Analysis of capping with G a A s S b N thin layers in (un)coupled I n A s/ G a A s multi quantum dot layers for enhanced solar cells

Abstract

Nowadays, stacked InAs/GaAs quantum dots (QDs) capped with layers different than GaAs are being applied in photovoltaic and photodetector technologies due to their potential for tailoring the optical properties and enhancing the device efficiency. 1,2 The particular use of GaAsSbN capping layers (CLs) allows controlling the QD size as well as the electron and hole confinement potentials in a wide range. 3 In addition, multi‐coupled QD structures formed by vertically aligned QDs represent a very interesting approach, which enhances the photoluminescence (PL) intensity and reduces the spectral width by utilizing carrier tunneling among QDs. However, in this type of nanostructures the control of the QD composition, size and alignment are important parameters as they affect the electronic coupling and the carrier lifetime. In this work, we analyze the effect of using 2.5 nm‐thick capping layers of GaAs 0.76 Sb 0.2 N 0.04 in an (un)coupled 10 stacked layer structure of InAs/GaAs QD by transmission electron microscopy (TEM) techniques. For this, we have compared three samples: two samples with uncoupled QDs (the spacer thickness is 40 nm) with ( SbN‐u ) and without CL ( Ref ), and a coupled sample with CL and a spacer thickness of 10 nm ( SbN‐c ). PL and photocurrent (PC) data are also used in the discussion. Cross sectional g200 dark field (DF) images sensitive to composition have been used to obtain a statistic of the buried QD size (Figure 1). For the uncoupled samples, our measurement on more than 90 QDs revealed an unexpected similar height for Ref and SbN‐u samples (4.7 nm) while the average diameter presents a slight increase in sample SbN‐u with respect to Ref sample (from 17 to 19 nm). The panorama changes totally for SbN‐c sample, where the average height and diameter increases layer by layer from 4.7 to 5.2 nm and from 23 to 34 nm, respectively. The relative volume rises quickly (Figure 1) up to the 4 th layer, and after slowly, becoming 7 times higher in the upper layer. The shield effect of Sb in the QD decomposition is not appreciated. Indeed, EDX measurements revealed only tiny traces of Sb in the CLs that we related with the typical delay observed in the Sb incorporation due to the CL small thickness. In addition, strain maps from high resolution TEM images acquired on the [110] pole axis were calculated. The comparison between the uncoupled samples discloses that N is incorporated in the CL. The same occurs in the coupled sample where a high compressive strain appears over the QDs (Figure 2). More important, the strain into the QDs decreases in a similar way that the volume increases. Finite element simulations suggest a modification of segregation process during the growth, with an increase of Ga incorporation in the QDs of the upper layers.