Continuous-film vs. device-level ferromagnetic resonance in magnetic tunnel junction thin films

We quantitatively compared film-level ferromagnetic resonance (FMR) measurements using standard vector network analyzer (VNA) techniques with device-level FMR measurements for both thermal FMR (T-FMR) and field-swept spin-torque FMR (FS-ST-FMR) techniques on magnetic tunnel junction (MTJ) thin films with in-plane magnetization. The film and FS-ST-FMR device determination of damping α are in agreement; however, α cannot be reliably determined by use of T-FMR device measurements due to bandwidth limitations. The device-level intercept of H_res vs. f is lower than film-level measurements of the effective magnetization (M_eff) due to the demagnetizing field and exchange coupling of the patterned free layer. The intercept shows device-to-device variations due to a combination of size variation and local film variations. At the device level, the inhomogeneous broadening (ΔH₀) is nearly zero, while in film-level measurements, μ₀ΔH₀ > 10 mT due to averaging of the local film variations detected explicitly in the intercept of H_res vs. f at the device level. These results suggest that continuous-film and FS-ST-FMR measurements on multiple devices can provide comparable information about thin-film M_eff, α, and ΔH₀ with minimal interpretation, but caution is necessary when using T-FMR to determine α or ΔH₀.