Composition Determination of Semiconductors at Different Scattering Angles with the Help of Energy-Filtered STEM and Four-Dimensional STEM
In this study, the structural characterization of nanomaterials is performed by an extension of the established method which is used for quantitative STEM based on comparing the ADF-STEM image simulations and experiments. The limitations and capabilities of the method towards single atom accuracy ar...
|PDF Full Text
No Tags, Be the first to tag this record!
|In this study, the structural characterization of nanomaterials is performed by an extension of the established method which is used for quantitative STEM based on comparing the ADF-STEM image simulations and experiments. The limitations and capabilities of the method towards single atom accuracy are investigated. The method is further elaborated by determining more complex material systems as well as optimizing the imaging conditions to increase the accuracy.
Accessing the composition determination method with high resolution in a single atom accuracy and potential of estimating the accuracy of the method emerges the idea to optimize the experimental condition by conducting a simulation study. A performed simulation study on several critical imaging parameters leads to optimization of the imaging condition and enhances the accuracy of the method. The result reveals the critical role of two ADF-STEM imaging parameters: the semi-convergence angle of the impinging beam as well as the detection angle of ADF detectors. The former can be simply tuned experimentally by the choice of STEM condenser aperture, while the latter demands a fast, pixelated detector allowing the flexible choice of detection angle. The study however indicated that the optimum imaging condition differs by sample thickness and material systems. This becomes more apparent in the case of material systems containing light elements, e.g. GaNxAs1-x. Due to their low amount of protons, the light elements do not efficiently scatter to the commonly used detection angles in Z-contrast HAADF STEM micrographs. Accordingly, lower detection angles should be chosen. In contrast to the HAADF in which the image intensity is dominated by only elastically scattered electrons leading to a perfect match between ADF-STEM image simulations and experimental results so many other parameters play a role in the intensity of the micrograph at a low angular regime. In this study, as the main source of discrepancy between STEM experiments and simulations at low scattering angles, the effect of plasmon excitations on the angular distribution of the STEM intensities is investigated. The comparison of energy-filtered and unfiltered diffraction patterns indicates the significant effect of inelastic scattering at angles in the range of 0-40 mrad.
Considering the effect of plasmon excitation at low scattering angles, a further method is developed for the composition determination of material systems containing light elements at a low scattering angle based on EFSTEM. It is confirmed that the strain contrast induced by SADs causes higher scattering intensity at low scattering angles. Consequently, a material system containing SADs, i.e. GaNxAs1-x, is intentionally chosen to make the composition determination more reliable. There are also other sources of discrepancy between simulated and experimental STEM images such as neglecting the phonon correlations in image simulations, the effect of mistilt from the targeted crystalline zone-axis, and the existence of surface amorphous layers on ADF images. These errors are resolved either experimentally or by optimizing the detection angle with the help of a fast, pixelated detector. Here, Si as a model material is used to obtain the angular range with the perfect match between experiment and simulation. The method was applied on a sample containing GaNxAs1-x QWs embedded in GaAs barriers. The composition and the width of the QWs obtained by the new method are in very good agreement with XRD results.
The new advanced four-dimensional detector in hand enables recording a full diffraction pattern for every electron probe position. So far the camera is utilized as an annular detector with the flexibility of choosing the optimum detection angle. However, it can be further expanded in optimizing the composition determination by the flexible choice of regions on diffraction pattern for which the intensity suits the best for quantitative STEM. As Pennycook suggests, the incoherent nature of HAADF-STEM results in simply interpretable micrographs with independent information at every atomic column of the crystalline materials. However, this interpretability is lost in the conventional high resolution TEM micrographs due to the dynamical scattering and the coherent nature of the image formation. Hence, avoiding any coherent information in the diffraction pattern such as Laue zones may lead to more localized information in real space and consequently a higher accuracy in composition determination.
The combination of an in-column energy filter a fast pixelated detector is utilized to quantify the composition of different material systems at high accuracy. The fourdimensional detectors however, are capable of detecting the shift of the diffraction pattern's center-of-mass (COM) which is correlated to the local electric field within the crystal such as atomic potentials. The effect of quasi-elastic TDS on electric field determination is investigated in a simulation study showing a significant effect on electric field measurements. Future researches can be focused on utilizing the combination of EFSTEM and 4D-STEM to investigate the effect of another source of inelastic scattering, i.e. plasmon excitation, on COM shift.