The photoluminescence spectrum and absorption spectrum as well as resonant Raman scattering tuned by pressure in heavily doped CdS have been studied. The pressure coefficients of shallow levels (I~and I,) and deep levels (G, and R) are measured. A new optical-absorption method under pressure was developed and the shift of band gap with pressure of CdS in the direct-band-gap wurtzite phase and indirect-band-gap rock-salt phase was measured. From the experimental results of resonant Raman scattering tuned by pressure we conclude that at about 27 kbar CdS undergoes a complete phase transition from the wurtzite phase to the rock-salt phase. After the phase transition there is no two-phase mixture, and the material still has semiconductor characteristics.

New experimental results on Raman scattering from porous silicon and silicon and gallium arsenide nanocrystals are reported. In all of these systems, almost all vibrational modes become Raman active and are remarkably soft. A carrier-induced strain model is proposed to explain the optical properties of these nanocrystal systems. According to this carrier-induced strain model, the selection rule of crystal momentum conservation for Raman scattering is greatly relaxed in Si and GaAs nanocrystals due to the dilatation strain caused by coupling of excited free carriers with the particle lattice and the optical properties of such systems are dominated by multiphonon assisted free-electron transition processes.

No abstract

No abstract

In recent years, with the rapid development of computer vision, increasing attention has been paid to remote sensing image scene classification. To improve the classification performance, many studies have increased the depth of convolutional neural networks (CNNs) and expanded the width of the network to extract more deep features, thereby increasing the complexity of the model. To solve this problem, in this paper, we propose a lightweight convolutional neural network based on attention-oriented multi-branch feature fusion (AMB-CNN) for remote sensing image scene classification. Firstly, we propose two convolution combination modules for feature extraction, through which the deep features of images can be fully extracted with multi convolution cooperation. Then, the weights of the feature are calculated, and the extracted deep features are sent to the attention mechanism for further feature extraction. Next, all of the extracted features are fused by multiple branches. Finally, depth separable convolution and asymmetric convolution are implemented to greatly reduce the number of parameters. The experimental results show that, compared with some state-of-the-art methods, the proposed method still has a great advantage in classification accuracy with very few parameters.