为研制高灵敏海水盐度传感器，本文基于CO₂激光技术成功制备出一种工作在色散转折点（DTP）附近的长周期光纤光栅（LPFG）。首先，利用CO₂激光器在80 μm细单模光纤上制备出工作在DTP附近的LPFG，证明了采用CO₂激光微加工技术制备较短周期LPFG的可能性。其次，通过调控CO₂激光器的制备周期，使高阶包层模式LP_1,9工作在DTP附近，从而显著提高了LPFG的折射率灵敏度。在双峰谐振增敏效应的作用下，当海水盐度从5.001‰变化到39.996‰时，光栅周期为115.4 μm的双峰谐振LPFG平均灵敏度高达0.279 nm/‰。研究结果表明，本文制备的LPFG海水盐度传感器具有谐振损耗大和灵敏度高的优点，其在海水盐度监测领域具有较好的应用前景。

Evidence for metallization in dense oxygen has been reported for over 30 years [Desgreniers et al., J. Phys. Chem. 94, 1117 (1990)] at a now routinely accessible 95 GPa [Shimizu et al., Nature 393, 767 (1998)]. However, despite the longevity of this result and the technological advances since, the nature of the metallic phase remains poorly constrained [Akahama et al., Phys. Rev. Lett. 74, 4690 (1995); Goncharov et al., Phys. Rev. B 68, 224108 (2003); Ma, Phys. Rev. B 76, 064101 (2007); and Weck et al., Phys. Rev. Lett. 102, 255503 (2009)]. In this work, through Raman spectroscopy, we report the distinct vibrational characteristics of metallic ζ-O₂ from 85 to 225 GPa. In comparison with numerical simulations, we find reasonable agreement with the C2/m candidate structure up to about 150 GPa. At higher pressures, the C2/m structure is found to be unstable and incompatible with experimental observations. Alternative candidate structures, C2/c and Ci, with only two molecules in the primitive unit cell, are found to be stable and more compatible with measurements above 175 GPa, indicative of the dissociation of (O₂)₄ units. Further, we report and discuss a strong hysteresis and metastability with the precursory phase ϵ-O₂. These findings will reinvigorate experimental and theoretical work into the dense oxygen system, which will have importance for oxygen-bearing chemistry, prevalent in the deep Earth, as well as fundamental physics.

科研试剂耗材采购、使用和报销是科学研究中不可或缺的一环，但是由于科研试剂耗材的特殊性，往往都是采用以课题组为单位自主选择、随买随用、实时报销的管理模式。课题组作为科研中的基本单元，可以通过做好试剂耗材管理和经费报销为科研工作和经费管理打下良好基础。本课题组在实际工作中，依靠在线表单系统——简道云搭建了一整套耗材试剂管理系统，覆盖了从采购、入库、领用、出库到报账的全流程，实现了耗材试剂管理的无纸化高效管理，并实现了经费管理中的使用追踪。

旋转目标检测是遥感图像处理领域中的重要任务，其存在的目标尺度变化大和目标方向任意等问题给自动目标检测带来了挑战。针对上述问题，提出了一种改进的RoI Transformer旋转目标检测框架：首先，利用RoI Transformer检测框架获取旋转的感兴趣区域特征(rotated region of interest, RRoI)用于鲁棒的几何特征提取；其次，在检测器中引入高分辨率网络(high-resolution network, HRNet)提取多分辨率特征图，在保持高分辨率特征同时适应目标的多尺度变化；最后，引入KLD(Kullback-Leibler divergence)损失，解决旋转目标表示的角度周期性的问题，提高检测方法对任意方向目标的适应性，并通过旋转目标边界框参数的联合优化提升目标定位精度。本文提出的旋转目标检测方法，即HRD-ROI Transformer (HRNet + KLD ROI Transformer)，在DOTAv1.0和DIOR-R两个公开数据集上与典型的旋转目标检测方法进行了比较。结果显示：相比于传统的RoI Transformer检测框架，本文方法在DOTAv1.0和DIOR-R数据集上检测结果的mAP(mean-average-precision)分别提高了3.7%和4%。