Heteroepitaxial Writing of Silicon-on-Sapphire Nanowires.

The heteroepitaxial growth of crystal silicon thin films on sapphire, usually referred to as SoS, has been a key technology for high-speed mixed-signal integrated circuits and processors. Here, we report a novel nanoscale SoS heteroepitaxial growth that resembles the in-plane writing of self-aligned silicon nanowires (SiNWs) on R-plane sapphire. During a low-temperature growth at <350 °C, compared to that required for conventional SoS fabrication at >900 °C, the bottom heterointerface cultivates crystalline Si pyramid seeds within the catalyst droplet, while the vertical SiNW/catalyst interface subsequently threads the seeds into continuous nanowires, producing self-oriented in-plane SiNWs that follow a set of crystallographic directions of the sapphire substrate. Despite the low-temperature fabrication process, the field effect transistors built on the SoS-SiNWs demonstrate a high on/off ratio of >5 × 104 and a peak hole mobility of >50 cm2/V·s. These results indicate the novel potential of deploying in-plane SoS nanowire channels in places that require high-performance nanoelectronics and optoelectronics with a drastically reduced thermal budget and a simplified manufacturing procedure.

In-plane epitaxial growth of silicon nanowires and junction formation on Si(100) substrates.

Growing self-assembled silicon nanowires (SiNWs) into precise locations represents a critical capability to scale up SiNW-based functionalities. We here report a novel epitaxy growth phenomenon and strategy to fabricate orderly arrays of self-aligned in-plane SiNWs on Si(100) substrates following exactly the underlying crystallographic orientations. We observe also a rich set of distinctive growth dynamics/modes that lead to remarkably different morphologies of epitaxially grown SiNWs/or grains under variant growth balance conditions. High-resolution transmission electron microscopy cross-section analysis confirms a coherent epitaxy (or partial epitaxy) interface between the in-plane SiNWs and the Si(100) substrate, while conductive atomic force microscopy characterization reveals that electrically rectifying p-n junctions are formed between the p-type doped in-plane SiNWs and the n-type c-Si(100) substrate. This in-plane epitaxy growth could provide an effective means to define nanoscale junction and doping profiles, providing a basis for exploring novel nanoelectronics.

Adaptive Synaptic Scaling in Spiking Networks for Continual Learning and Enhanced Robustness.

Synaptic plasticity plays a critical role in the expression power of brain neural networks. Among diverse plasticity rules, synaptic scaling presents indispensable effects on homeostasis maintenance and synaptic strength regulation. In the current modeling of brain-inspired spiking neural networks (SNN), backpropagation through time is widely adopted because it can achieve high performance using a small number of time steps. Nevertheless, the synaptic scaling mechanism has not yet been well touched. In this work, we propose an experience-dependent adaptive synaptic scaling mechanism (AS-SNN) for spiking neural networks. The learning process has two stages: First, in the forward path, adaptive short-term potentiation or depression is triggered for each synapse according to afferent stimuli intensity accumulated by presynaptic historical neural activities. Second, in the backward path, long-term consolidation is executed through gradient signals regulated by the corresponding scaling factor. This mechanism shapes the pattern selectivity of synapses and the information transfer they mediate. We theoretically prove that the proposed adaptive synaptic scaling function follows a contraction map and finally converges to an expected fixed point, in accordance with state-of-the-art results in three tasks on perturbation resistance, continual learning, and graph learning. Specifically, for the perturbation resistance and continual learning tasks, our approach improves the accuracy on the N-MNIST benchmark over the baseline by 44% and 25%, respectively. An expected firing rate callback and sparse coding can be observed in graph learning. Extensive experiments on ablation study and cost evaluation evidence the effectiveness and efficiency of our nonparametric adaptive scaling method, which demonstrates the great potential of SNN in continual learning and robust learning.

Brain-inspired multimodal hybrid neural network for robot place recognition.

Place recognition is an essential spatial intelligence capability for robots to understand and navigate the world. However, recognizing places in natural environments remains a challenging task for robots because of resource limitations and changing environments. In contrast, humans and animals can robustly and efficiently recognize hundreds of thousands of places in different conditions. Here, we report a brain-inspired general place recognition system, dubbed NeuroGPR, that enables robots to recognize places by mimicking the neural mechanism of multimodal sensing, encoding, and computing through a continuum of space and time. Our system consists of a multimodal hybrid neural network (MHNN) that encodes and integrates multimodal cues from both conventional and neuromorphic sensors. Specifically, to encode different sensory cues, we built various neural networks of spatial view cells, place cells, head direction cells, and time cells. To integrate these cues, we designed a multiscale liquid state machine that can process and fuse multimodal information effectively and asynchronously using diverse neuronal dynamics and bioinspired inhibitory circuits. We deployed the MHNN on Tianjic, a hybrid neuromorphic chip, and integrated it into a quadruped robot. Our results show that NeuroGPR achieves better performance compared with conventional and existing biologically inspired approaches, exhibiting robustness to diverse environmental uncertainty, including perceptual aliasing, motion blur, light, or weather changes. Running NeuroGPR as an overall multi-neural network workload on Tianjic showcases its advantages with 10.5 times lower latency and 43.6% lower power consumption than the commonly used mobile robot processor Jetson Xavier NX.

