A prediction of dislocation-free CdTe/CdS photovoltaic multilayers via nano-patterning and composition grading

Defects in multilayered films have long been a performance-limiting problem for the semiconductor industry. For instance, CdTe/CdS solar cell efficiencies have had significant improvement in the past 15years or more without addressing the problem of high misfit dislocation densities. Overcoming this stagnation requires a fundamental understanding of interfacial defect formation. Herein, we demonstrate a new first principles-based CdTe bond-order approach that enables efficient molecular dynamics to approach the fidelity of density functional theory. Stringent quantum-mechanical verification and experimental validation tests reveal that our new approach provides an accurate prediction of defects that earlier methods cannot. Using this new capability, we show that misfit dislocations in CdTe/CdS multilayers can be significantly reduced via nano-patterning and composition grading and more importantly, dislocation-free multilayers naturally arise when the pattern dimension is reduced below 90nm. Our predictive methods are generally applicable to other materials, highlighting a rational approach towards low-defect semiconductor films. Copyright © 2015 John Wiley & Sons, Ltd.