Search

Field-effect transistors (FETs) switched by quantum band-to-band tunneling (BTBT) mechanism, rather than conventional thermionic emission mechanism, are emerging as an exciting device candidate for future ultralow-power electronics due to their exceptional electronic properties of subthermionic subthreshold swing. However, fundamental limitations in drive current have hindered such technology encountering for high-performance and high-speed operations, especially for silicon-based device. Here, we demonstrate a novel pathway of integrating tunneling and thermionic emission mechanisms together, to circumvent their respective limitation and design a hybrid adaptively modulated FET (HamFET) that orients power saving and performance enhancement simultaneously. This transistor architecture, utilizing a nested source configuration without cost or area penalties, exhibits both ultrasteep (subthermionic) subthreshold swing and the largest “on” and “off” current ratio over the state-of-the-art tunneling transistors. Our design methodology of hybrid switching mechanism is also applicable to other mechanism, material, and architecture systems, opening the doors to a range of high-speed application opportunities for ultralow-power but performance-insufficient electronics.

In this paper, a vertically stacked nanosheet gate-all-around field-effect transistor (GAA-NSFET) as a label-free biosensor has been proposed and investigated. The influences of different biomolecules on the biosensor’s electrical properties are analyzed, and the proposed biosensor exhibits good sensitivity for both neutral and charged biomolecules. Furthermore, a deep sensitivity analysis is performed to evaluate the sensing ability of biosensors with different channel structures. The sensitivity variation of biosensors is analyzed in terms of subthreshold swing ( $SS $ ), threshold voltage ( $V_{\text {th}}$ ), and current switching ratio ( $I_{\text {on}} / I_{\text {off}}$ ). The results show that GAA-NSFET-based biosensor obtains the best sensitivity compared with other biosensors (including nanowire FET-based biosensor and vertically stacked nanowire FET-based biosensor) due to its larger channel width and multi-channels. Also, the influences of filling position on sensitivity are analyzed in various cases for the proposed biosensor. Its sensitivity depends on the filling amounts of biomolecules instead of the filling position. Finally, a status map is presented, which plots the sensitivity of some important works in biosensing application along with the sensitivity of the proposed biosensor, and GAA-NSFET-based biosensor is more sensitive compared with those works in terms of $I_{\text {on}}/I_{\text {off}}$ sensitivity.

Increasing demands for mass storage and new paradigm computing ask for non-volatile memories that can meet reliability requirements. Hf0.5Zr0.5O2-based (HZO) memory has attracted growing attention due to its excellent CMOS compatibility. This letter investigated the time dependent dielectric breakdown (TDDB) of HZO ferroelectric under both forward and reverse stress conditions, which is relevant to the memory’s practical operation. The key similarities and the differences for both breakdown conditions have been identified and the underlying mechanism is explored. It is found that the pre-existing oxygen vacancies near the bottom electrode play the key role and all the observed phenomenon can be explained. Therefore, the precise control of these pre-existing oxygen vacancies can be critical for future TDDB improvement.

The RRAM-based array is one of the most promising core functional primitives to accelerate the inference process of neural networks. However, the stress-induced disturbance can cause a significant accuracy drop during inference process where input vectors with different voltage levels are fed to the device. This kind of disturb can hardly be avoided by optimizing the fabrication process. Here, we investigate this phenomenon based on TaOx-based devices with different electrodes. The results indicate that the stress-induced disturb mainly appears in the intermediate resistance states when the voltage is applied on the device. The simulation result by COMSOL reveals the relationship between read disturb and electric field. Therefore, we propose a nonlinear weight quantification method to mitigate read disturb effect on inference accuracy by reducing the number of devices in intermediate resistance states. The simulation results based on the fully-connected networks for MNIST recognition indicate that stress disturb phenomenon can be well suppressed by nonlinear weight quantification compared with the conventional linear quantification method, which will advance the application of the RRAM-based accelerator.

Wireless sensor networks play an important role in Internet of Things systems and services but are prone and vulnerable to poor communication channel quality and network attacks. In this paper we are motivated to propose resilient routing algorithms for wireless sensor networks. The main idea is to exploit the link reliability along with other traditional routing metrics for routing algorithm design. We proposed firstly a novel deep-learning based link prediction model, which jointly exploits Weisfeiler-Lehman kernel and Dual Convolutional Neural Network (WL-DCNN) for lightweight subgraph extraction and labelling. It is leveraged to enhance self-learning ability of mining topological features with strong generality. Experimental results demonstrate that WL-DCNN outperforms all the studied 9 baseline schemes over 6 open complex networks datasets. The performance of AUC (Area Under the receiver operating characteristic Curve) is improved by 16% on average. Furthermore, we apply the WL-DCNN model in the design of resilient routing for wireless sensor networks, which can adaptively capture topological features to determine the reliability of target links, especially under the situations of routing table suffering from attack with varying degrees of damage to local link community. It is observed that, compared with other classical routing baselines, the proposed routing algorithm with link reliability prediction module can effectively improve the resilience of sensor networks while reserving high-energy-efficiency.

Network virtualization allows multiple isolated virtual networks (VNs) to share the same substrate network (SN). VN embedding (VNE) algorithms can efficiently allocate the limited SN resources to VNs and assign a unique identifier to each VN. However, the fixed bit width of VN identifier in the packet header limits the number of VNs, and extending the bit width leads to the increase of the network traffic. In this paper, we consider the label-combination method to generate VN identifiers by combining the link-grained labels with location information. This method requires the efficient allocation of labels, but the existing VN embedding works only consider the CPU and bandwidth resources. To address this issue, we propose a novel embedding model that considers the label, CPU and bandwidth resource constraints. Furthermore, two window-based heuristic algorithms called VNE-LIA and VNE-iLIA using the greedy algorithm and the proximity principle are presented to solve the VNE problem. The simulation experiments show that our proposed algorithms increase the number of VN identifiers and the revenue to cost ratio under the different resource conditions of SN.