Optimizing Performance in Supercapacitors through Surface Decoration of Bismuth Nanosheets.

Bismuth (Bi) exhibits a high theoretical capacity, excellent electrical conductivity properties, and remarkable interlayer spacing, making it an ideal electrode material for supercapacitors. However, during the charge and discharge processes, Bi is prone to volume expansion and pulverization, resulting in a decline in the capacitance. Deposition of a nonmetal on its surface is considered an effective way to modulate its morphology and electronic structure. Herein, we employed the chemical vapor deposition technique to fabricate Se-decorated Bi nanosheets on a nickel foam (NF) substrate. Various characterizations indicated that the deposition of Se on Bi nanosheets regulated their surface morphology and chemical state, while sustaining their pristine phase structure. Electrochemical tests demonstrated that Se-decorated Bi nanosheets exhibited a 51.1% improvement in capacity compared with pristine Bi nanosheets (1313 F/g compared to 869 F/g at a current density of 5 A/g). The energy density of the active material in an assembled asymmetric supercapacitor could reach 151.2 Wh/kg at a power density of 800 W/kg. These findings suggest that Se decoration is a promising strategy to enhance the capacity of the Bi nanosheets.

Automatic Sparse Connectivity Learning for Neural Networks.

Since sparse neural networks usually contain many zero weights, these unnecessary network connections can potentially be eliminated without degrading network performance. Therefore, well-designed sparse neural networks have the potential to significantly reduce the number of floating-point operations (FLOPs) and computational resources. In this work, we propose a new automatic pruning method-sparse connectivity learning (SCL). Specifically, a weight is reparameterized as an elementwise multiplication of a trainable weight variable and a binary mask. Thus, network connectivity is fully described by the binary mask, which is modulated by a unit step function. We theoretically prove the fundamental principle of using a straight-through estimator (STE) for network pruning. This principle is that the proxy gradients of STE should be positive, ensuring that mask variables converge at their minima. After finding Leaky ReLU, Softplus, and identity STEs can satisfy this principle, we propose to adopt identity STE in SCL for discrete mask relaxation. We find that mask gradients of different features are very unbalanced; hence, we propose to normalize mask gradients of each feature to optimize mask variable training. In order to automatically train sparse masks, we include the total number of network connections as a regularization term in our objective function. As SCL does not require pruning criteria or hyperparameters defined by designers for network layers, the network is explored in a larger hypothesis space to achieve optimized sparse connectivity for the best performance. SCL overcomes the limitations of existing automatic pruning methods. Experimental results demonstrate that SCL can automatically learn and select important network connections for various baseline network structures. Deep learning models trained by SCL outperform the state-of-the-art human-designed and automatic pruning methods in sparsity, accuracy, and FLOPs reduction.

Building Low-Cost, High-Performance Flexible Photodetector Based on Tetragonal Phase VO2 (A) Nanorod Networks.

We present a straightforward and cost-effective method for the fabrication of flexible photodetectors, utilizing tetragonal phase VO2 (A) nanorod (NR) networks. The devices exhibit exceptional photosensitivity, reproducibility, and stability in ambient conditions. With a 2.0 V bias voltage, the device demonstrates a photocurrent switching gain of 1982% and 282% under irradiation with light at wavelengths of 532 nm and 980 nm, respectively. The devices show a fast photoelectric response with rise times of 1.8 s and 1.9 s and decay times of 1.2 s and 1.7 s for light at wavelengths of 532 nm and 980 nm, respectively. In addition, the device demonstrates exceptional flexibility across large-angle bending and maintains excellent mechanical stability, even after undergoing numerous extreme bending cycles. We discuss the electron transport process within the nanorod networks, and propose a mechanism for the modulation of the barrier height induced by light. These characteristics reveal that the fabricated devices hold the potential to serve as a high-performance flexible photodetector.