MANY-BEAM SIMULATIONS AND OBSERVATIONS OF LARGE-ANGLE CONVERGENT-BEAM ELECTRON-DIFFRACTION IMAGING OF CRYSTAL DEFECTS

Many-beam dynamical simulations and observations have been made for large-angle convergent-beam electron diffraction (LACBED) imaging of crystal defects, such as stacking faults and dislocations. The simulations are based on a general matrix formulation of dynamical electron diffraction theory by Peng and Whelan, and the results are compared with experimental LACBED images of stacking faults and dislocations of Si angle crystals. Excellent agreement is achieved.

INVESTIGATION OF GAAS/SI MATERIAL BY X-RAY DOUBLE-CRYSTAL DIFFRACTION

GaAs epilayer films on Si substrates grown by molecular-beam epitaxy were investigated by the x-ray double-crystal diffraction method. The rocking curves were recorded for different diffraction vectors of samples. The results show that the unit-cell volumes of GaAs epilayers are smaller than that of the GaAs bulk material. The strained-layer superlattice buffer layer can improve the quality of the film, especially in the surface lamella. The parameter W' = W(expt)/(square-root \gamma-h\/gamma-0/sin 2-theta-B) is introduced to describe the quality of different depths of epilayers. As the x-ray incident angle is increased, W' also increases, that is, the quality of the film deteriorates with increasing penetration distance of the x-ray beam. Therefore, W' can be considered as a parameter that describes the degree of perfection of the epilayer along the depth below the surface. The cross-section transmission electron microscopy observations agree with the results of x-ray double-crystal diffraction.

CONVERGENT-BEAM ELECTRON-DIFFRACTION STUDY OF GE0.5SI0.5/SI STRAINED-LAYER SUPERLATTICES GROWN BY MOLECULAR-BEAM EPITAXY

Side bands due to purely composition and combined composition-strain modulation in plan-view specimens of a nominally Ge0.5Si0.5(5nm)/Si(25nm) superlattice have been obtained by large-angle convergent-beam electron diffraction. The intensities of the side bands have been calculated from a periodic tension-compression model of the superlattice bilayer using the kinematical theory of electron diffraction. Accurate values of elastic strains in the bilayer and of the Ge content can be obtained in this way.

Structures of an asymmetrically coupled double-well superlattice by double-crystal X-ray diffraction

An asymmetrically coupled (GaAs/AlAs/GaAs/AlAs)/GaAs (001) double-well supperlattice is studied by HRDCD (high resolution double-crystal X-ray diffractometry). The intensity of satellite peaks is modulated by wave packet of different sublayers. In the course of simulation, the satellite peaks in the vicinity of the node points of wave packet are very informative for precise determination of sublayer thickness and for improving accuracy.

Structural studies of NixFe100-x/Mo magnetic multilayers by x-ray small-angle reflection and high-angle diffraction

Magnetic multilayers [NixFe100-x/Mo-30] grown by dc-magnetron sputtering were investigated by x-ray small-angle reflection and high-angle diffraction. Structural parameters of the multilayers such as the superlattice periods, the interfacial roughness, and interplane distance were obtained. It was found that for our NixFe100-x/Mo system, the Mo layer has bcc structure with [110] preferential orientation, while the preferential orientation of the NixFe100-x layer changes from a fee structure with [111] preferential orientation to a bcc structure with [110] preferential orientation with decreasing values of x. An intermixing layer located in the interlayer region between the NixFe100-x and Mo layers exists in the multilayers, and its thickness is almost invariant with respect to an increase of Mo layer thickness and/or a decrease of x in the region of x greater than or equal to 39. The thickness of the intermixing layer falls to zero when x less than or equal to 23.

Electron diffraction and photoemission studies of clean and water-covered Si(5,5,12) and Si(112) surfaces

The structure of silicon surfaces in the orientation range (113)-(5,5,12)-(337)-(112) has been investigated using high resolution LEED and photoemission both on a spherical and on flat samples. We find that Si(5,5,12) [5.3 degrees from (113) and 0.7 degrees from (937)] is the only stable orientation between (113) and (111) and confirm the result of Baski et al. [Science 269, 1556 (1995)] that it has a 2 x 1 superstructure with a very large unit cell of 7.68 x 53.5 Angstrom(2). Adsorption measurements of water on Si(5,5,12) yield a mobile precursor kinetics with two kinds of regions saturating at 0.25 and 0.15 ML which are related to adsorption on different sites. Using these results, a modified structure model is proposed. Surfaces between (113) and (5,5,12) separate into facets of these two orientations; between (5,5,12) and (112), they separate into (5,5,12) and (111) facets. (337) facets in this range may be considered as defective (5,5,12) facets.

X-ray-diffraction study of quasipseudomorphic ErSi1.7 layers formed by channeled ion-beam synthesis

ErSi1.7 layers with high crystalline quality (chi(min) of Er is 1.5%) have been formed by 90 keV Er ion implantation to a dose of 1.6X10(17)/cm(2) at 450 degrees C using channeled implantation. The perpendicular and parallel elastic strain e(perpendicular to)=-0.94%+/-0.02% and e(parallel to)=1.24%+/-0.08% of the heteroepitaxial erbium silicide layers have been measured with symmetric and asymmetric x-ray reflections using a double-crystal x-ray diffractometer. The deduced tetragonal distortion e(T(XRD))=e(parallel to)-e(perpendicular to)=2.18%+/-0.10%, which is consistent with the value e(T(RBS))2.14+/-0.17% deduced from the Rutherford backscattering and channeling measurements. The quasipseudomorphic growth of the epilayer and the stiffness along a and c axes of the epilayer deduced from the x-ray diffraction are discussed.

