Action Filename Description Size Access License Resource Version
Show more files...


The fabrication of nanostructures using III-V semiconductors results in materials with different physical properties than the bulk materials. Their optical and electronic properties are tailored and developed on a large scale for the fabrication of optical and electronic devices (transistors, diodes, lasers). Among these materials, InGaAs/GaAs heterostructures are particular interesting as active layer in lasers for glass fibre assisted information transmission. These structures are strained due to the lattice mismatch between GaAs and InGaAs. Their geometry, the strain field and the local chemical composition have a direct impact on their luminescence properties. Among the analytical techniques, electron microscopy is an essential tool for nanoscale characterization. More specifically, dark-field imaging is particularly sensitive to both, local chemical composition and strain in the InGaAs/GaAs system. Although this type of contrast is well known, up to date, no quantitative study of the intensities in these strained semiconductor samples has been proposed. This quantitative aspect guided this work. In this work, we have analyzed quantitatively the evolution of contrasts in (200) dark-field images of planar InGaAs layers. Therefore we have optimized sample preparation, the observation techniques and the parameters influencing the (200) darkfield image intensities. By the exploration of the parameter space permitted we determined the possibilities and limits of the quantitative analysis of chemical composition from (200) dark-field images.We found an ideal sample thickness for this analysis of 60 to 100 nm. From a systematic analysis of the dark-field intensities as a function of excitation conditions we could evidence a sample bending of ± 1 mrad.To understand the observed contrasts quantitatively, we developed a simulation tool based on the finite element calculations of strain fields and their integration in a dynamic contrast calculation. These simulations were compared with the experimental observations. Based on these results, we developed a method for the quantitative analysis of the chemical composition of InxGa1-xAs/GaAs heterostructures. The obtained precision is better than Δx = 0.015 in an indium concentration range of 0.0 < x < 0.35 and a spatial resolution of about one nanometer. This method relies on the determination of calibration curves of chemical and strain contrasts. This method was also extended to the characterization of non-planar nanostructures. The analysis of planar InGaAs/GaAs structures is carried out under quasi-kinematical diffraction conditions by tilting the sample 14° from the [001] zone axis.In general, non-planar heterostructures cannot be tilted to these two-beam condition since projection effects strongly increase the width of the interfaces to be characterized. Experimental tilt series found that at a tilt angle of 4°, we obtain contrasts similar to the ones under two-beam conditions. These results were confirmed by simulations. Finally, the analysis of (200) dark-field images was extended to the case of quaternary InGaAsN for the simultaneous determination of indium and nitride composition. From (200) dark-field strain contrast curves we determined the latticeparameter with a precision of 0.1%. We could show that the experimental evolution of the chemical contrast is in contradiction with theory. The only possibility to approach theory and experiment consists in a modification of the structure factors.We propose the inclusion of static displacements for the definition of the unit cell decoration to take this effect into account. In the strained parts of the TEM samples, we observe a difference between the experimental and simulated intensities. As is the case of the InGaAsN layers, this discrepancy could be explained by local modifications of the structure factors due to atomic displacements and unit cell deformations (changes of symmetry) induced by strain.