Single-shot time-resolved measurements of spin-transfer-torque switching in 100×300 nm2100\times300\ \mathrm{nm}^2 in-plane CoFeB/MgO/CoFeB magnetic tunnel junctions (Devolder et al., 2008) showed that reversal proceeds by a fluctuating, electrically quiet nanosecond incubation delay followed by a reproducible 0.4 ns0.4\ \mathrm{ns} transition and damped ringing, and argued against FMR-pumping, C-bend, and odd-spin-wave triggering in favor of inhomogeneous tip nucleation, while deferring the quantitative micromagnetic modeling. The problem is to give this picture a predictive microscopic basis: identify the initiation mechanism and micromagnetic nucleation-to-propagation pathway of the incubation delay, determine the origin of the small offset couplings (the hard-axis astroid shift and the easy-axis dipolar coupling) and of the extrinsic pinned-layer linewidth, and model the full switching-time statistics as a function of device geometry, pulse amplitude, and temperature.

Background

Spin-transfer torque (STT) writing uses a spin-polarized current to reverse a nanomagnet: in a magnetic tunnel junction (MTJ) a 'free' ferromagnetic layer is separated by an MgO tunnel barrier from a 'pinned' reference layer, and a current pulse whose spin polarization is fixed by the pinned layer exerts a torque that can switch the free layer between the parallel (P, low-resistance) and antiparallel (AP, high-resistance) states, the write operation of spin-transfer magnetic random access memory (STT-MRAM). In the devices studied here the free layer is a 2.5 nm2.5\ \mathrm{nm} CoFeB film patterned into a 100×300 nm2100\times300\ \mathrm{nm}^2 rectangle with in-plane anisotropy field Hk≈8 mTH_k\approx 8\ \mathrm{mT}, and the pinned electrode is a synthetic antiferromagnet (two antiferromagnetically coupled ferromagnetic layers that largely cancel each other's stray field).

Earlier time-resolved studies of STT switching used stroboscopic, ensemble-averaged readout, which recovers mean switching paths but hides event-to-event variability; the experiment by Devolder et al. (2008) was among the first to record high-bandwidth (13 GHz13\ \mathrm{GHz}) single-shot resistance traces R(t)R(t) of individual switching events in MgO MTJs. Each P→\toAP event driven by a near-breakdown negative voltage pulse begins with an 'incubation delay' of order nanoseconds during which the resistance shows no detectable pre-switching dynamics; this delay fluctuates strongly from trial to trial (switching instants spread from 00 to beyond 10 ns10\ \mathrm{ns}, with mean 5.3 ns5.3\ \mathrm{ns} at Vpulse=−1.0 VV_{\text{pulse}}=-1.0\ \mathrm{V} and 2.5 ns2.5\ \mathrm{ns} at −1.1 V-1.1\ \mathrm{V}), whereas the subsequent transition (duration ≈0.4 ns\approx 0.4\ \mathrm{ns}) and the damped post-switching 'ringing' (1.4±0.2 GHz1.4\pm0.2\ \mathrm{GHz}, decay time 1.5±0.5 ns1.5\pm0.5\ \mathrm{ns}) are reproducible once the delay is passed. The ringing decay is consistent with the damping αEFF=0.006±0.001\alpha_{\text{EFF}}=0.006\pm0.001 extracted from the linewidth of the free-layer eigenmode in voltage-noise spectra using a macrospin relation (the macrospin, or single-domain, approximation treats the magnetization as spatially uniform), implying that the dynamics after the transition are spatially coherent. The absence of any pre-switching oscillation was used to argue against triggering by resonant pumping of the uniform ferromagnetic-resonance (FMR) mode, by beating of a symmetric 'C-bend' mode (a nonuniform eigenmode in which the magnetization bows symmetrically along the element's long axis), or by spin-wave modes with an odd number of nodal lines whose local precessions would cancel in the spatially averaged resistance. Instead the authors proposed an inhomogeneous nucleation picture: localized tilting of the magnetization at the tips or corners of the free layer under spin torque and thermal fluctuations, with the resulting local change of conductance redistributing the current and feeding back on the magnetization until the disturbance propagates inward and triggers the rapid, quasi-coherent reversal of the whole element. They explicitly deferred the extensive micromagnetic simulations needed to test this picture, to quantify how thermal sampling of the initial micromagnetic configurations sets the incubation-time distribution, and to predict the switching statistics as a function of device geometry, pulse amplitude, and temperature.

