulse amplitudes/durations, and statistics across devices are not provided, making it difficult to assess the magnitude, reproducibility, and practical impact of the mitigation.
minor comments (2)
  1. [Abstract] The abstract states qualitative agreement but omits any mention of the specific pulse parameters or WER reduction factors that appear in the main text; adding one or two quantitative highlights would improve clarity.
  2. Notation for the effective fields, torque terms, and the definition of the loss-of-determinism regime should be consolidated in a single table or appendix for easier reference.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments on our manuscript. We address each major comment point by point below.

read point-by-point responses

  1. Referee: [Modeling and simulation section] The macrospin model (two coupled LLG equations for free and reference layers) is invoked to identify the loss-of-determinism regime and to guide the STT pulse-shaping strategy. Because the manuscript explicitly attributes the backhopping and asymmetry to device imperfections in top-pinned stacks (which commonly produce spatial variations in anisotropy or pinning that nucleate non-uniform modes), the uniform-magnetization assumption requires explicit justification or cross-validation; without micromagnetic simulations or quantitative WER matching, it is unclear whether the model's regime corresponds to the experimental error mechanisms.

    Authors: We acknowledge that the macrospin model represents a simplification and that device imperfections in top-pinned stacks can induce non-uniform magnetization dynamics. The model is employed to isolate the coupled SOT-STT dynamics and to qualitatively reproduce the experimentally observed asymmetry between AP-to-P and P-to-AP transitions, thereby identifying the loss-of-determinism regime that motivates the pulse-shaping approach. We have added a dedicated paragraph in the revised manuscript that explicitly discusses the limitations of the uniform-magnetization assumption, notes the potential influence of micromagnetic effects, and clarifies that the model serves as a qualitative guide rather than a quantitative predictor. Full micromagnetic validation lies beyond the present scope but remains a valuable direction for future work. revision: partial

  2. Referee: [Experimental results and discussion] The experimental demonstration of reduced WER via pulse shaping is presented as qualitative agreement with the model. Specific numerical WER values (before/after shaping), error bars, exact pulse amplitudes/durations, and statistics across devices are not provided, making it difficult to assess the magnitude, reproducibility, and practical impact of the mitigation.

    Authors: We agree that quantitative details are necessary to properly evaluate the practical impact. In the revised manuscript we have incorporated the specific WER values measured before and after pulse shaping, together with error bars, the precise pulse amplitudes and durations employed, and statistics obtained across multiple devices to demonstrate reproducibility. revision: yes

Circularity Check

0 steps flagged

No significant circularity; model informed by experiment and mitigation validated separately

full rationale

The derivation proceeds from experimental observations of WER asymmetry and backhopping in top-pinned SOT-MRAM devices, to a macrospin model (two coupled LLG equations) that qualitatively reproduces the asymmetry and identifies an intermediate loss-of-determinism regime, to proposed STT pulse-shaping mitigation strategies that are then tested and shown to reduce WER experimentally. No load-bearing step reduces by construction to its inputs: the model is not claimed to be a first-principles derivation but a qualitative tool; the mitigation is not a fitted parameter renamed as prediction but a strategy derived from simulation and independently verified in hardware. No self-citations, ansatzes smuggled via citation, or uniqueness theorems are invoked as load-bearing. The paper is self-contained against external benchmarks (experiment) and receives the default low score.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Based solely on abstract; relies on standard micromagnetic modeling and experimental observations of device behavior. No new free parameters or invented entities are described.

axioms (2)
  • standard math Magnetization dynamics of free and reference layers are governed by two coupled Landau-Lifshitz-Gilbert equations
    Foundation of the macrospin model used to reproduce asymmetry and loss-of-determinism
  • domain assumption Top-pinned devices exhibit STT-induced backhopping and pronounced AP-to-P vs P-to-AP asymmetry under combined torques
    Directly stated as observed in experiments

pith-pipeline@v0.9.0 · 5523 in / 1427 out tokens · 39110 ms · 2026-05-08T10:18:47.318537+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages

  1. [1]

    Ultra-Fast and High-Reliability SOT-MRAM: From Cache Replacement to Normally-Off Computing,

    Lamard, O Boulle, K Garello, J Langer, B Ocker, M -C Cyrille, P Gambardella, M Tahoori, and G Gaudin (2016), “Ultra-Fast and High-Reliability SOT-MRAM: From Cache Replacement to Normally-Off Computing,” IEEE Trans. Multi-Scale Comput. Syst., vol. 2, 1, 49–60, doi: 10.1109/TMSCS.2015.2509963. V D Nguyen, S Rao, K Wostyn, and S Couet (2024), “Recent progres...

    work page doi:10.1109/tmscs.2015.2509963 2016

  2. [2]

    Basic principles of STT-MRAM cell operation in memory arrays,

    Driskill-Smith, W H Butler, P B Visscher, D Lottis, E Chen, V Nikitin and M Kroumbi (2013), “Basic principles of STT-MRAM cell operation in memory arrays,” J. Phys. D: Appl. Phys., vol. 46, 7, 074001, doi: 10.1088/0022-3727/46/7/074001. T Yang, J Gao, Y Cui, J Xu, J Li, W Wang, J Luo, M Yang (2021), “Thermal stability of SOT-MTJ thin films tuning by multi...

    work page doi:10.1088/0022-3727/46/7/074001 2013

  3. [3]

    Complementary Field-Effect Transistor (CFET) Demonstration at 48nm Gate Pitch for Future Logic Technology Scaling,

    Fukuda, T Koike, M Saitoh, J Ito, and K Koi (2023), “14nm High-Performance MTJ with Accelerated STT -Switching and High -Retention Doped Co -Pt Alloy Storage Layer for 1Znm MRAM,” in 2023 IEDM , 1–4. doi: 10.1109/IEDM45741.2023.10413856. E Liu, J Swerts, T Devolder, S Couet, S Mertens, T Lin, V Spampinato, A Franquet, T

    work page doi:10.1109/iedm45741.2023.10413856 2023

  4. [4]

    Seed Layer Impact on Structural and Magnetic Properties of [Co/Ni] Multilayers with Perpendicular Magnetic Anisotropy,

    Conard, S Van Elshocht, A Furnemont, J De Boeck, and G Kar (2017), “Seed Layer Impact on Structural and Magnetic Properties of [Co/Ni] Multilayers with Perpendicular Magnetic Anisotropy,” J. Appl. Phys. , vol. 121, 4, doi: 10.1063/1.4974885. Y Tomczak, J Swerts, S Mertens, T Lin, S Couet, E Liu, K Sankaran, G Pourtois, W Kim, L Souriau, S Van Elshocht, G ...

    work page doi:10.1063/1.4974885 2017