Vertical GaAs nanowires on Si (111) substrate were grown by metal organic chemical vapor deposition via Au-catalyst vapor-liquid-solid mechanism. Stacking-faults-free zinc blende nanowires were realized by using AlGaAs/GaAs buffer layers and growing under the optimized conditions, that the alloy droplet act as a catalyst rather than an adatom collector and its size and composition would keep stable during growth. The stable droplet contributes to the growth of stacking-faults-free nanowires. Moreover, by using the buffer layers, epitaxial growth of well-aligned NWs was not limited by the misfit strain induced critical diameter, and the unintentional doping of the GaAs nanowires with Si was reduced.

No abstract

The dependence of crystal structure on contributions of adatom diffusion (ADD) and precursor direct impingement (DIM) was investigated for vapor-liquid-solid growth of InAs nanowires (NWs). The ADD contributions from the sidewalls and substrate surface can be changed by using GaAs NWs of different length as the basis for growing InAs NWs. We found that pure zinc-blende structure is favored when DIM contributions dominate. Moreover, without changing the NW diameter or growth parameters (such as temperature or V/III ratio), a transition from zinc-blende to wurtzite structure can be realized by increasing the ADD contributions. A nucleation model is proposed in which ADD and DIM contributions play different roles in determining the location and phase of the nucleus.

In recent research, deep neural network (DNN) has been used to solve the monaural source separation problem. According to the training objectives, DNN-based monaural speech separation is categorized into three aspects, namely masking, mapping and signal approximation (SA) based techniques. However, the performance of the traditional methods is not robust due to variations in real-world environments. Besides, in the vanilla DNN-based methods, the temporal information cannot be fully utilized. Therefore, in this paper, the long short-term memory (LSTM) neural network is applied to exploit the long-term speech contexts. Then, we propose the complex signal approximation (cSA) which is operated in the complex domain to utilize the phase information of the desired speech signal to improve the separation performance. The IEEE and the TIMIT corpora are used to generate mixtures with noise and speech interferences to evaluate the efficacy of the proposed method. The experimental results demonstrate the advantages of the proposed cSA-based LSTM RNN method in terms of different objective performance measures.

No abstract