Impurity-mediated near-infrared (NIR) photoresponse in silicon is of great interest for photovoltaics and photodetectors. In this paper, we have fabricated a series of n+/p photodetectors with hyperdoped silicon prepared by ion-implantation and femtosecond pulsed laser. These devices showed a remarkable enhancement on absorption and photoresponse at NIR wavelengths. The device fabricated with implantation dose of 1014 ions/cm2 has exhibited the best performance. The proposed method offers an approach to fabricate low-cost broadband silicon-based photodetectors.

No abstract

In this paper, we investigate the dependence of NIR absorption of hyper-doped silicon on different sulfur and nitrogen doping ratio. With different molecular proportion of N 2 and SF 6 background gas, femtosecond laser irradiation was used to implant co-doping of sulfur and nitrogen into silicon. The hyper-doped silicon presents high absorption properties in NIR and visible range. The results of first-principles calculations demonstrate the high absorption in NIR is ascribed to the induced impurity energy levels in hyper-doped silicon. The nitrogen doping process improves the crystallinity in the doped layer because the doped nitrogen repairs defects in silicon lattices. Given the thermal stability of nitrogen, the co-doping dopants limit the diffusion of sulfur during the annealing process. The co-doping process proposed in this paper provides a method to fabricate high performance NIR silicon optoelectronic device.

In the case where prior knowledge such as frequency, bandwidth, and modulation mode is unknown, the RF(radio frequency) signal from the receiver must be demodulated to a bitstream. In this paper, unsupervised clustering of logical channels such as communication system signals, service synchronization or physical layer signaling using deep learning methods is analyzed. After analysis, it is found that the current channel clustering mainly faces two problems: one is the need to manually annotate the key features of the signal dock; Second, different representations of signals, such as square spectrum and time-frequency graph, contain different feature information. At present, most of them are only analyzed for specific signal representations.Therefore, we propose to use the mixed-domain attention mechanism to automatically locate the areas that need to be focused on instead of manual participation. At the same time, use product layer for feature fusion of different signal representations (Identical/orthogonal format, Amplitude/Phase format, Time and Frequency diagram) to improve the recognition accuracy, which is a feasible direction for future research.

Automatic Modulation Recognition (AMR) is a fundamental research topic in the field of signal processing and wireless communication, which has widespread applications in cognitive radio, non-collaborative communication, etc. However, current AMR methods are mostly based on unimodal inputs, which suffer from incomplete information and local optimization. In this paper, we focus on the modality utilization in AMR. The proxy experiments show that different modalities achieve a similar recognition effect in most scenarios, while the personalities of different inputs are complementary to each other for particular modulations. Therefore, we mine the universal and complementary characteristics of the modality data in the domain-agnostic and domain-specific aspects, yielding the Universal and Complementary subspaces accordingly (dubbed as UCNet). To facilitate the subspace construction, we propose universal and complementary losses accordingly, where the former minimizes the heterogeneous feature gap by an adversarial constraint and the latter consists of an orthogonal constraint between universal and complementary features. The extensive experiments on the RadioML2016.10A dataset demonstrate the effectiveness of UCNet, which has achieved the highest recognition accuracy of 93.2% at 10 dB, and the average accuracy is 92.6% at high SNR greater than zero.