In magnetic tunnel junctions (MTJs), inelastic electron-magnon (e-m) scattering shapes the measured inelastic electron tunneling spectra (IETS): experiments show a zero-bias anomaly (ZBA) with two peaks of the second derivative symmetric about zero bias and a dip in the differential conductance at , while elastic theories of the spin-transfer torque (STT) deviate from experiment at high bias. Mahfouzi and Nikolić [Phys. Rev. B 90, 045115 (2014)] developed a conserving, self-consistent nonequilibrium Green function many-body perturbation treatment of coupled e-m transport, but could only solve it in quasi-one-dimensional models with a few transverse sites: the Hartree diagram was dropped, the electron-hole polarization bubble was excluded from the widest calculation, the IETS peaks shifted away from zero bias, the zero-bias dip in was not reproduced at all, and the bias dependence of the STT parameters was not computed. The open problem is whether the framework itself — extended to realistic multi-channel (three-dimensional, transverse--sampled) junctions with the conserving diagram set (self-consistent Fock electronic self-energy with fully dressed lines, bubble-dressed magnon Green function, and the previously omitted Hartree term either stably included or justifiably neglected) — can reproduce the essential features of these measured spectra.
Background
A magnetic tunnel junction (MTJ) is a ferromagnet/insulator/ferromagnet (F/I/F) stack whose resistance depends on the relative orientation of the two magnetizations; the effect is called tunneling magnetoresistance (TMR). Inelastic electron tunneling spectroscopy (IETS) measures the second derivative of the current with respect to bias voltage : inelastic channels open at thresholds set by excitation energies, producing characteristic peaks. In CoFeB/MgO and AlO-based MTJs, IETS shows a zero-bias anomaly (ZBA): peaks of opposite sign at mV attributed to magnons (quantized spin-wave excitations of the ferromagnetic electrodes), together with a dip in the differential conductance at , and TMR decreases with increasing temperature (V. Drewello et al., Phys. Rev. B 77, 014440 (2008); A. A. Khan et al., Phys. Rev. B 82, 064416 (2010)). Separately, the spin-transfer torque (STT) — the torque exerted by the spin-polarized tunneling current on the magnetization, parameterized by field-like and damping-like coefficients and — deviates from purely elastic phase-coherent theories at high bias, especially for (H. Kubota et al., Nat. Phys. 4, 37 (2008); C. Wang et al., Nat. Phys. 7, 496 (2011)), and inelastic scattering has been invoked to explain the discrepancy. An early alternative theory attributed the ZBA to inelastic tunneling via two-level impurities in the barrier (L. Sheng, D. Y. Xing, and D. N. Sheng, Phys. Rev. B 70, 094416 (2004)), so the microscopic origin of the ZBA features has competing candidate mechanisms.
Mahfouzi and Nikolić (Phys. Rev. B 90, 045115 (2014)) formulated a nonequilibrium many-body perturbation theory of the coupled e-m system, based on the nonequilibrium Green function (NEGF) formalism on the Keldysh contour, with the electron and magnon Green functions and self-energies handled as matrices in Keldysh space. The localized spins were mapped to magnons via the Holstein-Primakoff transformation, the e-m coupling arises from the local - exchange interaction (the contact exchange coupling between itinerant electron spins and the localized magnetic moments), and the electronic subsystem was a tight-binding lattice modeling F/I/F MTJs. To conserve charge current at finite bias they used a self-consistent Born approximation (SCBA) — resumming the Fock diagram for the electronic self-energy with fully dressed Green function lines, mutually self-consistent with the electron-hole polarization bubble diagram for the magnon self-energy (analogous to self-consistent GW for electron-electron interaction). Their key physical finding was that the bubble diagram substantially broadens the magnon density of states to low energies, enabling inelastic e-m scattering even at zero bias and producing a two-peak structure in the partial (elastic or inelastic) IETS around .