A surface structure model for Si(337)

The surface reconstruction on Si(337) at room temperature has been studied by low energy electron diffraction (LEED). It has been found that: (I) the Si(337) gave a clear LEED pattern which indicates the existence of another high index stable surface besides Si(113); (II) in addition to a strong Si(337)-(1 X 1), we observed for the first time a (2 X 1) LEED pattern indicating a surface reconstruction along the [1(1) over bar0$] direction; (III) a surface model has been proposed for the observed Si(337)-(2 X 1) structure.

Hydrogen adsorption induced surface reconstructions on Si(113) studied by LEED

Surface reconstructions on Si(113) induced by dissociated hydrogen adsorption have been studied using low energy electron diffraction (LEED). It has been found that: (1) at 300 K and 80 K temperatures, with the increase of hydrogen coverage on the surface, the (3 x 1) phase transferred continuously into a hydrogen saturated (1 x 1)-2H phase; (2) flashing of the (1 x 1)-2H surface at about 1100 degrees C resulted in a complete new phase of(1 x 3) and further annealing of the sample at 1250 degrees C gave back the starting surface of (3 x 1); (3) saturated hydrogen adsorption at a sample temperature of 700 degrees C resulted in a stable new phase of(1 x 2)-H and further saturation doses of hydrogen at other temperatures below 700 degrees C did not change the (1 x 2) LEED pattern; (4) annealing of the (I x 2)-H surface in the same manner as (2) gave similar results.

Study of double barrier superlattice by synchrotron radiation and double-crystal x-ray diffraction

An (A1As/GaAs/A1As/A1GaAs)/GaAs(001) double-barrier superlattice grown by molecular beam epitaxy (MBE) is studied by combining synchrotron radiation and double-crystal x-ray diffraction (DCD). The intensity of satellite peaks is modulated by the wave function of each sublayer in one superlattice period. Simulated by the x-ray dynamical diffraction theory, it is discovered that the intensity of the satellite peaks situated near the modulating wave node point of each sublayer is very sensitive to the variation of the layer structural parameters, The accurate layer thickness of each sublayer is obtained with an error less than 1 Angstrom. Furthermore, x-ray kinematical diffraction theory is used to explain the modulation phenomenon. (C) 1996 American Institute of Physics.

A method of Biomimetic Pattern Recognition for face recognition

A new method of face recognition, based on Biomimetic Pattern Recognition and Multi-Weights Neuron Network, had been proposed. A model for face recognition that is based on Biomimetic Pattern Recognition had been discussed, and a new method of facial feature extraction also had been introduced. The results of experiments with BPR and K-Nearest Neighbor Rules showed that the method based on BPR can eliminate the error recognition of the samples of the types that not be trained, the correct rate is also enhanced.

Biomimetic (topological) pattern recognition - A new model of pattern recognition theory and its application

A new theoretical model of Pattern Recognition principles was proposed, which is based on "matter cognition" instead of "matter classification" in traditional statistical Pattern Recognition. This new model is closer to the function of human being, rather than traditional statistical Pattern Recognition using "optimal separating" as its main principle. So the new model of Pattern Recognition is called the Biomimetic Pattern Recognition (BPR)(1). Its mathematical basis is placed on topological analysis of the sample set in the high dimensional feature space. Therefore, it is also called the Topological Pattern Recognition (TPR). The fundamental idea of this model is based on the fact of the continuity in the feature space of any one of the certain kinds of samples. We experimented with the Biomimetic Pattern Recognition (BPR) by using artificial neural networks, which act through covering the high dimensional geometrical distribution of the sample set in the feature space. Onmidirectionally cognitive tests were done on various kinds of animal and vehicle models of rather similar shapes. For the total 8800 tests, the correct recognition rate is 99.87%. The rejection rate is 0.13% and on the condition of zero error rates, the correct rate of BPR was much better than that of RBF-SVM.

An improved BP algorithm for pattern recognition

An improved BP algorithm for pattern recognition is proposed in this paper. By a function substitution for error measure, it resolves the inconsistency of BP algorithm for pattern recognition problems, i.e. the quadratic error is not sensitive to whether the training pattern is recognized correctly or not. Trained by this new method, the computer simulation result shows that the convergence speed is increased to treble and performance of the network is better than conventional BP algorithm with momentum and adaptive step size.

A new development on ANN in China - Biomimetic pattern recognition and multi weight vector neurons

A new model of pattern recognition principles-Biomimetic Pattern Recognition, which is based on "matter cognition" instead of "matter classification", has been proposed. As a important means realizing Biomimetic Pattern Recognition, the mathematical model and analyzing method of ANN get breakthrough: a novel all-purpose mathematical model has been advanced, which can simulate all kinds of neuron architecture, including RBF and BP models. As the same time this model has been realized using hardware; the high-dimension space geometry method, a new means to analyzing ANN, has been researched.