| 摘要: | Tomography 原先被使用在生命科學上，其表現非結晶型的3D物體結構具有相當的成果，由於近年來半導體元件尺寸日漸縮小，光靠二維圖片無法輕易地解釋複雜的結構，因此在未來利用Tomography作為分析用途將是一大利器。 Tomography 主要分為影像拍攝、影像校正、影像重組等部份，然而在影像校正方面分為 Feature Tracking 與 Cross-Correlation兩種方式，良好的影像校正對於接下來的影像品質有很大的關聯。 在Feature Tracking中，金粒子分布及數目是決定校正品質兩大因素；因此，第一部分的實驗將利用不同濃度Protein A金粒子懸浮液附著於碳膜上，觀察其表面金粒子分佈情形。 實驗結果發現碳膜上有大量的汙染物，因此無法順利拍攝到金粒子的分佈情形，觀察其成分發現Protein A懸浮液裡面的緩衝溶液含有20%的甘油，我們決定更換另一種成分只有金與水的懸浮液,使用此懸浮液可以獲得金粒子分佈均勻且清晰的影像。 在Cross-Correlation實驗方面,良好的高通濾波與低通濾波參數會使Cross-Correlation peak清晰，經由校正過後連續影像間偏移也會有所減少，本次實驗所使用的軟體是FEI公司的Inspect 3D，將高通濾波0.01、0.03、0.05、0.07、0.09等五種數值分別在低通濾波0.1、0.3、0.5、0.7、0.9等五種數值，在TEM 與STEM連續影像中比較25對參數哪一組校正的效果最好；發現高通濾波0.09在低通濾波0.5的時候TEM影像校正在這25組中有最好的效果；在STEM測試中，發現高通濾波0.03在低通濾波0.1在STEM的影像在這25組中有最好的結果。 Tomography is a well-established technique for three dimensional structure determination amorphous specimens in life sciences applications; Semiconductor device size is diminished in recent year, it is difficult to explain complicated structure by two-dimension image. For this reason, it is advantage for tomography used in fault analysis. Tomography consist of image acquisition、image alignment and image reconstruction. Cross-Collection and Feature Tracking are two methods in image alignment. Alignment well or not can influence with three-dimension image quantity seriously. In the Feature Tracking, the gold particle number and distribution are two important factors in alignment quality. Therefore, in the first experiment, using different concentration of gold particle suspension adhere on the carbon film, and watching gold particles distribution situation. Experiment results show large range of pollutants on carbon film, and therefore can not acquire gold distribution image, because of buffer solution in gold particle suspension contains twenty percentages glycerol. We decide to change another suspension only consists gold and water, and we get clear and uniform distribution gold image by using this suspension. In Cross-Correlation experiment, well high pass filter and low pass filter parameters can make Cross-Correlation peak clear and sharp; the shift between image and above image can be reduced by well alignment. The software used in this experiment is Inspect 3D made by FEI company, we set high pass filter up 0.01, 0.03, 0.05, 0.07 and 0.09 in low pass filter 0.1, 0.3, 0.5, 0.7 and 0.9. Compare effect of alignment by changing parameters which is the best in twenty-five sets in TEM and STEM modes. After experiment, we find high pass filter 0.09 in low pass filter 0.5 has the best TEM image alignment in twenty-five sets, and high pass filter 0.03 in low pass filter 0.1 has the best STEM image alignment in twenty-five sets. |
