Intermixing during growth of InAs self-assembled quantum dots in InP: A photoluminescence and tight-binding investigation

Atomic intermixing during the growth of self-assembled InAs quantum dots (QDs) in InP(001) by chemical beam epitaxy (CBE) can result in the formation of graded interfaces or even alloyed InAs1-xPx QDs. Taking advantage of a series of samples in which photoluminescence (PL) spectra are characterized by the presence of a large number of distinct peaks corresponding to the emission from QD families having the same thicknesses (h(QD)) in terms of an integer number of monolayers (MLs), we investigate intermixing by matching the experimentally observed transition energies to those calculated from tight-binding simulations. Two calculation frameworks are considered: (i) structures in which QDs are alloys with uniform P concentration [P] and (ii) nominally pure InAs QDs bordered by graded interfaces characterized by a diffusion length L-D. Excellent agreement between theory and experiment is achieved with either framework. The analysis reveals that the two frameworks yield similar composition profiles as the composition at the center of the QD in the case of graded interfaces is modified to a point where it becomes essentially equal to that calculated for a QD of uniform composition. The calculations also indicate that in both frameworks, two substantially different solution sets are compatible with the experimental results. The first one is characterized by all QDs having the same P composition independent of their size, while the second solution shows a decreasing degree of intermixing with increasing h(QD). To discriminate between the two solutions, we use them as input data in a Bloch-wave simulation of transmission electron microscopy (TEM) image contrast, providing a sequence of contrast versus h(QD) values. Only the scheme in which the amount of [P] is the same for all QD ensembles is compatible with the TEM observations. From the analysis of an extensive array of CBE-deposited QD samples, it is concluded that the observed PL transitions can be attributed to a 3 ML thick wetting layer and 4-14 ML thick QDs, with [P] ranging from 6% to 10% depending on the growth conditions. The determined values of [P] and L-D suggest that surface As/P exchange and strain-driven alloying are the most probable mechanisms for substantial P incorporation.