Lyotropic Boron Nitride Nanotube Liquid Crystals: Preparation, Characterization, and Wet-Spinning for Fabrication of Composite Fiber.

Microfiber fabrication via wet-spinning of lyotropic liquid crystals (LCs) with anisotropic nanomaterials has gained increased attention due to the microfibers' excellent physical/chemical properties originating from the unidirectional alignment of anisotropic nanomaterials along the fiber axis with high packing density. For wet-spinning of the microfibers, however, preparing lyotropic LCs by achieving high colloidal stability of anisotropic nanomaterials, even at high concentrations, has been a critically unmet prerequisite, especially for recently emerging nanomaterials. Here, we propose a cationically charged polymeric stabilizer that can efficiently be adsorbed on the surface of boron nitride nanotubes (BNNTs), which provide steric hindrance in combination with Coulombic repulsion leading to high colloidal stability of BNNTs up to 22 wt %. The BNNT LCs prepared from the dispersions with various stabilizers were systematically compared using optical and rheological analysis to optimize the phase behavior and rheological properties for wet-spinning of the BNNT LCs. Systematic optical and mechanical characterizations of the BNNT microfibers with aligned BNNTs along the fiber axis revealed that properties of the microfibers, such as their tensile strength, packing density, and degree of BNNT alignment, were highly dependent on the quality of BNNT LCs directly related to the types of stabilizers.

Direct 3D-printed CdSe quantum dots via scanning micropipette.

The micropipette, pencil-shaped with an aperture diameter of a few micrometers, is a potentially promising tool for the three-dimensional (3D) printing of individual microstructures based on its capability to deliver low volumes of nanomaterial solution on a desired spot resulting in micro/nanoscale patterning. Here, we demonstrate a direct 3D printing technique in which a micropipette with a cadmium selenide (CdSe) quantum dot (QD) solution is guided by an atomic force microscope with no electric field and no piezo-pumping schemes. We define the printed CdSe QD wires, which are a composite material with a QD-liquid coexistence phase, by using photoluminescence and Raman spectroscopy to analyze their intrinsic properties and additionally demonstrate a means of directional falling.

Optimal Architecture of Floating-Point Arithmetic for Neural Network Training Processors.

The convergence of artificial intelligence (AI) is one of the critical technologies in the recent fourth industrial revolution. The AIoT (Artificial Intelligence Internet of Things) is expected to be a solution that aids rapid and secure data processing. While the success of AIoT demanded low-power neural network processors, most of the recent research has been focused on accelerator designs only for inference. The growing interest in self-supervised and semi-supervised learning now calls for processors offloading the training process in addition to the inference process. Incorporating training with high accuracy goals requires the use of floating-point operators. The higher precision floating-point arithmetic architectures in neural networks tend to consume a large area and energy. Consequently, an energy-efficient/compact accelerator is required. The proposed architecture incorporates training in 32 bits, 24 bits, 16 bits, and mixed precisions to find the optimal floating-point format for low power and smaller-sized edge device. The proposed accelerator engines have been verified on FPGA for both inference and training of the MNIST image dataset. The combination of 24-bit custom FP format with 16-bit Brain FP has achieved an accuracy of more than 93%. ASIC implementation of this optimized mixed-precision accelerator using TSMC 65nm reveals an active area of 1.036 × 1.036 mm2 and energy consumption of 4.445 µJ per training of one image. Compared with 32-bit architecture, the size and the energy are reduced by 4.7 and 3.91 times, respectively. Therefore, the CNN structure using floating-point numbers with an optimized data path will significantly contribute to developing the AIoT field that requires a small area, low energy, and high accuracy.

Spatially resolved Raman spectroscopy of defects, strains, and strain fluctuations in domain structures of monolayer graphene.

We report spatially resolved Raman scattering results of polycrystalline monolayer graphene films to study the effects of defects, strains, and strain fluctuations on the electrical performance of graphene. Two-dimensional Raman images of the integrated intensities of the G and D peaks (I G and I D) were used to identify the graphene domain boundaries. The domain boundaries were also identified using Raman images of I D/I G and I 2D/I G ratios and 2D spectral widths. Interestingly, the I D maps showed that the defects within individual domains significantly increased for the graphene with large domain size. The correlation analysis between the G and 2D peak energies showed that biaxial tensile strain was more developed in the graphene with large domain size than in the graphene with small domain size. Furthermore, spatial variations in the spectral widths of the 2D peaks over the graphene layer showed that strain fluctuations were more pronounced in the graphene with large domain size. It was observed that the mobility (sheet resistance) was decreased (increased) for the graphene with large domain size. The degradation of the electrical transport properties of the graphene with large domain size is mainly due to the defects, tensile strains, and local strain fluctuations within the individual domains.