Analysis of the Sb and N distribution in GaAsSb/GaAsN superlattices for solar cell applications

Abstract

Dilute nitride III–V alloys have attracted a lot of attention in the last decade due to its wide tunability of both band gap and lattice constant that makes them a potential candidate in multi‐junction solar cell technology. For certain, these alloys can be used to improve the conventional lattice matched three‐junction solar cell by the replacement of Ge bottom cell and/or also by adding a fourth junction with a bandgap of 1 eV, which is predicted to provide efficiencies beyond 50%. 1 In this sense, the GaAs 1‐ x‐y Sb x N y quaternary alloys have a surplus due to the Sb‐surfactant effect and the possibility of an independent tuning of the electron and hole confinements. 2 However, the composition control in GaAsSbN materials present important challenges, and undesired phenomena such as segregation, clustering or phase separation are typically observed. A possible route to overcome these handicaps is the use of superlattices, based on the fine stacking of ternary and/or binary thin layers, which could, in addition, offer extra advantages (such as a greater collection efficiency). In any case, it is expected that the spatial separation of Sb and N atoms would improve the composition control and therefore facilitate a smaller lattice mismatch deviation with respect to the substrate. The present work analyses the Sb and N distribution in different GaAsSb/GaAsN superlattice structures grown lattice matched to GaAs substrates by molecular beam epitaxy. The nominal contents of Sb and N are 6.12% and 2.7%, respectively, and the periodicity is varied from a few MLs to ~20 nm. The samples were then characterised by different transmission electron microscopy techniques (TEM) and the results supported by X‐ray diffraction (XRD) and cross‐sectional scanning tunneling microscopy (X‐STM) data. Firstly, compositionally sensitive 002 dark field (DF) imaging was carried out in order to estimate the compositional distribution and periodicity, where the Sb and N rich areas depict a brighter or darker intensity than the GaAs ones, respectively (Figure 1a). Although the periodicity is close to planned, the intensity analysis revealed a compositional deviation with respect to the nominal design, especially in the Sb distribution. Certainly, although N atoms seem to be confined into the GaAsN layers exhibiting greater stability against interdiffusion, Sb atoms are spread along the whole structure. The precise control over the N position in the supperlattice is confirmed by atomic scale X‐STM images in which individual N atoms are directly observed. The low contents of Sb and N unable the application of high angle annular dark field (HAADF) conditions in STEM images (Figure 1.b). Instead, ADF images are used, since N yields a brighter contrast due to the distortion that produces in the structure 3,4 (Figure 1c) and compared with electron dispersive x‐ray (EDX) maps acquired simultaneously, that allow us to determine the distribution and the overall content of Sb (Figure 1d). All these analysis pointed out to a lower incorporation both of Sb and N that depends on the period of the structure. More important, a strong interdiffusion of Sb is evidenced. The Sb content reaches its maximum value at the top of the GaAsSb layer and a minimum at the top of the GaAsN layer (Figure 2). As consequence, there is a phase difference between the ADF images and the EDX maps that must be taken in consideration. The system moves away from the original design, where the Sb content adopts a sawtooth distribution and the square superlattice structure is preserved mainly by the N distribution. The influence of the period of the superlattice on these effects is discussed.