Accuracy of circular polarization as a measure of spin polarization in quantum dot qubits

TL;DR

This paper investigates how the geometry of quantum dots affects the accuracy of using circular polarization of emitted light as a measure of the actual spin polarization within the dot, revealing significant geometric influences.

Contribution

It provides detailed calculations of optical matrix elements considering quantum dot geometry, showing how emission direction impacts the correlation between spin and photon polarization.

Findings

Circular polarization of emitted light is strongly influenced by quantum dot geometry.

Observed polarizations of ~1% suggest actual spin polarization of ~20%.

Measuring along the growth direction yields near-perfect conversion of spin to photon polarization.

Abstract

A quantum dot spin LED provides a test of carrier spin injection into a qubit, as well as a means of analyzing carrier spin injection in general and local spin polarization. The polarization of the observed light is, however, significantly influenced by the dot geometry so the spin may be more polarized than the emitted light would naively suggest. We have calculated carrier polarization-dependent optical matrix elements using 8-band strain-dependent k.p theory for InAs/GaAs self-assembled quantum dots (SAQDs) for electron and hole spin injection into a range of quantum dot sizes and shapes, and for arbitrary emission directions. The observed circular polarization does not depend on whether the injected spin-polarized carriers are electrons or holes, but is strongly influenced by the SAQD geometry and emission direction. Calculations for typical SAQD geometries with emission along [110] show light that is only ~5% circularly polarized for spin states that are 100% polarized along [110]. Therefore observed polarizations [Chye et al. PRB 66, 201301(R)] of ~1% imply a spin polarization within the dot of ~20%. We also find that measuring along the growth direction gives near unity conversion of spin to photon polarization, and is the least sensitive to uncertainties in SAQD geometry.