Detecting out-of-distribution samples via variational auto-encoder with reliable uncertainty estimation.

Variational autoencoders (VAEs) are influential generative models with rich representation capabilities from the deep neural network architecture and Bayesian method. However, VAE models have a weakness that assign a higher likelihood to out-of-distribution (OOD) inputs than in-distribution (ID) inputs. To address this problem, a reliable uncertainty estimation is considered to be critical for in-depth understanding of OOD inputs. In this study, we propose an improved noise contrastive prior (INCP) to be able to integrate into the encoder of VAEs, called INCPVAE. INCP is scalable, trainable and compatible with VAEs, and it also adopts the merits from the INCP for uncertainty estimation. Experiments on various datasets demonstrate that compared to the standard VAEs, our model is superior in uncertainty estimation for the OOD data and is robust in anomaly detection tasks. The INCPVAE model obtains reliable uncertainty estimation for OOD inputs and solves the OOD problem in VAE models.

Sorafenib inhibits LPS-induced inflammation by regulating Lyn-MAPK-NF-kB/AP-1 pathway and TLR4 expression.

Sorafenib is an anti-tumor drug widely used in clinical treatment, which can inhibit tyrosine kinase receptor on cell surface and serine/threonine kinase in downstream Ras/MAPK cascade signaling pathway of cells. Tyrosine kinase phosphorylation plays an important role in inflammatory mechanism, such as TLR4 tyrosine phosphorylation, MAPK pathway protein activation, and activation of downstream NF-кB. However, the effects of sorafenib on LPS-induced inflammatory reaction and its specific mechanism have still remained unknown. We found that sorafenib inhibited the phosphorylation of tyrosine kinase Lyn induced by LPS, thereby reducing the phosphorylation level of p38 and JNK, inhibiting the activation of c-Jun and NF-κB, and then inhibiting the expression of inflammatory factors IL-6, IL-1ß, and TNF-α. Furthermore, sorafenib also decreased the expression of TLR4 on the macrophage membrane to inhibit the expression of inflammatory factors latterly, which may be related to the inactivation of Lyn. These results provide a new perspective and direction for the clinical treatment of sepsis.

Brain-inspired global-local learning incorporated with neuromorphic computing.

There are two principle approaches for learning in artificial intelligence: error-driven global learning and neuroscience-oriented local learning. Integrating them into one network may provide complementary learning capabilities for versatile learning scenarios. At the same time, neuromorphic computing holds great promise, but still needs plenty of useful algorithms and algorithm-hardware co-designs to fully exploit its advantages. Here, we present a neuromorphic global-local synergic learning model by introducing a brain-inspired meta-learning paradigm and a differentiable spiking model incorporating neuronal dynamics and synaptic plasticity. It can meta-learn local plasticity and receive top-down supervision information for multiscale learning. We demonstrate the advantages of this model in multiple different tasks, including few-shot learning, continual learning, and fault-tolerance learning in neuromorphic vision sensors. It achieves significantly higher performance than single-learning methods. We further implement the model in the Tianjic neuromorphic platform by exploiting algorithm-hardware co-designs and prove that the model can fully utilize neuromorphic many-core architecture to develop hybrid computation paradigm.

A framework for the general design and computation of hybrid neural networks.

There is a growing trend to design hybrid neural networks (HNNs) by combining spiking neural networks and artificial neural networks to leverage the strengths of both. Here, we propose a framework for general design and computation of HNNs by introducing hybrid units (HUs) as a linkage interface. The framework not only integrates key features of these computing paradigms but also decouples them to improve flexibility and efficiency. HUs are designable and learnable to promote transmission and modulation of hybrid information flows in HNNs. Through three cases, we demonstrate that the framework can facilitate hybrid model design. The hybrid sensing network implements multi-pathway sensing, achieving high tracking accuracy and energy efficiency. The hybrid modulation network implements hierarchical information abstraction, enabling meta-continual learning of multiple tasks. The hybrid reasoning network performs multimodal reasoning in an interpretable, robust and parallel manner. This study advances cross-paradigm modeling for a broad range of intelligent tasks.

