Giant universal conductance fluctuations in the antiferromagnetic topological insulator MnBi2Te4

Abstract

Intrinsic magnetic topological insulators can host quantum states with quantized magneto-electric responses, such as the axion and Chern insulators states evidenced in ultra-thin MnBi2Te4 films. Yet, whereas quantization is investigated thoroughly, transport properties related to the phase of charge carriers remains unexplored. Here, we study quantum coherent transport in mesoscopic Hall bars fabricated from thick exfoliated MnBi2Te4 flakes, and reveal the longest phase-coherence length ever observed in a mesoscopic magnet (about 500nm at 1K), associated to 2D topological surface states. In the fully-coherent regime, significant non-local contributions to quantum interference up to the micron scale lead to giant-amplitude universal conductance fluctuations (about 20e2/h). In the self-averaging regime, the statistical properties of conductance fluctuations confirm the 2D nature of quantum interference and different dephasing mechanisms are identified, as due to either magnetism or magnetic flux through coherent loops. Remarkably, the weak decoherence in magnetic topological insulator nanostructures show their potential to realize novel quantum spin interferometers based on dephasing by local magnetic textures at liquid-helium temperatures.

Figures (9)

• Figure 1: Giant conductance fluctuations in MnBi$_2$Te$_4$ nanostructures. a,b), AFM images of Hall-bar shaped devices A and B, with Ti/Au ohmic contacts. These two devices have a different cross-section (device A : width $W=100$ nm, thickness $t=110$ nm; device B : $W=500$ nm, $t=55$ nm), and conductance fluctuations are studied for three different lengths of mesoscopic conductors ($L_{14}=3500$ nm, $L_{23}=1700/1100$ nm, $L_{56}=200/300$ nm, for A/B respectively); c,d), Magneto-conductance measured at $T=1$ K, for three different lengths, with a magnetic field $B_\perp$ applied perpendicular to the sample plane (out-of-plane OOP configuration). The three different magnetic states are identified by colored panels (antiferromagnet - AFM, pink; canted antiferromagnet - c-AFM, green; saturated magnetization, aligned along the applied field, blue).; e,f), Giant-amplitude universal conductance fluctuations $\delta G_{5-6}$ for short-length conductors, after subtraction of the classical magneto-conductance.
• Figure 2: Dephasing mechanisms and weak decoherence : correlation field, self averaging and scaling analysis.a), Magnetic field dependence of the universal conductance fluctuations $\delta G_{1-4}^A$, as measured at $T=1$ K for the 3.5 µm-long conductor in device A, after subtraction of the classical magneto-conductance; b), Temperature dependence of the UCF amplitude $\delta G_{1-4}^A$ due to self-averaging ($L_\varphi < L_{1-4}$), showing the weak damping at higher temperatures; c), Temperature dependence of the UCF correlation field $B_C$, measured for the different magnetic states. Whereas $B_C$ is directly related to the phase-coherence length in the uniform magnetization regime, the constant value in the AFM regime is due to spin dephasing; d), Scaling-law temperature dependence of the UCF amplitude $\delta G_{rms}.L^{\alpha}$ in the UM regime, revealing the slow decoherence ($\beta\approx0.25$). Inset (right): Length dependence of $\delta G_{rms}$. In the long-length limit ($L\gg L_\varphi$), the reduction of the UCF amplitude due to self averaging follows an usual power-law dependence $\delta G_{rms}\propto L^{-\alpha}$, with $\alpha \approx 3$. Inset (left): Temperature-dependence of $\alpha$
• Figure 3: Angular dependence and evidence for the TSS contribution to UCF.a), Magneto-conductance $G^{A}_{2-3}(B)$ measured at $T=1$ K in the OOP and IP configurations, with saturation fields $B^{sat}_{OOP}\approx7.5$ T and $B^{sat}_{IP}\approx10$ T. The grey trace is the first-magnetization curve measured in the IP configuration after cooldown in zero field; b,d), Universal conductance fluctuations $\delta G^{A}_{2-3}(B)$, as measured in the OOP (b) and IP (d) configurations. The direct comparison in the uniform magnetization regime ($B>B^{sat}$) reveals the change in $B_C$ (expected from the geometry) and, strikingly, a significant change in the UCF amplitude (not observed in other mesoscopic magnets). c) Comparison of $\delta G^{B}_{5-6}(B)$ at high magnetic fields, for both the OOP and IP configurations, showing the strong reduction of the UCF amplitude for the in-plane configuration (surface loops not probed by $B_{\vert\vert}$).
• Figure 4: Non-local measurements of quantum coherent transport at $T=$ 1K.a), Current-injection path and voltage probes used for the non-local measurements of the magneto-conductance in Device A; b), Universal resistance fluctuations $\delta R_{1-6}^A$, as measured in the local configuration with a perpendicular field $B_{\perp}$, after subtraction of the classical magneto-resistance background; c), Non-local magneto-resistance $R_{5-4}^{NL-A} (B_{\perp})$, representative of the UCF mostly probed by contact 5, about 200 nm away from the current path; d), Non-local magneto-resistance $R_{3-4}^{NL-A} (B_{\perp})$, representative of the UCF probed by contacts 3 and 4, about 2µm away from the current path.
• Figure S1: Universal conductance fluctuations measured under an out-of-plane magnetic field for a short-length $\delta G_{5-6}$, a medium-length $\delta G_{2-3}$ and a long-length $\delta G_{1-4}$ conductor at 1 K. Data shown for device A (a,b,c) and device B (d,e,f). Color code: antiferromagnetic (AFM) phase in pink, canted antiferromagnetic (cAFM) phase in yellow and phase of uniform magnetization (UM) in cyan. The perfect reproducibility of the UCF under both sweep directions in clearly seen in the UM phase, while in the cAFM phase the dephasing by local magnetic textures leads to some hysteresis.