Direct demonstration of time-reversal-symmetry-breaking spin injection from a compensated magnet

TL;DR

Direct $T$-symmetry-breaking spin injection from a compensated magnet Mn5Si3 is demonstrated in a lateral spin valve, using two bias geometries to inject spins into a Cu channel and detect with Py electrodes. Despite a vanishing net magnetization, switching between time-reversed states of Mn5Si3 alters the non-local spin signal, corroborated by a remanent anomalous Hall effect that switches sign with the time-reversed state. Density functional theory shows a $d$-wave altermagnetic Fermi surface yielding distinct spin-up and spin-down transport channels with about 10% current polarization, aligning with the observed spin injection and its geometry-dependent behavior. The work establishes compensated magnets as viable, energy-efficient spin injectors for scalable spintronic devices, enabling TRS-breaking spin transport without bulk magnetization.

Abstract

The injection, propagation and detection of spin currents are essential physical processes in spintronics. So far, the separation of charge and spin currents was facilitated by the electrical spin injection from a ferromagnet (FM) or the injection by a relativistic spin Hall effect. The devices employed are lateral spin valves comprising spatially separated injection and detection electrodes, connected by a spin-propagation channel. The time-reversal symmetry (TRS) breaking FM spin injection is realized in a geometry with an electrical bias applied between the injection electrode and the channel and is modelled by a conserved spin-polarized drift current. In contrast, the spin injection by the T-symmetric relativistic spin Hall mechanism is driven by an electrical bias applied across the injection electrode alone, and is modelled by a non-conserved spin current transverse to the applied bias. In this work, we use a lateral spin valve with a Mn5Si3 injection electrode to directly demonstrate a TRS-breaking spin injection from a compensated magnet with a vanishing net magnetization. Specifically, the TRS-breaking is demonstrated by the fact that switching between time-reversed states of the compensated magnet changes the detected spin signal. Moreover, the TRS-breaking nature of the spin injection is observed in both experimental geometries with the different electrical biasing, while using the same detection electrode. We show that this unconventional spin-injection is consistent with different magnitudes and propagation angles of electrical currents in the spin-up and spin-down channel in a d-wave altermagnet. Here our symmetry analysis and first-principles calculations are based on the compensated collinear altermagnetic order which has provided a comprehensive microscopic interpretation of earlier structural, magnetic, and anomalous Hall and Nernst measurements in Mn5Si3 thin films.

Figures (4)

• Figure 1: Spin-injection mechanisms and direct detection in a lateral spin valve. Top row: A lateral spin valve comprising injection and detection electrodes connected by a transverse channel. An applied electrical bias ($\textit{I}_\text{bias}$) injects spins into the channel, which then diffuse towards the detection electrode, generating a non-local voltage, $\textit{V}_\text{NL}$. The left and right schematics illustrate the first and the second electrical biasing geometry in which the respective spin-injection mechanisms were directly observed. Second row: $\cal{T}$-symmetric relativistic spin injection from a non-magnetic electrode, arising form a transverse spin-Hall current, in the second biasing geometry. In the first biasing geometry, spin injection from a non-magnetic electrode is not observed. Third row: $\cal{T}$-symmetry-breaking spin injection in the first biasing geometry, originating from the finite magnetization of a ferromagnet ($M \neq 0$), which generates a spin-polarized longitudinal current. In the second biasing geometry, the spin injection is due to the $\cal{T}$-symmetric spin Hall effect as in the non-magnetic electrode. Bottom row: In this work, we demonstrate $\cal{T}$-symmetry-breaking spin injection from a compensated magnet ($M = 0$), independent of the biasing geometry.
• Figure 2: Magnetotransport measurements for the lateral spin valve device at 100 K.a False-colored scanning electron microscope image of the Py/Cu/$\text{Mn}_5\text{Si}_3$ lateral spin valve device (right) and a reference lateral spin device with two Py electrodes (left). b Non-local resistance (detected between contacts 1 and 2) for the reference Py/Cu/Py lateral spin valve in the biasing geometry across the Py/Cu interface (contacts 3 to 8). The magnetic field is applied along the y-direction. Black arrows indicate the magnetization direction of each Py electrode. The spin signal, $\Delta\textit{R}_\text{NL}$, is indicated by a grey arrow. c Non-local resistance (detected between contacts 1 and 2) for the reference Py/Cu/Py lateral spin valve in the biasing geometry along the Py electrode (contacts 3 to 4). The magnetic field is applied along the x-direction; inset: the same measurement with the field along the y-direction. Black arrows direction indicate the magnetization state of the Py detector electrode. The spin signal, $\Delta\textit{R}_\text{NL}$, is indicated by a grey arrow. d Non-local resistance for the Py/Cu/$\text{Mn}_5\text{Si}_3$ lateral spin valve (detected between contacts 7 and 8) in the biasing geometry across the $\text{Mn}_5\text{Si}_3$/Cu interface (contacts 5 to 1). The magnetic field is applied along the y-direction. e Non-local resistance for the Py/Cu/$\text{Mn}_5\text{Si}_3$ lateral spin valve (detected between contacts 7 and 8) in the biasing geometry along the $\text{Mn}_5\text{Si}_3$ wire (contacts 5 to 6). The magnetic field is applied along the y-direction.
• Figure 3: $\cal T$-symmetry-breaking nature of the spin injection from $\text{Mn}_5\text{Si}_3$. a False color SEM image of the device with defined Hall cross on the $\text{Mn}_5\text{Si}_3$ electrode. b Anomalous Hall effect measured in the $\text{Mn}_5\text{Si}_3$ wire (left axis) despite the vanishing net magnetic moment (right axis). The linear slope of the magnetization curve corresponds to the diamagnetic signal of the substrate (see Supplementary Fig. S2 for details). The magnetic field is applied along the y-axis. c Non-local spin signal (detected between contacts 6 to 7) for the Py/Cu/$\text{Mn}_5\text{Si}_3$ lateral spin valve with biasing geometry across the $\text{Mn}_5\text{Si}_3$/Cu interface (contacts 3 to 1). The magnetic field is applied along the y-axis. The non-local signal exhibits intermediate steps as in the anomalous Hall signal. Stars mark states which are probed by measuring minor loops shown in panels d and e after presetting the system magnetization with -2 T and +2 T in-plane magnetic fields, respectively.
• Figure 4: Spin-dependent transport in $\text{Mn}_5\text{Si}_3$ films. Magnetic and crystal structure (a) of the the altermagnetic phase of $\text{Mn}_5\text{Si}_3$ and the corresponding first-principle calculations of the non-relativistic $\cal T$-symmetry-breaking Fermi surface (b) with separated spin-up and spin-down channels and the corresponding calculated spin polarization (c). The spin polarization of the electrical current reaches 10% and it is only weakly influenced by the relativistic spin-orbit coupling (SOC) (dotted line). Considering the Fermi surfaces are dominated by the non-relativistic altermagnetic exchange interactions (solid line marked by nosoc) and schematically shown in panel e the biasing geometry shown in panel d will generate spin currents with non-zero net component injected in the Cu channel. In the biasing geometry shown in panel f the same Fermi surfaces will result in net spin polarization injected in the transverse direction into the Cu channel. The time-reversed Fermi surface (h) will lead to opposite spin polarizations in both geometries (g,i).