Analysis of Point-contact Andreev Reflection Spectra in Spin Polarization Measurements

TL;DR

This paper systematically analyzes point-contact Andreev reflection spectra in ferromagnetic materials, examining transport regimes, modeling effects, and the influence of interface parameters on spin polarization measurements.

Contribution

It provides a detailed comparison of ballistic and diffusive models for PCAR spectra and highlights the impact of transport regimes and interface effects on spin polarization estimation.

Findings

Diffusive and ballistic models yield similar spin polarization values within ~3%.

Variation of the superconducting gap can cause spurious Z dependence of Pc.

System-dependent Pc(Z) variation likely due to spin-flip scattering at interfaces.

Abstract

We present a systematic analysis of point-contact Andreev reflection (PCAR) spectra for ferromagnetic materials, using both modeling and experimental data. We consider the relationship between ballistic and diffusive transport, the effect of different transport regimes on spin polarization measurements, and the importance of unambiguous identification of the type of transport regime. We find that in a realistic parameter range, the analysis of PCAR spectra of purely diffusive character by a ballistic model yield approximately the same (within ~3%) values of the spin polarization and the barrier strength Z larger by ~ 0.5-0.6. We also consider the dependence of polarization values on Z, and have shown by simple modeling that letting the superconducting gap vary as an adjustable parameter can result in a spurious dependence of the spin-polarization Pc on Z. At the same time we analyzed the effects of finite Z on the apparent value of Pc measured by the PCAR technique, using a large number of examples from both our own measurements and from the literature. We conclude that there is a system-dependent variation in Pc (Z), presumably due to spin-flip scattering at the interface. However, the exact type of this dependence is hard to determine with any statistical certainty.