Adversarial symmetric GANs: Bridging adversarial samples and adversarial networks.

Generative adversarial networks have achieved remarkable performance on various tasks but suffer from training instability. Despite many training strategies proposed to improve training stability, this issue remains as a challenge. In this paper, we investigate the training instability from the perspective of adversarial samples and reveal that adversarial training on fake samples is implemented in vanilla GANs, but adversarial training on real samples has long been overlooked. Consequently, the discriminator is extremely vulnerable to adversarial perturbation and the gradient given by the discriminator contains non-informative adversarial noises, which hinders the generator from catching the pattern of real samples. Here, we develop adversarial symmetric GANs (AS-GANs) that incorporate adversarial training of the discriminator on real samples into vanilla GANs, making adversarial training symmetrical. The discriminator is therefore more robust and provides more informative gradient with less adversarial noise, thereby stabilizing training and accelerating convergence. The effectiveness of the AS-GANs is verified on image generation on CIFAR-10, CIFAR-100, CelebA, and LSUN with varied network architectures. Not only the training is more stabilized, but the FID scores of generated samples are consistently improved by a large margin compared to the baseline. Theoretical analysis is also conducted to explain why AS-GAN can improve training. The bridging of adversarial samples and adversarial networks provides a new approach to further develop adversarial networks.

Synergistic Effect of Pressurization Rate and ß-Form Nucleating Agent on the Multi-Phase Crystallization of iPP.

Using a homemade pressure device, we explored the synergistic effect of pressurization rate and ß-form nucleating agent (ß-NA) on the crystallization of an isotactic polypropylene (iPP) melt. The obtained samples were characterized by combining small angle X-ray scattering and synchrotron wide angle X-ray diffraction. It was found that the synergistic application of pressurization and ß-NA enables the preparation of a unique multi-phase crystallization of iPP, including ß-, γ- and/or mesomorphic phases. Pressurization rate plays a crucial role on the formation of different crystal phases. As the pressurization rate increases in a narrow range between 0.6-1.9 MPa/s, a significant competitive formation between ß- and γ-iPP was detected, and their relative crystallinity are likely to be determined by the growth of the crystal. When the pressurization rate increases further, both ß- and γ-iPP contents gradually decrease, and the mesophase begins to emerge once it exceeds 15.0 MPa/s, then mesomorphic, ß- and γ- iPP coexist with each other. Moreover, with different ß-NA contents, the best pressurization rate for ß-iPP growth is the same as 1.9 MPa/s, while more ß-NA just promotes the content of ß-iPP under the rates lower than 1.9 MPa/s. In addition to inducing the formation of ß-iPP, it shows that ß-NA can also significantly promote the formation of γ-iPP in a wide pressurization rate range between 3.8 to 75 MPa/s. These results were elucidated by combining classical nucleation theory and the growth theory of different crystalline phases, and a theoretical model of the pressurization-induced crystallization is established, providing insight into understanding the multi-phase structure development of iPP.

A new opportunity for the emerging tellurium semiconductor: making resistive switching devices.

The development of the resistive switching cross-point array as the next-generation platform for high-density storage, in-memory computing and neuromorphic computing heavily relies on the improvement of the two component devices, volatile selector and nonvolatile memory, which have distinct operating current requirements. The perennial current-volatility dilemma that has been widely faced in various device implementations remains a major bottleneck. Here, we show that the device based on electrochemically active, low-thermal conductivity and low-melting temperature semiconducting tellurium filament can solve this dilemma, being able to function as either selector or memory in respective desired current ranges. Furthermore, we demonstrate one-selector-one-resistor behavior in a tandem of two identical Te-based devices, indicating the potential of Te-based device as a universal array building block. These nonconventional phenomena can be understood from a combination of unique electrical-thermal properties in Te. Preliminary device optimization efforts also indicate large and unique design space for Te-based resistive switching devices.

Self-Doping Memristors with Equivalently Synaptic Ion Dynamics for Neuromorphic Computing.

The accumulation and extrusion of Ca2+ ions in the pre- and post-synaptic terminals play crucial roles in initiating short- and long-term plasticity (STP and LTP) in biological synapses, respectively. Mimicking these synaptic behaviors by electronic devices represents a vital step toward realization of neuromorphic computing. However, the majority of reported synaptic devices usually focus on the emulation of qualitatively synaptic behaviors; devices that can truly emulate the physical behavior of the synaptic Ca2+ ion dynamics in STP and LTP are rarely reported. In this work, Ag/Ag:Ta2O5/Pt self-doping memristors were developed to equivalently emulate the Ca2+ ion dynamics of biological synapses. With conductive filaments from double sources, these memristors produced unique double-switching behavior under voltage sweeps and demonstrated several essential synaptic behaviors under pulse stimuli, including STP, LTP, STP to LTP transition, and spike-rate-dependent plasticity. Experimental results and nanoparticle dynamic simulations both showed that Ag atoms from double sources could mimic Ca2+ dynamics in the pre- and post-synaptic terminals under stimuli. A perceptron network with an STP to LTP transition layer based on the self-doping memristors was also introduced and evaluated; simulations showed that this network could solve noisy figure recognition tasks efficiently. All of these results indicate that the self-doping memristors are promising components for future hardware creation of neuromorphic systems and emulate the characteristics of the brain.

