Moving towards nano-TCAD through multimillion atom quantum dot simulations matching experimental data

TL;DR

This paper demonstrates that large-scale atomistic simulations can accurately match experimental data for quantum dot wavelengths, advancing nano-TCAD and providing new insights into quantum dot shape changes.

Contribution

It presents the first systematic multimillion-atom simulations that match experimental data without parameter adjustments, showing the potential of nano-TCAD for quantum dot design.

Findings

Simulations match experimental nonlinear behavior closely.

Quantum dots change shape during tuning.

Atomistic models provide deeper insights than simple models.

Abstract

Low-loss optical communication requires light sources at 1.5um wavelengths. Experiments showed without much theoretical guidance that InAs/GaAs quantum dots (QDs) may be tuned to such wavelengths by adjusting the In fraction in an InxGa1-xAs strain-reducing capping layer (SRCL). In this work systematic multimillion atom electronic structure calculations qualitatively and quantitatively explain for the first time available experimental data. The NEMO 3-D simulations treat strain in a 15 million atom system and electronic structure in a subset of ~9 million atoms using the experimentally given nominal geometries and without any further parameter adjustments the simulations match the nonlinear behavior of experimental data very closely. With the match to experimental data and the availability of internal model quantities significant insight can be gained through mapping to reduced order models and their relative importance. We can also demonstrate that starting from simple models has in the past led to the wrong conclusions. The critical new insight presented here is that the QD changes its shape. The quantitative simulation agreement with experiment without any material or geometry parameter adjustment in a general atomistic tool leads us to believe that the era of nano Technology Computer Aided Design (nano-TCAD) is approaching. NEMO 3-D will be released on nanoHUB.org where the community can duplicate and expand on the results presented here through interactive simulations.