However, the solved models were quasi-one-dimensional: a strictly 1D active region ( sites) produced an overly complex total IETS (interference between few spin-resolved conducting channels), and the widest calculation ( sites) recovered the simple two-peak structure but only after dropping the bubble diagram for computational expense, so the peaks shifted away from . Three limitations were explicitly flagged: (1) the Hartree diagram was omitted because its direct evaluation is numerically unstable when the junction is invariant under spin rotations about the magnetization axis (the spin-flip rate in the Hartree term can acquire an arbitrary phase); (2) the experimentally observed dip of at was not reproduced in any of their calculations, plausibly due to the small number of transverse channels; (3) extending the approach to fully 3D MTJs requires transverse -point sampling, turning the self-consistent coupled nonlinear integral equations into prohibitively expensive five-dimensional integrals, so carefully crafted approximations are needed. They also did not compute the bias dependence of the STT parameters and (the announced follow-up never appeared in peer-reviewed form), and the TMR temperature dependence was shown only in the linear-response limit .
Problem Statement
Extend the conserving, self-consistent nonequilibrium many-body perturbation theory of the coupled electron-magnon system in ferromagnet/insulator/ferromagnet magnetic tunnel junctions — electronic Fock self-energy with fully dressed Green function lines, mutually self-consistent with the electron-hole polarization bubble self-energy for the magnons — to a realistic multi-channel junction that stands in for measured 3D MTJs (active region whose number of transverse sites, or transverse momentum mesh density, is converged in the sense that further enlargement leaves the computed spectra qualitatively unchanged, with suitable controlled approximations to keep the five-dimensional self-consistency integrals tractable), and settle the status of the previously omitted Hartree diagram — either by including it in a numerically stable, charge-conserving way, or by establishing within the same framework that its effect on the observables below is negligible, as is customarily assumed for the analogous Hartree term in electron-phonon SCBA. Within that framework, determine whether inelastic electron-magnon scattering alone reproduces the essential features of the measured inelastic magnetotransport signatures of MgO- and AlO-based MTJs, namely: (a) the zero-bias anomaly in the total IETS — the experimentally measured — a two-peak structure centered at on the meV scale of the observed ZBA, together with the dip in at , i.e., rising in absolute value as increases away from the zero-bias minimum; (b) the TMR ratio as a function of both temperature and finite bias — within the temperature range in which the low-magnon-occupation Holstein-Primakoff approximation underlying the framework remains valid () — including its interplay with the ZBA; and (c) the bias dependence of the spin-transfer torque parameters and , where purely elastic theories are known to deviate from experiment at high bias. All three sub-targets (a), (b) and (c) are required, as are the multi-channel extension and the settlement of the Hartree diagram (by either of the two options stated above), and each sub-target is resolved by whether the extended framework, with its converged active region, reproduces the qualitative feature it names — for (a) the two zero-bias IETS peaks together with the zero-bias dip in , for (b) the temperature- and finite-bias dependence of the TMR ratio together with its interplay with the ZBA, and for (c) a bias-dependent and — the agreement being qualitative reproduction of these enumerated features rather than a quantitative fit to a single measured curve. A definite demonstration, obtained within the same converged framework, that inelastic electron-magnon scattering alone cannot reproduce one or more of these features resolves the corresponding sub-target equally; partial reproduction of a sub-target, or a conclusion resting on a non-converged or uncontrolled calculation, does not.