Two device-characterization issues were left open alongside this main question. The quasistatic hysteresis loops are off-centered by a coupling field of about 2.8 mT2.8\ \mathrm{mT} favoring the P state, attributed to dipolar coupling with the uncompensated synthetic antiferromagnet of the pinned stack; the measurements additionally show a small hard-axis offset of the switching 'astroid' — the Stoner–Wohlfarth switching boundary plotted in the plane of easy- and hard-axis applied fields, whose displaced center signals an internal bias field — of about −0.7 mT-0.7\ \mathrm{mT}, whose microscopic origin was not identified. Second, the voltage-noise spectra show two eigenmodes, one attributed to the free layer and one to the pinned layer; the pinned-layer linewidth is broad due to extrinsic contributions (inhomogeneous broadening, two-magnon scattering, or similar mechanisms), and the extrinsic versus intrinsic decomposition and its consequence for the switching dynamics were not established.

Problem Statement

Establish a quantitative, predictive microscopic account of spin-transfer-torque switching in in-plane CoFeB/MgO/CoFeB magnetic tunnel junctions of the type measured by Devolder et al. (2008), replacing their qualitative interpretation of single-shot data with validated theory and simulation. Specifically, with one common objective — a model that reproduces the observed incubation-delay and total-switching-time distributions as functions of device geometry, pulse amplitude, and temperature — determine: (a) the microscopic initiation mechanism of the incubation delay and the micromagnetic pathway of the nucleation-to-propagation event, by simulations that test the proposed inhomogeneous nucleation at the tips of the free layer, quantify how thermal variations of the initial micromagnetic configurations set the incubation-time distribution, and include the feedback between local conductivity/current redistribution and the magnetization evolution; (b) the microscopic origin of the small offset couplings in this device class — the hard-axis offset of the switching astroid (reported as about −0.7 mT-0.7\ \mathrm{mT} (uncertain) in the Devolder et al. (2008) measurements, alongside a 2.8 mT2.8\ \mathrm{mT} easy-axis loop offset attributed to dipolar coupling with the uncompensated synthetic-antiferromagnet pinned stack) — and its quantitative effect on switching speed and variability; and (c) the extrinsic versus intrinsic contributions to the pinned-layer eigenmode linewidth and their effect on the switching dynamics. The account must discriminate among the candidate triggering mechanisms (FMR-mode pumping, symmetric C-bend mode beating, odd spin-wave modes, inhomogeneous tip nucleation) on mechanistic grounds rather than from the absence of a pre-switching electrical signal alone, and must reproduce the full single-shot switching-time statistics and not only their means.

