Darwin3: a large-scale neuromorphic chip with a novel ISA and on-chip learning.

Spiking neural networks (SNNs) are gaining increasing attention for their biological plausibility and potential for improved computational efficiency. To match the high spatial-temporal dynamics in SNNs, neuromorphic chips are highly desired to execute SNNs in hardware-based neuron and synapse circuits directly. This paper presents a large-scale neuromorphic chip named Darwin3 with a novel instruction set architecture, which comprises 10 primary instructions and a few extended instructions. It supports flexible neuron model programming and local learning rule designs. The Darwin3 chip architecture is designed in a mesh of computing nodes with an innovative routing algorithm. We used a compression mechanism to represent synaptic connections, significantly reducing memory usage. The Darwin3 chip supports up to 2.35 million neurons, making it the largest of its kind on the neuron scale. The experimental results showed that the code density was improved by up to 28.3× in Darwin3, and that the neuron core fan-in and fan-out were improved by up to 4096× and 3072× by connection compression compared to the physical memory depth. Our Darwin3 chip also provided memory saving between 6.8× and 200.8× when mapping convolutional spiking neural networks onto the chip, demonstrating state-of-the-art performance in accuracy and latency compared to other neuromorphic chips.

Simultaneous Quantification of Biomarkers for Bis-(2-chloroethyl) Sulfide and 1,2-Bis(2-chloroethylthio) Ethane Exposure in Human Urine at Trace Exposure Levels by Gas Chromatography Tandem Mass Spectrometry via Simultaneous Incubation and Extraction.

Sulfur mustard [HD; bis-(2-chloroethyl) sulfide] and other analogues are a kind of highly toxic vesicant and have been prohibited by the Organization for the Prohibition of Chemical Weapons (OPCW) since 1997. Exposures to HD could generate several adducts in the plasma and hydrolysis products in the urine, which are widely applied as biomarkers to identify HD exposure in forensic analysis. Several methods have been developed for the detection of related biomarkers. However, most methods are based on complex derivatization, and not enough attention is paid to HD analogues. A modified and convenient analytical method reported herein includes simultaneous incubation and organic solvent extraction. The biomarkers such as thiodiglycol and 1,2-bis (2-hydroxyethylthio) are transferred to HD and 1,2-bis(2-chloroethylthio) ethane via hydrochloric acid at the appropriate temperature. The analytes are analyzed by gas chromatography tandem mass spectrometry (GC-MS/MS) with 2-chloroethyl ethyl sulfide (2-CEES) applied as the internal standard. The interday and intraday study according to FDA rules has been achieved to evaluate the accuracy and precision of the method. The two targets are detected with a good linearity (R2 > 0.99) in the concentration ranges from 5 to 1000 ng/mL and 10 to 1000 ng/mL, with small relative standard deviations (RSD ≤6.62% and RSD ≤6.93%) and favorable recoveries between 90.3 and 107.3% and between 89.4 and 108.7%, respectively. The established method can be used for retrospective detection of sulfur mustards in biological samples and successfully applied in the biomedical proficiency testing organized by the OPCW.

Fluoride derivatization-enabled sensitive and simultaneous detection of biomarkers for nitrogen mustard in human plasma and urine via gas chromatography tandem mass spectrometry.

Methyl-diethanolamine (CAS: 105-59-9), ethyl-diethanolamine (CAS: 139-87-7), and triethanolamine (CAS: 102-71-6) were identified as the degradation products and bio-markers of nitrogen mustard exposure. Sensitive and convenient detection methods for amino alcohol are of great importance to identify nitrogen mustard exposure in forensic analysis. Herein, analytical methods including gas chromatography-tandem mass spectrometry combined with heptafluorobutyryl derivatization and solid phase extraction were established for retrospective detection of the biomarkers in human plasma and urine samples. The efficiency of the method was improved by optimizing the conditions for sample preparation and the GC-MS/MS method. The optimization included the derivatization temperature, reaction time, reagent dosage and solid phase extraction cartridges, eluent and pH of the loading sample. The results indicated that the SCX cartridge resulted in better enrichment and purification effects, and the best recovery could be obtained with pH = 3-4 for the loading samples and an eluent of 2 mL 10% NH4OH/MeOH. The GC-MS/MS parameters were also optimized for better specificity and sensitivity. The established method was fully validated for each analyte both in plasma and urine matrixes. The linear range of analytes in plasma was 1.0-1000 ng mL-1 with a correlation parameter (R2) of ≥0.994, intra-day/inter-day accuracy of 93.7-117%, and relative standard deviation (RSD) of ≤6.5%. Meanwhile the results in urine were 1.0-1000 ng mL-1 with R2 of ≥0.996, intra-day/inter-day accuracy of 94.3-122%, and RSD of ≤6.6%. The detection limit of the analytes was 1.0 ng mL-1. The method was applied for the detection and identification of trace amino alcohols present in urine samples dispatched by the Organization for the Prohibition of Chemical Weapons (OPCW) and the results were confirmed to be correct.

A charge control method for space-mission inertial sensor using differential UV LED emission.

Various space missions and applications require the charge on isolated test masses to be strictly controlled because any unwanted disturbances will introduce acceleration through the Coulomb interaction between the test masses and their surrounding conducting surfaces. In many space missions, charge control has been realized using ultraviolet (UV) photoemission to generate photoelectrons from metal surfaces. The efficiency of photoelectron emission strongly depends on multiple physical parameters of gold-coated surfaces, such as the work function, reflectivity, and quantum yield. Therefore, to achieve satisfactory charge control performance, these parameters need to be measured accurately. This paper describes a charge control method that achieves self-adaptive charge neutralization while removing the need to measure the above-mentioned physical parameters. First, to explain the principle, a differential illumination model is constructed based on the typical structure of an inertial sensor. A charge management system based on a torsion pendulum system is then introduced along with an UV light emitting diode coupling system. Finally, experimental results are obtained in a vacuum chamber system with a pressure of 10-7 mbar, showing that precise calibration allows the test mass potential to be automatically controlled below 10 mV without considering the physical parameters or measuring the potential of the test mass before or after the control process.

Effective decrease of photoelectric emission threshold from gold plated surfaces.

Many applications require charge neutralization of isolated test bodies, and this has been successfully done using photoelectric emission from surfaces which are electrically benign (gold) or superconducting (niobium). Gold surfaces nominally have a high work function (∼5.1 eV) which should require deep UV photons for photoemission. In practice, it has been found that it can be achieved with somewhat lower energy photons with indicative work functions of (4.1-4.3 eV). A detailed working understanding of the process is lacking, and this work reports on a study of the photoelectric emission properties of 4.6 × 4.6 cm2 gold plated surfaces, representative of those used in typical satellite applications with a film thickness of 800 nm, and measured surface roughnesses between 7 and 340 nm. Various UV sources with photon energies from 4.8 to 6.2 eV and power outputs from 1 nW to 1000 nW illuminated ∼0.3 cm2 of the central surface region at angles of incidence from 0° to 60°. Final extrinsic quantum yields in the range 10 ppm-44 ppm were reliably obtained during 8 campaigns, covering a period of ∼3 years but with intermediate long-term variations lasting several weeks and, in some cases, bake-out procedures at up to 200 °C. Experimental results were obtained in a vacuum system with a baseline pressure of ∼10-7 mbar at room temperature. A working model, designed to allow accurate simulation of any experimental configuration, is proposed.