Quantitative Evaluation of the Interfaces in III/V Semiconductors with Scanning Transmission Electron Microscopy
In order to understand the principles of HAADF imaging and also to implement the contrast simulations efficiently and effectively, simulations about HAADF imaging are carried out. The results clearly show that TDS can significantly influence the collected intensity at the detector. In order to in...
Keine Tags, Fügen Sie den ersten Tag hinzu!
|Zusammenfassung:||In order to understand the principles of HAADF imaging and also to implement the contrast simulations efficiently and effectively, simulations about HAADF imaging are carried out. The results clearly show that TDS can significantly influence the collected intensity at the detector. In order to include TDS during the simulation in this study, frozen phonon approach can be applied. Frozen phonon considers TDS through averaging HAADF images calculated from different atom configurations. To save computing time and resources, the influences of the number of phonon configurations on simulated image intensities are investigated. It is proven that 15 phonon configurations is enough to pro- vide a robust result for the multi slice simulation within a reasonable computing time. In addition, intensities of group III and group V atomic columns together with back- ground positions for GaAa, GaP, (Ga0.5In0.5)P are calculated to investigate the origin of background intensity and its influences on individual sublattice intensities. Two quantitative evaluation methods, namely Q-method and BIMS, are introduced in this study. They have been successfully applied to characterize material systems with both the non-chemically sensitive background and the chemically sensitive background, respectively. With respect to the (GaIn)As/GaAs material systems with non-chemically sensitive backgrounds, Q-method is applied. To obtain trustable results, a 2-dimensional thickness gradient correction is performed, and the influences of factors (such as cross talk, strain relaxation, projection effect), which can significantly affect the characterization, are also discussed. To make fair comparisons between different samples, two criteria to determine the interface abruptness, namely 90/10 evaluation approach and error function fitting approach, are given and applied for all the samples investigated in the current study. In addition, the chemical homogeneity of the QW is also introduced as a characteristic for evaluation. Followed by the introduction of these fundamental properties, the influences of growth conditions, like growth temperature, growth interruption temperature and time, on the chemical composition and on the interface abruptness are investigated. For samples with the same growth and growth interruption temperature, it can be stated that the growth temperature of (GaIn)As can strikingly influence the indium dis- tribution at both the QW and the interface. Growth at 625 ◦C leads to an inhomogeneous indium distribution at the QW as well as an intermixed interface. Meanwhile, the ap- plication of growth interruptions can significantly homogenize the indium distribution at both the QW and the interface. Through the comparison with simulations, the bottom interfaces of (GaIn)As grown on GaAs are shown to be abrupt. Therefore, for samples with different growth and growth interruption temperatures, only the top interfaces of (GaIn)As are investigated. Besides the above mentioned results, it can be concluded that an abrupt interface, generated at low temperature (525 ◦C), can be easily degenerated into an intermixing interface with a high interruption temperature of 625 ◦C. Similarly, an intermixed interface, formed at 625 ◦C, can also be improved by growth interruption at 525 ◦C. With respect to the (GaIn)P/GaAs material systems with chemically sensitive back- grounds, the above mentioned method can not be applied any more. As a result, BIMS, based on Q-method, is developed. The image background intensity and its influences on quantitative evaluation of the chemical composition are presented. It is found that the chemical composition characterization across the interface is impossible, if the image background intensity is not subtracted from the original image. With this method, the atomic resolution after background intensity subtraction can be kept for further evalua- tion. As expected, composition depth profiles and interface morphology strongly depend on the growth conditions. A reduction of the growth temperature from 625 ◦C to 525 ◦C can lead to a more abrupt heterointerface. The introduction of a GaP interlayer can improve the interface morphology. Nonetheless, this interlayer also results in an increased separation between the constituent QW and the barrier. With this method, the existence of an island-like structure at the interface can be shown and analyzed quantitatively. In the current study, the determination of the chemical composition map is based on a linear assumption between the collected intensity of the atomic column and its chemical concentration. Although the quantitative evaluation of the interface morphology and chemical homogeneity of the QW is hardly influenced by the assumption, it is still of great importance to derive the exact chemical composition of each atomic column, especially at the interface, in order to better control the lattice constant and the band gap. To aid these processes, a massive number of simulations of (GaIn)As super cells with different indium compositions needs to be carried out. Then a five-dimensional database can be created, which is composed of x, y, z space dimensions, one compositional dimension as well as one dimension of the collection angle covering both high and low angle ranges. Based on the database, the relationship between the sublattice intensity and the corresponding chemical composition can be derived and applied for the calculation of the chemical composition of individual atomic columns with great accuracy. In this study, only electron signals from high angles are made use of to carry out the quantitative evaluation. With the database, electrons scattered at low angles can also be used for the analysis. It is found that the image background intensity can significantly influence the quanti- tative evaluation of chemical composition maps. In fact, the image background intensity can also be used to determine the local sample thickness and the chemical composition of corresponding atomic columns, since it depends on both the thickness and the average atomic number of the crystal. Therefore, the calculated background intensity map in this study can be converted into a thickness map or a chemical composition map. To fulfill the purpose, background intensities of experimental images needs to be compared with those of the five-dimensional database. Then, a thickness map or a chemical composition map can be derived.|