Microscopic foundation and experimental validation of the effective-circuit theory for the cusplike zero-bias anomaly in tunneling magnetoresistance

You, Sheng, and Xing (Phys. Rev. B 81, 052402, 2010) explained the cusplike zero-bias anomaly (ZBA) of tunneling magnetoresistance (TMR) by an effective circuit theory: impurity-mediated tunneling coupled to an environmental LC ladder through a capacitance $C_0$ suppresses the indirect conductance as a power law in $|V|$ with exponent $g$ (the dimensionless environmental hardness), producing a cusp for $g<1$. The theory is phenomenological — $g$, $C_0$, $L$, and $C$ are fit parameters. The framework has remained undeveloped relative to adjacent proposals for the zero-bias anomaly (magnon-assisted tunneling, Kondo/cotunneling, and interface spin-flip scattering): no later work is known that derives the effective circuit parameters from microscopic material properties, delimits the $Z(\omega)\approx Z(0)$ approximation, incorporates junction band structure or resonant impurity states, extends the theory quantitatively to double-barrier junctions, or tests whether $g<1$ correlates with observed cusps across material systems. The problem of giving the effective-circuit theory of the cusplike TMR zero-bias anomaly a microscopic and experimental foundation remains open.

Background

A magnetic tunnel junction (MTJ) consists of two ferromagnetic electrodes separated by a thin insulating barrier such as alumina or MgO. Its tunneling magnetoresistance (TMR) — the change of junction resistance between the parallel (P) and antiparallel (AP) alignments of the electrode magnetizations — is the operating principle of hard-disk read heads, magnetic field sensors, and magnetic random-access memory. In most MTJs the differential conductance, and often the TMR itself, is strongly suppressed in a narrow window around zero bias voltage at low temperature. This is the zero-bias anomaly (ZBA). In many junctions the suppression is not smooth but cusplike: the slope of the conductance is discontinuous at $V=0$, behavior that corresponds to a power law $G(V)-G(0)\sim|V|^{p}$ with $p\leq1$; this cusplike regime comprises a strict cusp with divergent slope when $p<1$ and a marginal linear kink with a finite jump of the slope at $p=1$, whereas standard inelastic-tunneling theories typically give $p>1$, for which the slope is continuous and there is no cusp.

You, Sheng, and Xing proposed a phenomenological 'effective circuit' theory for this cusplike ZBA (Phys. Rev. B 81, 052402 (2010)). The junction conductance is split into a direct tunneling channel and an indirect channel through impurities in the barrier; the bare impurity-channel conductance is written in closed two-spin-channel circuit form, with spin-flip scattering between the majority- and minority-spin channels represented by a connecting resistor. The impurity channel is coupled to the environmental degrees of freedom of the junction (phonons, magnons, electromagnetic modes), which are lumped into an infinite ladder of inductors (inductance $L$) and capacitors (capacitance $C$) coupled to the tunneling spot through a capacitance $C_0$. This is in close analogy with the dynamical Coulomb blockade framework (also called $P(E)$ theory) used for single-electron and molecular tunnel junctions, in which low-energy tunneling is suppressed because the tunneling electron must polarize an electromagnetic environment of impedance $Z(\omega)$. The indirect differential conductance is a convolution of the bare channel with the environmental excitation spectrum, and, using the low-frequency approximation $Z(\omega)\approx Z(0)=\sqrt{L/C}$ justified in the paper by the dominance of small $\omega$ in the emission integral, it reduces at small bias to a power law $G_{I}(V)=c\,|V/V_{0}|^{g}\,G_{I0}$ with a dimensionless 'environmental hardness' $g$ that measures the strength of coupling to the collective modes. A strict cusp of the indirect conductance appears for $g<1$ — the slope of the indirect conductance then diverges at $V=0$, the marginal $g=1$ case being a slope kink rather than a strict cusp — while the TMR-versus-bias curve is cusplike for $g\leq1$; with a small set of parameters (a representative fit to the low-temperature TMR data of Mattana et al. on a GaMnAs/AlAs-based double-barrier magnetic tunnel structure used $g=0.8$ and $C_0=0.06$ fF) the model reproduces the cusplike TMR-versus-bias curves of experiments.

The same authors had earlier given a microscopic counterpart in one specific model: inelastic tunneling via magnetic impurities with phonon emission, treated by the closed-time path-integral technique (a nonperturbative real-time method for dissipative quantum transport), which yields an infrared-singular emission spectrum and $G(V)-G(0)\sim|V|^{2g_{\mathrm{ph}}}$ with a slope discontinuity for $g_{\mathrm{ph}}\leq0.5$ (Phys. Rev. B 70, 094416 (2004)); they also treated impurity resonant TMR (Phys. Rev. B 69, 132414 (2004)). In the 2010 circuit theory, however, the parameters $g$, $C_0$, $L$, and $C$ are not derived from any microscopic model of a particular junction: they are fit parameters, the analytic results rely on the uncontrolled approximation $Z(\omega)\approx Z(0)$, junction-specific electronic structure (for example the complex band structure of MgO and narrow interface states) and resonant tunneling through localized states are excluded, and double-barrier geometries are only suggested. The unresolved target is therefore whether this phenomenology can be grounded in the microscopic physics of real junction materials and validated experimentally — in particular whether the predicted criterion $g<1$ for a cusp is corroborated when the parameters are determined independently of the fits.

