Table of Contents
Fetching ...

Low-temperature spin dynamics in LAFO thin films: from cubic anisotropy to TLS-limited coherence

Srishti Pal, Guanxiong Qu, Hervé M. Carruzzo, Katya Mikhailova, Lerato Takana, Qin Xu, Yuri Suzuki, Clare C. Yu, Gregory D. Fuchs

TL;DR

This work investigates low-temperature spin dynamics in epitaxial LAFO thin films using broadband FMR across 0.44–68 K. It establishes a crossover from conventional cubic anisotropy–driven behavior along [100] to TLS-limited coherence with an unusually large sixth-order cubic anisotropy along [110], driven by disorder-induced local symmetry breaking. A two-TLS damping model, with distinct exchange splittings ω_ex,1 and ω_ex,2, accounts for the nonmonotonic linewidth and its temperature dependence, indicating both exchange-coupled impurities and nearly free paramagnetic centers contribute to damping. The findings highlight defect-induced TLS as the fundamental limiter of magnon coherence in ferrimagnetic insulators at ultralow temperatures and provide a framework for anisotropy engineering and growth optimization toward low-loss, coherent magnonics and quantum transduction applications.

Abstract

We investigate the low-temperature spin dynamics of epitaxial lithium aluminum ferrite (LAFO) thin films using broadband ferromagnetic resonance (FMR) spectroscopy from 0.44 K to 68 K. The results reveal a crossover from conventional cubic anisotropy-dominated behavior at higher temperatures to pronounced linewidth broadening and higher-order anisotropy contributions at cryogenic temperatures. With the magnetic field oriented along the [100] crystallographic direction, the resonance is well-captured by four-fold in-plane and out-of-plane uniaxial anisotropies. In contrast, measurements with the field along the [110] direction reveal the presence of an unusually large sixth-order cubic anisotropy term that is symmetry-suppressed for [100] but becomes apparent under this field orientation at ultralow temperatures, indicating a substantial modification of the anisotropy landscape. Independent linewidth analysis shows a pronounced peak near 8 K and a subtle monotonic enhancement with decreasing temperatures below 2 K, features consistent with dissipation mediated by a bath of two-level systems (TLS) arising from antisite defects and localized Fe$^{3+}$ moments. Comparison with TLS-based models demonstrates that both exchange-coupled impurities and nearly free paramagnetic centers contribute to the observed damping. Our results establish LAFO as a model ferrite system where disorder-induced TLS limit spin coherence at ultralow temperatures, providing new insights into anisotropy engineering, magnetic relaxation, and the design of ferrimagnetic insulators for coherent magnonics. These findings offer a framework for future optimization of growth conditions.

Low-temperature spin dynamics in LAFO thin films: from cubic anisotropy to TLS-limited coherence

TL;DR

This work investigates low-temperature spin dynamics in epitaxial LAFO thin films using broadband FMR across 0.44–68 K. It establishes a crossover from conventional cubic anisotropy–driven behavior along [100] to TLS-limited coherence with an unusually large sixth-order cubic anisotropy along [110], driven by disorder-induced local symmetry breaking. A two-TLS damping model, with distinct exchange splittings ω_ex,1 and ω_ex,2, accounts for the nonmonotonic linewidth and its temperature dependence, indicating both exchange-coupled impurities and nearly free paramagnetic centers contribute to damping. The findings highlight defect-induced TLS as the fundamental limiter of magnon coherence in ferrimagnetic insulators at ultralow temperatures and provide a framework for anisotropy engineering and growth optimization toward low-loss, coherent magnonics and quantum transduction applications.

Abstract

We investigate the low-temperature spin dynamics of epitaxial lithium aluminum ferrite (LAFO) thin films using broadband ferromagnetic resonance (FMR) spectroscopy from 0.44 K to 68 K. The results reveal a crossover from conventional cubic anisotropy-dominated behavior at higher temperatures to pronounced linewidth broadening and higher-order anisotropy contributions at cryogenic temperatures. With the magnetic field oriented along the [100] crystallographic direction, the resonance is well-captured by four-fold in-plane and out-of-plane uniaxial anisotropies. In contrast, measurements with the field along the [110] direction reveal the presence of an unusually large sixth-order cubic anisotropy term that is symmetry-suppressed for [100] but becomes apparent under this field orientation at ultralow temperatures, indicating a substantial modification of the anisotropy landscape. Independent linewidth analysis shows a pronounced peak near 8 K and a subtle monotonic enhancement with decreasing temperatures below 2 K, features consistent with dissipation mediated by a bath of two-level systems (TLS) arising from antisite defects and localized Fe moments. Comparison with TLS-based models demonstrates that both exchange-coupled impurities and nearly free paramagnetic centers contribute to the observed damping. Our results establish LAFO as a model ferrite system where disorder-induced TLS limit spin coherence at ultralow temperatures, providing new insights into anisotropy engineering, magnetic relaxation, and the design of ferrimagnetic insulators for coherent magnonics. These findings offer a framework for future optimization of growth conditions.
Paper Structure (23 sections, 24 equations, 12 figures)

This paper contains 23 sections, 24 equations, 12 figures.

Figures (12)

  • Figure 1: (a)-(b) Microscope images of the 15 nm LAFO/MAO flipped on broadband coplanar waveguide chip along [100] and [110] crystallographic directions, respectively. The magnified microscope image in (b) shows the alignment of the sample along the [110] direction. (c)-(d) Broadband FMR response (dark cyan curve) at 13 GHz at 0.44 K fit to a combination of absorptive and dispersive contributions (red solid curves) along [100] and [110] directions, respectively.
  • Figure 2: (a), (c) Frequency dependence of resonance field at 0.44 K with fits to the Smit-Beljers forms for [100] and [110] directions, respectively. (b), (d) Fitted frequency variation of FMR linewidth at 0.44 K along [100] and [110] directions, respectively. The error bars shown correspond to the standard errors obtained from the nonlinear least-squares fits of the FMR spectrum using Eq. \ref{['eq:FMR_Fit']} at each frequency.
  • Figure 3: (a)-(b) Temperature evolution of the anisotropy fields $H_{4,||}$ and $H_{2,\perp}$ measured with the magnetic field along [100]. (c)-(d) Temperature evolution of Gilbert damping $\alpha$ and the inhomogeneous broadening $\Delta H_0$ also with the field along [100]. Error bars indicate the standard errors from nonlinear least-squares fits performed for the resonance frequency and linewidth variations using Eqs. \ref{['eq:SB_Fit']} and \ref{['eq:linewidth_model']}, respectively.
  • Figure 4: TLS contribution to the temperature evolution of the linewidth with the field along [100] at 13 GHz. Blue symbols denote the experimentally obtained linewidths obtained from Lorentzian fits (Eq. \ref{['eq:FMR_Fit']}) to the raw FMR spectra, with error bars representing the standard errors from the nonlinear least-square fittings. The red solid curve shows the result of fitting with a two-TLS model, as described in Section \ref{['MS:Results-TLS']}.
  • Figure 5: Correlation between linewidth and anisotropy with the field along the [100] direction at 13 GHz. Black symbols (left axis) show the temperature dependence of the FMR linewidth, while blue symbols (right axis) show the corresponding evolution of the 4-fold in-plane the anisotropy field. The shaded background serves as a visual guide to temperature, highlighting the crossover from a low-temperature regime dominated by paramagnetic impurity spins to a high-temperature regime where exchange-coupled impurities become relevant, as discussed in Section \ref{['MS:Discussion']}. Error bars represent the standard errors obtained from nonlinear least-squares fits to the frequency and linewidth data.
  • ...and 7 more figures