[Deviation of mass peak position in time-of-flight mass spectrum].

Time-of-flight (TOF) mass spectrometer is widely used in many research areas including the study of dynamics of laser ionization/dissociation molecular clusters. A new phenomenon was observed when we used YAG 355 nm laser to ionize water/methanol mixture molecular beam and detect the ions with time-of-flight mass spectrometer: the peak positions of mass ions shift when changing the laser delay time to the molecular beam and keeping other conditions unchanged. The flight time shortens to a minimum value when the laser ionizes the middle part of the molecular beam, while the ions intensity reaches the maximum. The peak shift does not mean a new mass number ion appearing through mass spectra identification. The reason is deduced to be the voltage fluctuation on the extract electrode of TOF when ions pass through the grating electrode and part of them are being absorbed. The numerical simulation of charge-particle movement in electrical field also supports the deduction, and is consistent with the experimental results.

High performance transparent in-plane silicon nanowire Fin-TFTs via a robust nano-droplet-scanning crystallization dynamics.

High mobility, scalable and even transparent thin-film transistors (TFTs) are always being pursued in the field of large area electronics. While excimer laser-beam-scanning can crystallize amorphous Si (a-Si) into high mobility poly-Si, it is limited to small areas. We here demonstrate a robust nano-droplet-scanning strategy that converts an a-Si:H thin film directly into periodic poly-Si nano-channels, with the aid of well-coordinated indium droplets. This enables the robust batch-fabrication of high performance Fin-TFTs with a high hole mobility of >100 cm2 V-1 s-1 and an excellent subthreshold swing of only 163 mV dec-1, via a low temperature <350 °C thin film process. More importantly, precise integration of tiny poly-Si channels, measuring only 60 nm in diameter and 2 µm apart on glass substrates, provides an unprecedented transparent Si-based TFT technology to visible light, which is widely sought for the next generation of high aperture displays and fully transparent electronics. The successful implementation of such a reliable nano-droplet-scanning strategy, rooted in the strength of nanoscale growth dynamics, will enable eventually the batch-manufacturing and upgrade of high performance large area electronics in general, and high definition and scalable flat-panel displays in particular.

Engineering island-chain silicon nanowires via a droplet mediated Plateau-Rayleigh transformation.

The ability to program highly modulated morphology upon silicon nanowires (SiNWs) has been fundamental to explore new phononic and electronic functionalities. We here exploit a nanoscale locomotion of metal droplets to demonstrate a large and readily controllable morphology engineering of crystalline SiNWs, from straight ones into continuous or discrete island-chains, at temperature <350 °C. This has been accomplished via a tin (Sn) droplet mediated in-plane growth where amorphous Si thin film is consumed as precursor to produce crystalline SiNWs. Thanks to a significant interface-stretching effect, a periodic Plateau-Rayleigh instability oscillation can be stimulated in the liquid Sn droplet, and the temporal oscillation of the Sn droplets is translated faithfully, via the deformable liquid/solid deposition interface, into regular spatial modulation upon the SiNWs. Combined with a unique self-alignment and positioning capability, this new strategy could enable a rational design and single-run fabrication of a wide variety of nanowire-based optoelectronic devices.

Operating principles of in-plane silicon nanowires at simple step-edges.

Growing silicon nanowires (SiNWs) into precise locations is a key enabling technology for SiNW-based device applications. This can be achieved via in-plane growth of SiNWs along a simple step-edge, where metal catalyst droplets absorb an amorphous Si matrix to produce c-SiNWs. However, a comprehensive understanding of this phenomenon is still lacking. We here establish an analytical model to address the driving force that dictates the growth dynamics under various droplet-step contact configurations, and to identify the key control parameters for effective guided growth. These new principles were verified against a series of experimental observations and proved to be powerful in designing a stable guided growth configuration. Furthermore, we propose and demonstrate a unique ability to achieve in situ capturing, guiding and release of the in-plane SiNWs along curved step-edges. We suggest that such a new understanding and results provide a fundamental basis and a practical guide for positioning and integrating self-assembled nanowires in a wide variety of material systems.