An arXiv preprint reports that a passive spin-torque microwave detector could identify one setting of a repeating positive rectangular current pulse from a single time-averaged voltage, as long as the pulse’s other two settings are known. The result appears only in the modeled linear regime: in a separate nonlinear regime, the simulated voltage can jump or drop, undermining an unambiguous readout.
The work is a preprint based on numerical and simplified theoretical analysis. It is identified as arXiv version 1 in the physics.app-ph category and carries the arXiv date 20 August 2026. The authors examine whether a passive spin-torque microwave detector, or STMD, can detect the pulses.
A detector tested by three pulse settings
At the center of the calculation is a circular three-layer magnetic-tunnel junction, or MTJ. It has a free magnetic layer and a pinned magnetic layer separated by a non-magnetic dielectric spacer; the pinned layer’s magnetization is fixed along the x direction. The detector response is calculated with the Landau-Lifshitz-Gilbert-Slonczewski equation in the macrospin approximation, a simplified model of the device’s magnetization dynamics.
The input is a periodic positive rectangular current, written as I(t). The sweeps vary three quantities: amplitude I0, pulse duration τ and repetition period T. The tested amplitude runs from just above 0 to 0.7 mA, duration from 0.1T to 0.875T, and repetition period from 0.5 to 4 ns.
To calculate the detector’s reported output, the analysis averages the instantaneous voltage over an integer number of pulse periods after transient behavior has been excluded. The averaging begins at t0 = 10T, so the central quantity is Udc, the time-averaged voltage, rather than the voltage at one instant.
The parameter set is deliberately specific. The modeled free layer has a radius of 50 nm and a thickness of 1 nm; the calculation uses a spin-polarization efficiency of 0.7, a resistance-area product of 7.854 Ω·µm², a perpendicular-state resistance of 1 kΩ, a Gilbert damping parameter of 0.01, normalized saturation magnetization of 800 mT, and a 200 mT bias field.
Where the response stays orderly
In the in-plane, or IP, configuration, the averaged voltage follows a linear relationship with pulse amplitude for any tested pulse duration. The reported approximation is Udc ≈ I0 R0 (τ/T). In ordinary terms, the modeled output tracks both how strong the pulse is and what fraction of each repetition period it occupies.
That relationship is the study’s clearest route to parameter identification. If two of the three pulse settings are known, one measured Udc can be used to identify the remaining I0, τ or T in linear operation. The conclusion is therefore conditional: the calculation describes a one-unknown-at-a-time readout, not a claim that one voltage independently reveals every pulse setting.
The clean response is not reported for every magnetization configuration and pulse setting. For out-of-plane, or OOP, dynamics at the largest tested repetition period, T = 4 ns, Udc as a function of I0 is described as precisely linear, with q approximately 1.02.
When the curve starts to jump
At the shorter OOP period of T = 2 ns, the curve behaves differently. It is nearly linear at small amplitudes, then becomes nonlinear once I0 reaches the reported threshold Ith. In one displayed curve, with τ = 1.75 ns, that threshold is approximately 0.29 mA; beyond it, the voltage shows jumps or drops.
The paper links the linear regime to unambiguous determination and the nonlinear OOP regime to voltage jumps and drops that prevent unambiguous determination.
The authors group the results into two operating regimes: a linear regime and a nonlinear regime. They interpret the IP configuration as preferable for pulse detection, while identifying the linear relationship among Udc, I0 and τ/T as a limitation.
The simulations do not settle why the OOP curve departs from linearity. The authors offer a preliminary explanation involving particular magnetization dynamics, but say that additional study is needed to explain the voltage deviations.
A model result, not a hardware verdict
The evidence has a narrow scope. It comes from deterministic numerical and simplified theoretical calculations for one specified passive circular MTJ macrospin model, comparing in-plane and out-of-plane magnetization dynamics across the stated pulse sweeps. The reported threshold is from one displayed curve, and no uncertainty interval is reported for it.
That specificity limits what can be inferred beyond the calculation. The work tests positive periodic rectangular pulses and the listed amplitude, duration and period ranges; the threshold is tied to the displayed calculation rather than offered as a universal OOP value, and the same caution applies to the reported voltage jumps and drops. The authors’ interpretation remains that IP operation is preferable for the study’s pulse-detection setting.
The result also remains conditional on having two pulse parameters in hand before reading the third. Further work would need to test whether the predicted regimes appear in fabricated hardware and how other device geometries, bias fields or pulse waveforms affect identification. For the nonlinear cases, the specific magnetization-dynamics features behind the jumps and drops remain to be explained.
The work was supported by grant No. 2025.07/0237 from the National Research Foundation of Ukraine.
Paper data and sources
Original title: Spin-torque microwave detectors of positive rectangular pulse signals
Authors: V. Prokopenko, O. Shtanko, I. Sotnyk, O. Prokopenko
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text