Abstract
2‐dimensional nanostructures consisting only of a single layer of material have attracted great research interest in the past years. Among them are e.g. transition metal dichalcogenides like MoSe 2 , WS 2 and their ternary alloys. But also In x Ga 1‐x N/GaN short period superlattices (SPSL) built up of In x Ga 1‐x N monolayer quantum wells belong to this material class [1]. For such kind of nanostructures relevant structural parameters, which determine material properties and thus device performance, are interface quality, alloy composition and possible ordering phenomena. Characterization of these quantities, at atomic scale, is commonly performed by high angle annular dark field imaging using a scanning transmission electron microscope (STEM HAADF). In the past years even quantitative composition analysis at atomic scale by STEM HAADF imaging has been demonstrated for various material systems, including In x Ga 1‐x N quantum wells [2]. What makes STEM HAADF imaging so attractive for that purpose is the generally valid monotonic relationship between image intensity and the mean atomic number Z of the probed material (commonly expressed by the Z 1.7 rule of thumb for the image intensity). However, in our combined experimental and theoretical STEM HAADF analysis of In 0.33 Ga 0.67 N/GaN SPSL consisting of ordered In 0.33 Ga 0.67 N monolayers we have observed an anomalous contrast behavior. Within the ordered In 0.33 Ga 0.67 N monolayers In atoms are arranged in a periodic √3x√3R30° structure, resulting in pure In atomic columns in a GaN matrix along the cross‐sectional viewing directions of the wurtzite lattice (see Fig. 1). This has been experimentally confirmed ex‐situ by high resolution (S)TEM and in‐situ by reflection high‐energy electron diffraction (RHEED). Expecting intuitively a high contrast in high‐resolution STEM HAADF images of the ordered In 0.33 Ga 0.67 N monolayers, the experimental contrast between pure In and Ga atomic columns, however, was far below the Z 1.7 rule of thumb. To verify this result, we have performed frozen phonon simulations of a relaxed structure model consisting of a √3x√3R30° ordered In 0.33 Ga 0.67 N monolayer coherently embedded in a GaN matrix. Although the simulations agree with our experimental finding, even on a quantitative level, the explanation for the low contrast is far from intuitive. Even more surprisingly, for specimen thicknesses above 45 nm a contrast inversion occurs, i.e. the peak intensity of pure In atomic columns becomes lower than that of adjacent Ga atomic columns (see Fig. 2). Our frozen phonon simulations reveal that the origin for this anomalous contrast behavior lies in a strongly enhanced de‐channeling of the electron probe if it is positioned on the In atomic column of the ordered In 0.33 Ga 0.67 N monolayer. This in turn is caused by a complex interplay of increased disorder in the direct vicinity of the In atomic column in terms of chemistry (In atomic column is surrounded by material with a different atomic potential) and lattice periodicity (stronger local distortions around the In column because of differences in the In‐N vs. Ga‐N bond length).
Published Version (Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have