Problem Statement

Give the effective-circuit theory of the cusplike zero-bias anomaly in tunneling magnetoresistance a microscopic and experimental foundation. Concretely, with the circuit model defined as in the source theory (impurity-mediated tunneling coupled through a capacitance $C_0$ to an environmental $LC$ ladder of inductance $L$ and capacitance $C$, giving a low-bias indirect conductance $G_I(V)\propto|V/V_0|^{g}\,G_{I0}$, a cusplike conductance anomaly for $g<1$ (a strict cusp; the marginal $g=1$ case is a slope kink), and a cusplike TMR peak for $g\leq1$):

The following directions trace the limitations stated by the source paper itself, which presents an explicitly phenomenological model: $g$ enters as an adjustable 'material-dependent quantity', $C_0$ and the lumped $LC$ environment are not derived from any microscopic model, the analytic power law relies on the approximation $Z(\omega)\approx Z(0)=\sqrt{L/C}$ justified only by the dominance of small $\omega$, the complex band structure of MgO, narrow interface states, and resonant tunneling are explicitly excluded from the simplified model, and applicability to double-barrier junctions is only suggested. A full resolution addresses all five directions within one consistent microscopic-and-experimental framework; each direction is also independently meaningful, so a result that fully settles a single direction counts as progress on that direction while leaving the others open. Direction 2 is not contained in direction 1: deriving the effective (lossless) $Z(\omega)$ does not by itself bound the error of the $Z(\omega)\approx Z(0)$ replacement, nor does it treat environments with dissipative losses beyond the ideal $LC$ ladder.

1. Starting from a microscopic description of a concrete junction material system that comprises the barrier impurity distribution, phonon and magnon spectra, dielectric screening, and electrode/barrier electronic structure, derive the effective circuit parameters: the environmental hardness $g$, the coupling capacitance $C_0$, and the effective environmental impedance $Z(\omega)$ (equivalently the lumped $LC$ spectrum), so that these become predictions of material properties rather than fit parameters.
2. Delimit the validity of the low-frequency approximation $Z(\omega)\approx Z(0)=\sqrt{L/C}$ used to obtain the analytic power law, in particular for realistic frequency-dependent environments including dissipative losses beyond the ideal $LC$ ladder.
3. Within the same microscopic framework, determine when junction-specific electronic structure — such as the complex band structure of MgO, narrow interface states, and resonant tunneling through impurity- or interface-localized states — qualitatively modifies the low-bias exponent or produces additional bias-dependent structure.
4. Extend the theory to more complex geometries, in particular double-barrier junctions, quantitatively.
5. Test the theory experimentally beyond the source paper's single illustrative fit to the Mattana et al. data: determine the circuit parameters independently of the TMR fits and check the source's cusp criteria — a strict cusp of the indirect conductance for $g<1$ (with a slope kink at the marginal $g=1$) and a cusplike TMR peak for $g\leq1$ — against measured conductance and TMR-versus-bias curves.

References

1. D. You, L. Sheng, D. Y. Xing, 'Effective circuit theory for the cusplike zero-bias anomaly in tunneling magnetoresistance', Physical Review B 81, 052402 (2010). DOI: 10.1103/PhysRevB.81.052402, https://doi.org/10.1103/PhysRevB.81.052402
2. L. Sheng, D. Y. Xing, D. N. Sheng, 'Theory of the zero-bias anomaly in magnetic tunnel junctions: Inelastic tunneling via impurities', Physical Review B 70, 094416 (2004). DOI: 10.1103/PhysRevB.70.094416; arXiv:cond-mat/0408552, https://arxiv.org/abs/cond-mat/0408552
3. L. Sheng, D. Y. Xing, D. N. Sheng, 'Theory of impurity resonant tunnel magnetoresistance', Physical Review B 69, 132414 (2004). DOI: 10.1103/PhysRevB.69.132414, https://doi.org/10.1103/PhysRevB.69.132414
4. R. Mattana, J.-M. George, H. Jaffrès, F. Nguyen Van Dau, A. Fert, B. Lépine, A. Guivarc'h, G. Jézéquel, 'Electrical detection of spin accumulation in a p-type GaAs quantum well', Physical Review Letters 90, 166601 (2003). DOI: 10.1103/PhysRevLett.90.166601; arXiv:cond-mat/0211409, https://arxiv.org/abs/cond-mat/0211409
5. H.-X. Wei, Q.-H. Qin, Q.-L. Ma, X.-G. Zhang, X.-F. Han, 'Inelastic electron tunneling spectrum from surface magnon and magnetic impurity scatterings in magnetic tunnel junctions', Physical Review B 82, 134436 (2010). DOI: 10.1103/PhysRevB.82.134436, https://doi.org/10.1103/PhysRevB.82.134436
6. H. Yang, S.-H. Yang, G. Ilnicki, J. Martinek, S. S. P. Parkin, 'Zero-Bias Anomaly in Magnetic Tunnel Junctions', Acta Physica Polonica A 118, 316 (2010). DOI: 10.12693/aphyspola.118.316, https://doi.org/10.12693/aphyspola.118.316
7. J. Z. Sun, P. L. Trouilloud, G. P. Lauer, P. Hashemi, 'Bias dependent conductance in CoFeB-MgO-CoFeB magnetic tunnel junctions as an indicator for electrode magnetic condition at barrier interfaces', AIP Advances 9, 015034 (2019). DOI: 10.1063/1.5058265, https://doi.org/10.1063/1.5058265