Review and Analysis of Measurements of the Spin Hall Effect in Platinum

TL;DR

This paper reviews various experimental measurements of the spin Hall effect in platinum, identifies reasons for discrepancies, and provides a new measurement of the spin diffusion length that reconciles previous conflicting results.

Contribution

It offers a critical analysis of past measurements, corrects previous misinterpretations, and presents a new, more accurate measurement of the spin diffusion length in platinum.

Findings

Discrepancies in spin Hall measurements are due to experimental assumptions and device effects.

The new spin diffusion length in Pt is 1.4 .3 nm, much smaller than previously assumed.

Revised values suggest the spin Hall conductivity is (1.4 - 3.4) ^5 /(2e) (Ohm-m)^-1 and the spin Hall angle exceeds 0.05.

Abstract

Several different experimental techniques have been used in efforts to measure the spin Hall conductivity and the spin Hall angle in Pt samples at room temperature, with results that disagree by more than a factor of 20, with spin Hall conductivities from 2.4 x 10^4 to 5.1 x 10^5 [hbar/(2e)] (Ohm-m)^-1 and spin Hall angles from 0.0037 to 0.08. We review this work, and analyze possible reasons for the discrepancies. We explain that the smallest values for the spin Hall angle that have been reported, based on measurements of lateral permalloy/copper/platinum devices, are incorrect because the original analyses did not properly take into account that copper layers in these devices will shunt charge current flowing through adjacent platinum wires, thereby greatly reducing the size of the spin-Hall-related signals. We suggest that differences between the results for the spin Hall angle found by other experimental techniques are primarily a consequence of different assumptions about the value of the spin diffusion length in Pt. We present a new measurement of the spin diffusion length in Pt within sputtered Pt/permalloy bilayer thin films at room temperature, finding 1.4 \pm 0.3 nm, a much smaller value than has generally been assumed previously. With this value for the spin diffusion length, the previously-discordant results can be brought into much better agreement, with the result that the spin Hall conductivities are (1.4 - 3.4) x 10^5 [hbar/(2e)] (Ohm-m)^-1 and the spin Hall angles are greater than 0.05. These values are sufficiently large that the spin Hall effect in Pt can be used to generate spin currents and spin transfer torques strong enough for efficient manipulation of magnetic moments in adjacent ferromagnetic layers.