Comparison and Combination of Imaging Techniques for Three Dimensional Analysis of Electrical Trees

Research output: Contribution to journalArticle

Abstract

Imaging of electrical trees has been an important tool for studying the phenomenon. The authors have previously shown that electrical trees can be three-dimensionally (3D) imaged and virtual replicas generated using X-ray Computed Tomography (XCT) or Serial Block-Face Scanning Electron Microscopy (SBFSEM). Here these techniques are evaluated and compared for 3D analysis of electrical trees along with conventional optical methods. A number of types of laboratory created trees showing range of morphologies were grown and examined to delineate the capabilities of each technique. Cross-sectional images and virtual replicas of the electrical trees from XCT and SBFSEM techniques were compared both qualitatively and quantitatively. SBFSEM provides greater detail than XCT, evidenced by imaging smaller sub-branches and when comparing parameters such as the number of tree channels, tree length or tree volume captured. On average, SBFSEM captures almost double the number of tree channels per slice than XCT, and virtual replicas in most of the cases have larger volumes. However, SBFSEM is a destructive technique, which makes the imaging process less reliable than XCT and not suitable for multi-stage of tree growth experiments. For full analysis, a combination of imaging techniques is proposed. Optical methods are used first to monitor tree growth. Then, micro-XCT which provides pixel size down to ~0.4 µm with a field of view of around 1 mm × 1 mm, can be used to reveal the overall 3D structure of a normal/mature electrical tree. Nano-XCT can be used to explore in more detail regions of interest, with a pixel size of ~65 nm, but a limited field of view of 65 µm. Finally, sections of the tree can be analyzed in even greater detail using SBFSEM, which can provide resolutions below 50 nm. Using this approach, a deeper and more complete analysis of the structure of electrical trees can be achieved.

Bibliographical metadata

Original languageEnglish
Pages (from-to)709-719
Number of pages10
JournalI E E E Transactions on Dielectrics and Electrical Insulation
Volume22
Issue number2
DOIs
Publication statusPublished - 2015

Related information

Researchers

View all