References

  1. T. Devolder, J. Hayakawa, K. Ito, H. Takahashi, S. Ikeda, P. Crozat, N. Zerounian, Joo-Von Kim, C. Chappert, and H. Ohno, 'Single-Shot Time-Resolved Measurements of Nanosecond-Scale Spin-Transfer Induced Switching: Stochastic Versus Deterministic Aspects', Phys. Rev. Lett. 100, 057206 (2008). DOI: 10.1103/PhysRevLett.100.057206, https://doi.org/10.1103/PhysRevLett.100.057206
  2. R. H. Koch, J. A. Katine, and J. Z. Sun, 'Time-Resolved Reversal of Spin-Transfer Switching in a Nanomagnet', Phys. Rev. Lett. 92, 088302 (2004). DOI: 10.1103/PhysRevLett.92.088302, https://doi.org/10.1103/PhysRevLett.92.088302
  3. T. Devolder, C. Chappert, J. A. Katine, M. J. Carey, and K. Ito, 'Distribution of the magnetization reversal duration in subnanosecond spin-transfer switching', Phys. Rev. B 75, 064402 (2007). DOI: 10.1103/PhysRevB.75.064402, https://doi.org/10.1103/PhysRevB.75.064402
  4. H. Tomita, K. Konishi, T. Nozaki, H. Kubota, A. Fukushima, K. Yakushiji, S. Yuasa, Y. Nakatani, T. Shinjo, M. Shiraishi, and Y. Suzuki, 'Single-Shot Measurements of Spin-Transfer Switching in CoFeB/MgO/CoFeB Magnetic Tunnel Junctions', Appl. Phys. Express 1, 061303 (2008). DOI: 10.1143/APEX.1.061303, https://doi.org/10.1143/APEX.1.061303
  5. Y.-T. Cui, G. Finocchio, C. Wang, J. A. Katine, R. A. Buhrman, and D. C. Ralph, 'Single-Shot Time-Domain Studies of Spin-Torque-Driven Switching in Magnetic Tunnel Junctions', Phys. Rev. Lett. 104, 097201 (2010); preprint arXiv:0906.2476. DOI: 10.1103/PhysRevLett.104.097201, https://doi.org/10.1103/PhysRevLett.104.097201 (preprint: https://arxiv.org/abs/0906.2476)
  6. T. Devolder, L. Bianchini, K. Miura, K. Ito, Joo-Von Kim, P. Crozat, V. Morin, A. Helmer, C. Chappert, S. Ikeda, and H. Ohno, 'Spin-torque switching window, thermal stability, and material parameters of MgO tunnel junctions', Appl. Phys. Lett. 98, 162502 (2011). DOI: 10.1063/1.3576937, https://doi.org/10.1063/1.3576937
  7. R. Matsumoto, H. Arai, S. Yuasa, and H. Imamura, 'Theoretical analysis of thermally activated spin-transfer-torque switching in a conically magnetized nanomagnet', Phys. Rev. B 92, 140409 (2015). DOI: 10.1103/PhysRevB.92.140409, https://doi.org/10.1103/PhysRevB.92.140409
  8. T. Pramanik, U. Roy, P. Jadaun, L. F. Register, and S. K. Banerjee, 'Write error rates of in-plane spin-transfer-torque random access memory calculated from rare-event enhanced micromagnetic simulations', J. Magn. Magn. Mater. 467, 96–107 (2018). DOI: 10.1016/j.jmmm.2018.07.042, https://doi.org/10.1016/j.jmmm.2018.07.042
  9. P. Bouquin, Joo-Von Kim, O. Bultynck, S. Rao, S. Couet, G. S. Kar, and T. Devolder, 'Stochastic Processes in Magnetization Reversal Involving Domain-Wall Motion in Magnetic Memory Elements', Phys. Rev. Applied 15, 024037 (2021); preprint arXiv:2009.07492. DOI: 10.1103/PhysRevApplied.15.024037, https://doi.org/10.1103/PhysRevApplied.15.024037 (preprint: https://arxiv.org/abs/2009.07492)
  10. E. Grimaldi, V. Krizakova, G. Sala, F. Yasin, S. Couet, G. S. Kar, K. Garello, and P. Gambardella, 'Single-shot dynamics of spin–orbit torque and spin transfer torque switching in three-terminal magnetic tunnel junctions', Nat. Nanotechnol. 15, 111–117 (2020). DOI: 10.1038/s41565-019-0607-7, https://doi.org/10.1038/s41565-019-0607-7
  11. G. Sala, J. Meyer, A. Flechsig, L. Gabriel, and P. Gambardella, 'Deterministic and stochastic aspects of current-induced magnetization reversal in perpendicular nanomagnets', Phys. Rev. B 107, 214447 (2023). DOI: 10.1103/PhysRevB.107.214447, https://doi.org/10.1103/PhysRevB.107.214447
  12. F. Xu, L. Zhou, W. Cai, K. Xu, K. Cao, K. Shi, and W. Zhao, 'Time-resolved nanosecond magnetization switching induced by spin-transfer torque for cryogenic memories', Appl. Phys. Lett. 126, 232405 (2025). DOI: 10.1063/5.0266924, https://doi.org/10.1063/5.0266924