Low-temperature electron dephasing rates indicate magnetic disorder in superconducting TiN films

Abstract

We investigate electron transport and phase-breaking processes in thin titanium nitride (TiN) films of epitaxial quality. Previous studies show that a minute surface magnetic disorder significantly reduces the critical temperature ($T_\mathrm{c}$) and broadens the superconducting transition as the film thickness and device size decrease. We measure electron dephasing rates via magnetoresistance from $T_\mathrm{c}$ to $\sim 4T_\mathrm{c}$ in various-thickness TiN films. Electron dephasing occurs on the picosecond timescale and is nearly independent of temperature, differing from the expected inelastic scattering due to the electron-phonon and electron-electron interactions near $T_\mathrm{c}$, which occur over a nanosecond timescale. We propose spin-flip scattering as a possible additional phase-breaking mechanism. The significant increase in the dephasing rate for the thinnest film indicates that magnetic disorder resides near the surface of naturally oxidized films. Our research suggests that magnetic disorder may be a significant contributor to RF dissipation in superconducting devices based on TiN.

Figures (4)

• Figure 1: TiN film characterization. (a) TEM image of a 12 nm thick TiN layer on c-sapphire. (b) High-resolution HAADF-STEM image of a TiN/Al$_2$O$_3$ interface and enlarged fragments, showing atomic packing in both structures. The projections of the structures superimposed onto the experimental images (Ti – turquoise, Al – gray). [112] Fourier transform of TiN is shown as inset in left upper corner. (c)-(e) EDX maps confirms a homogeneous distribution of Ti and N in the film and presence of thin oxidized layer on the interface. (f) Wide scan XPS spectra of 3 nm (XPS3) and 20 nm (XPS20) TiN films. (g) Chemical and phase depth profiles of XPS3 and XPS20.
• Figure 2: (a) $R_\mathrm{s}(T)$ dependencies over a wide $T$ range on semi-log scale. Experimental data shown by symbols, the Bloch-Grüneisen fits by dashed lines. (b) Main: $T$-dependencies of $R_\mathrm{s}$ on magnetic field for sample MR2. Inset: $T$-dependence of second critical magnetic field, $B_\mathrm{c2}$.
• Figure 3: The dimensionless magnetoconductance, $\delta G(B,T)$, is plotted versus magnetic field for a representative sample (MR3) at various bath temperatures (a) and compared for four samples (MR1 - MR4) at a fixed $T/T_\mathrm{c} \approx 1.7$ (b). Experimental data are represented by symbols, and black lines show the best fits using Eq.\ref{['eq:fit_MC_1']}.
• Figure 4: Phase-Breaking Rates and Scattering Processes in Epitaxial TiN Films. (a) Electron phase-breaking rate $\tau_\phi^{-1}$ for all TiN samples. MR1 data are blue diamonds, MR2 orange squares, MR3 green circles, MR4 purple triangles. Data are plotted in log-log scaled. Solid colored lines show $2\tau_\mathrm{s}^{-1}$ estimates from the AG model using experimental $T_\mathrm{c}$ values. Black lines indicate contributions from the e-ph (dashed), e-e (solid), and e-fl (dot-dashed) scattering for the thinnest sample (MR1). Error bars for $\tau_\mathrm{\phi}^{-1}$ are obtained from temperature measurement uncertainty, estimated from resistance fluctuations at $B = 0$. (b) Estimated spin-flip scattering rate $\tau_\mathrm{s}^{-1}$ vs. inverse thickness $d^{-1}$. Dashed line shows a fit to the data, indicating that the dominant spin-flip scattering originates from near-surface magnetic disorder. (c) Main: $R_\mathrm{s}$ dependencies for MR3 and MR5 samples (MR5 has a 1-nm Cr layer on TiN). Inset: Experimental $\tau_\mathrm{\phi}^{-1}$ values for MR5, compared to initial MR3 data and AG model predictions.