公开(公告)号：US5355212A

公开(公告)日：1994-10-11

申请号：US93999

申请日：1993-07-19

申请人： Keith B. Wells ,  Hung Nguyen ,  Ralph T. Johnson ,  Brian C. Leslie

发明人： Keith B. Wells ,  Hung Nguyen ,  Ralph T. Johnson ,  Brian C. Leslie

IPC分类号： G01N21/94 ,  G01N21/956 ,  G01N21/88

CPC分类号： G01N21/94 ,  G01N21/95607

摘要： A method of locating particle and defect features on a periodically patterned surface uses multiple threshold intensity levels to identify features in the data stream produced by scanning the surface with a light beam and detecting the light scattered from the surface. High thresholds are assigned to regions of the surface with high background scatter, while low thresholds are assigned to regions of the surface with low background scatter. The scattered light is detected with a wide dynamic range detector producing high resolution 12-bit pixel data capable of resolving the smallest particles and defects of interest in low scatter regions, while avoiding saturation in high scatter regions. Periodic pattern features are removed from the data by mapping features from a plurality of periodically repeating die on the surface to a single die map and looking for overlapping features. Unique, nonoverlapping features are determined to correspond to particles and defects. In one embodiment, only a portion of all the die on the surface are mapped at one time to reduce the possibility of accidental overlap of particle and defect features.

摘要翻译： 在周期性图案化表面上定位颗粒和缺陷特征的方法使用多个阈值强度水平来识别通过用光束扫描表面而产生的数据流中的特征并且检测从表面散射的光。 高阈值被分配给具有高背景散射的表面的区域，而低阈值被分配给具有低背景散射的表面的区域。 用宽动态范围检测器检测散射光，产生高分辨率12位像素数据，能够在低散射区域中分辨最小的粒子和感兴趣的缺陷，同时避免高散射区域的饱和。 通过将表面上的多个周期性重复的模具的特征映射到单个模具图并查找重叠的特征，从数据中去除周期性模式特征。 确定独特的非重叠特征以对应于颗粒和缺陷。 在一个实施例中，仅一次表面上的所有模具的一部分被映射，以减少颗粒和缺陷特征的意外重叠的可能性。