References
- F. Mahfouzi and B. K. Nikolić, "Signatures of electron-magnon interaction in charge and spin currents through magnetic tunnel junctions: A nonequilibrium many-body perturbation theory approach", Phys. Rev. B 90, 045115 (2014). DOI: 10.1103/PhysRevB.90.045115; arXiv:1310.8551. https://doi.org/10.1103/PhysRevB.90.045115
- V. Drewello, J. Schmalhorst, A. Thomas, and G. Reiss, "Evidence for strong magnon contribution to the TMR temperature dependence in MgO based tunnel junctions", Phys. Rev. B 77, 014440 (2008). DOI: 10.1103/physrevb.77.014440. https://doi.org/10.1103/PhysRevB.77.014440
- A. A. Khan, J. Schmalhorst, G. Reiss, G. Eilers, M. Münzenberg, H. Schuhmann, and M. Seibt, "Elastic and inelastic conductance in Co-Fe-B/MgO/Co-Fe-B magnetic tunnel junctions", Phys. Rev. B 82, 064416 (2010). DOI: 10.1103/physrevb.82.064416. https://doi.org/10.1103/PhysRevB.82.064416
- L. Sheng, D. Y. Xing, and D. N. Sheng, "Theory of the zero-bias anomaly in magnetic tunnel junctions: Inelastic tunneling via impurities", Phys. Rev. B 70, 094416 (2004). DOI: 10.1103/physrevb.70.094416. https://doi.org/10.1103/PhysRevB.70.094416
- H. Kubota, A. Fukushima, K. Yakushiji, T. Nagahama, S. Yuasa, K. Ando, H. Maehara, Y. Nagamine, K. Tsunekawa, D. D. Djayaprawira, N. Watanabe, and Y. Suzuki, "Quantitative measurement of voltage dependence of spin-transfer torque in MgO-based magnetic tunnel junctions", Nat. Phys. 4, 37 (2008). DOI: 10.1038/nphys784. https://doi.org/10.1038/nphys784
- C. Wang, Y.-T. Cui, J. A. Katine, R. A. Buhrman, and D. C. Ralph, "Time-resolved measurement of spin-transfer driven ferromagnetic resonance and spin torque in magnetic tunnel junctions", Nat. Phys. 7, 496 (2011). DOI: 10.1038/nphys1928. https://doi.org/10.1038/nphys1928
- S. A. Bender, R. A. Duine, and Y. Tserkovnyak, "Quantum-kinetic theory of spin-transfer torque and magnon-assisted transport in nanoscale magnetic junctions", Phys. Rev. B 99, 024434 (2019). DOI: 10.1103/PhysRevB.99.024434. https://doi.org/10.1103/PhysRevB.99.024434
- A. Suresh, U. Bajpai, and B. K. Nikolić, "Magnon-driven chiral charge and spin pumping and electron-magnon scattering from time-dependent quantum transport combined with classical atomistic spin dynamics", Phys. Rev. B 101, 214412 (2020). DOI: 10.1103/physrevb.101.214412. https://doi.org/10.1103/PhysRevB.101.214412
- U. Bajpai, A. Suresh, and B. K. Nikolić, "Quantum many-body states and Green's functions of nonequilibrium electron-magnon systems: Localized spin operators versus their mapping to Holstein-Primakoff bosons", Phys. Rev. B 104, 184425 (2021). DOI: 10.1103/PhysRevB.104.184425. https://doi.org/10.1103/PhysRevB.104.184425
- D. Karlsson, R. van Leeuwen, Y. Pavlyukh, E. Perfetto, and G. Stefanucci, "Fast Green's Function Method for Ultrafast Electron-Boson Dynamics", Phys. Rev. Lett. 127, 036402 (2021). DOI: 10.1103/physrevlett.127.036402. https://doi.org/10.1103/PhysRevLett.127.036402
- B. K. Nikolić, K. Dolui, M. D. Petrović, P. Plecháč, T. Markussen, and K. Stokbro, "First-Principles Quantum Transport Modeling of Spin-Transfer and Spin-Orbit Torques in Magnetic Multilayers", Handbook of Materials Modeling (Springer, 2018). DOI: 10.1007/978-3-319-50257-1_112-1. https://doi.org/10.1007/978-3-319-50257-1_112-1
- R. E. Troncoso, A. Brataas, and R. A. Duine, "Many-body theory of spin-current driven instabilities in magnetic insulators", Phys. Rev. B 99, 104426 (2019). DOI: 10.1103/PhysRevB.99.104426. https://doi.org/10.1103/PhysRevB.99.104426