Research Of Rough Positioning Technology For Industrial Scanning Probe Microscopy

Industrial scanning probe microscopy(SPM) is mainly applied in the measurement of critical dimension of large size(about 4-8 inch) wafer samples in semiconductor manufacturing industry. Industrial SPM is capable of obtaining the real three-dimensional morphology of the micro-structure and requires more efficient and automatic than the ordinary SPM. This thesis is concentrated on the demand of coarse location of interest region on lager size samples and proposes a fast positioning method based on the digital image processing technology. In this thesis, a advanced auto-focusing technology is presented for focusing the samples with the optical microscope and a image matching technology is applied in searching the positioning marks on the surface of samples, so the coordinate values of interest region can be achieved through coordinate transformation. Robert algorithm is selected to be the most appropriate focusing evaluation function based on features of wafer samples' surface and the climbing searching algorithm combined with variable step searching strategy is adopted to search the focal plane. On the other hand, SURF algorithm is used to extract features of marks and the mismatching points can be eliminated as many as possible by the bidirectional matching method. The main work in this thesis is shown as the following:1. A brief review about the development history of SPM is presented, and the current research situation and drawbacks of SPM technology is analyzed in this thesis. The significance of this project and the necessity of the research of coarse positioning technology is indicated.2. The design of the overall structure of industrial SPM is introduced in detail, including the installation and calibration of mechanical system and the completion of establishing the system coordinate and coordinate transformation.3. The fast auto-focusing technology applied in industrial SPM is discussed from the aspect of selecting the focusing evaluation algorithm and determining the focal plane searching strategy. Robert algorithm is selected to be the most appropriate focusing evaluation function and the climbing searching algorithm combined with variable step searching strategy is adopted to search the focal plane.4. The features of marks is extracted by SURF algorithm and the mismatching points can be eliminated effectively with the bidirectional matching method. According to the distance and angle between mark on samples and the target location calculated by the coordinate transformation equation, the positioning function can be achieved.5. The PC control software concentrated on auto focusing and positioning function is designed based on the industrial SPM experimental platform. And the experimental results show that the positioning accuracy can reach to ±8 ?m and the time spent in positioning can be limited in 30 seconds with this method.