1/ f noise and two-level systems in MBE-grown Al thin films

TL;DR

This study quantitatively links $1/f$ resistance noise in 10-nm Al films to the activation-energy distribution of two-level systems (TLS). By comparing MBE-grown and EB evaporated Al, the authors find a pronounced TLS activation peak at $E_p \approx 0.78$ eV and show that MBE-grown films have a roughly $3\times$ lower noise at 300 K and about an order of magnitude lower TLS density than EB films, indicating grain-boundary diffusion of Al atoms as the TLS source. The peak occurs near $T_p \approx 320$ K, consistent with diffusion along grain boundaries, and the TLS density relative to total defects remains a small but temperature-dependent fraction ($n_{TLS}(T_p) \sim 0.5\%$–$2\%$). The work underscores the importance of grain-boundary control in ultra-thin Al films for minimizing TLS-related losses in superconducting and quantum devices.

Abstract

Aluminum thin films are essential to the functionalities of electronic and quantum devices, where two-level systems (TLS) can degrade device performance. MBE-grown Al films may appeal to these applications due to their low TLS densities. We studied the energy distributions of TLS densities, g(E), in 10-nm-thick MBE-grown and electron-beam evaporated Al films through 1/f noise measurements between 80 and 360 K. At 300 K, the noise magnitudes in MBE-grown films are about three times lower than in the electron-beam evaporated films, corresponding to the g(E) values about ten times lower in the former than in the latter. Compared with previously established observations, we identified that the 1/f noise was generated by thermally activated TLS at grain boundaries.

Figures (5)

• Figure 1: (a) An optical micrograph of a M10 film and a schematic of the $1/f$ noise measurement setup. The ballast resistor $R_1$ was typically a factor $\approx$ 5–10 times greater than the sample resistance. The adjustable resistor $R_2$ was used to balance the bridge. (b) A XRD scan across the Al (11$\overline{1}$) diffraction peak along the Al [11$\overline{2}$] direction for a M10 film. X-axis is the scattering vector $q$, and y-axis is intensity.
• Figure 2: (a) Resistivity as a function of temperature for five Al films. The lowest temperature shown here is 2 K. The M10A and P10A films exhibit a superconducting transition at 1.3 and 1.7 K, respectively (not shown). (b) PSD of M10A film at 300 K under various bias voltages $V_{\rm rms}$. The dashed line indicates $S_V \propto f^{-1}$ and is a guide to the eye. In the AC resistance bridge measurement scheme [Fig. 1(a)], the total bias voltage drop across the sample $V=2V_{\rm rms}$ [Eq. (1)], where $V_{\rm rms}$ is the root-mean-square voltage drop across one-half of the sample.Scofield1987 The inset shows $\left \langle f S_V \right \rangle \propto V^2$. The straight solid line is a linear fit.
• Figure 3: (a) Variation of $\gamma$ with temperature for five Al films. (b) $\gamma$ versus $T$ for M10A, M10C and P10A between 240 and 360 K. The solid curves are guides to the eye.
• Figure 4: Variation of $\alpha (f = 1\,{\rm Hz})$ with temperature for M10A and P10A films. Theoretical (open symbols) and experimental (closed symbols) values satisfactorily agree.
• Figure 5: Averaged activation energy distribution $g(E)$ for M10 and P10 films. Red circles are averaged over M10A, M10B, and M10C films, while blue squares are averaged over P10A and P10B films.