先进ATHENATM选用策略与对准功能

基于先进科技对准规格的减缩,对准仪量测数据的应用也变得更为重要。不同对准策略的选择所引起的结果,对全部对准预算有很重要的影响。介绍两个主要科技研发项目:先进对准配套和对准标的设计。     

A novel overlay process for imprint lithography using load release and alignment error pre-compensation method

As a low-cost and high-efficiency alternative to conventional photolithography, imprint lithography has been paid serious attention for it can realize the pattern transfer easily and repeatedly. Even so, there is still a long way to go before this technique can be used to manufacture ICs since the layer-to-layer alignment is a main obstacle. In this paper, the pattern distortion and position shift induced by loading force and friction force are investigated and an optimized process is proposed to realize high precision overlay, in which an optimal loading force and a load release and an alignment error pre-compensation (LRAEPC) method are adopted. The optimal loading force is acquired by comparing the variety trends of the residual layer and position shift with the loading force. Using the optimal loading force, transferred patterns with proper resist thickness and the minimal distortion are achieved. In LRAEPC method, load release is used to correct the pattern distortion and to alleviate the position shift. To avoid subsequent position adjustment in the fluid after load release, the pre-compensation alignment is performed before the mold is in contact with the wafer. This process does not locate correct alignment position and there is an alignment compensation value determined by the statistical data of alignment position shift after load release. Using this pre-compensation alignment method, the correct alignment can be obtained after load release.     

光面位置测量仪及研制

对二维视觉传感器和现有标定方法进行了简单介绍,建立了标定模型。基于模型介绍了标定调整方法,研制了光面位置测量仪,即利用了光电二极管对光强感应的原理。详细介绍了光面位置测量仪的结构,包括机械部分和电路部分,并进行了误差分析。此测量仪操作简单,克服了以往方法的复杂繁琐。经实验验证,测量仪的重复性误差达到1.03°,达到应用系统的要求。     

基于改进的TextRank算法的计算机辅助定密研究

针对传统定密方式定密不严谨、定密尺度难以把握、经验难以积累等问题,提出基于改进的TextRank算法的计算机辅助定密方法,该方法通过定密规则的词性特点,将句向量分解为名词向量和非名词向量,构造基于词性的句向量,利用改进的TextRank算法对文档语句排序,获取在定密细则影响下的关键语句权重,计算文档密级分数,判断文档密级。实验结果表明,该方法比目前传统定密方式准确率有所提高。     

A deep feature alignment adaptation network for rolling bearing intelligent fault diagnosis

Fault diagnostic methods based on deep learning achieve impressive progress recently, but most studies assume that signals from the source domain and target domain share a similar probability distribution. However, the domain shift phenomenon is often unavoidable in practical engineering because of changeable conditions, which hinders the performance of some intelligent methods in fault diagnosis. To tackle the above issue, an unsupervised domain adaptation approach called Deep Feature Alignment Adaptation Network (DFAAN) is proposed in this paper to raise the domain adaptability of fault diagnosis. Firstly, the latent distributions of the two domains are aligned indirectly guided by a Gaussian prior to create a common latent space, which can promote the feature alignment across different domains. Secondly, to better narrow the discrepancy of the feature distribution with the Gaussian prior, a novel discriminative reconstruction distance based on the mechanism of the autoencoder is introduced. Thirdly, an entropy minimum technique is incorporated in the objective function to further increase the transferability of the adaptation method. Diagnostic experiments are conducted on two bearing datasets to illustrate the effectiveness of the proposed approach. The results reveal the superiority of the proposed approach over other typical methods and validate the versatility in multiple diagnostic tasks.     

Vertically aligned carbon nanotubes grown on geometrically different types of surface

Aligned carbon nanotubes (CNTs) which were always perpendicular to the surface of substrate were synthesized on geometrically different surface through microwave plasma chemical vapour deposition (MWPCVD) at the low temperature of 550 °C. Growth was performed in a flowing mixture of CH₄ and H₂. Scanning electron microscopy (SEM) shows that the vertically aligned growth occurred under the effect of plasma. When the substrate was not contacted with plasma, only randomly entangled CNTs had grown on the substrate. The research results demonstrate that the electrical self-bias imposed on the surface is the primary mechanism responsible for the alignment of CNTs.     

Reconstruction of grains and subgrains from electron backscatter diffraction maps

Electron backscatter diffraction maps are capable of yielding a substantial amount of quantitative information about grains, subgrains and boundaries, and the amount and quality of the data may be substantially increased if the pixels of the map are re‐analysed so as to ‘reconstruct’ complete grains or subgrains. The paper discusses the various methods of grain reconstruction and the use of such methods to obtain microstructural information correlating the parameters of dimension, position, orientation and misorientation, which cannot usually be obtained by other means. Grain reconstruction also reveals the nature, location and contacts of all the triple junctions in the microstructure, and the paper discusses two important examples of how these data may be further analysed using automated routines. Boundary connectivity and the length and direction of likely paths along which grain boundary events such as creep fracture or stress corrosion may occur can readily be determined. The overall alignment of boundaries in deformed metals, with respect to the crystallography and the deformation geometry, may be determined as a function of the length and misorientation of the boundary segments.     

Accurate overlaying for mobile augmented reality

Mobile augmented reality requires accurate alignment of virtual information with objects visible in the real world. We describe a system for mobile communications to be developed to meet these strict alignment criteria using a combination of computer vision, inertial tracking and low-latency rendering techniques. A prototype low-power and low-latency renderer using an off-the-shelf 3D card is discussed.