Obtaining porous Si characteristic from SEM images via Non-destructible method; image segmentation

Abstract

This work aims to introduce an alternative method of obtaining porous density characteristics of porous silicon material by making use of images obtained from the Scanning Electron Microscope. The available and most commonly used method of obtaining the porous density characteristics of semiconductor materials is the gravimetric or quasi-gravimetric method. Using the gravimetric approach requires the sample material to go through various measurements during the multiple stages of processing and it would ultimately result in the destruction of the sample material. The gravimetric approach is flawed as the results it produces are questionable as it is less accurate. Also, it is refutable due to its destructive nature which is caused by the use of alkaline solution which dissolves the sample material at the final stage of the gravimetric process. Therefore, this research introduces an alternative image processing technique which would require only images of the sample material as an input which is obtained via the Scanning Electron Microscope. The image obtained is processed by segmenting the SEM images into two significant black and white regions which allow for the number of pores present on the image to be numerated. From the data obtained from the image, the porosity and porous density of the sample material can be calculated. While being a much simpler process than the commonly used gravimetric method, it is also non-destructive to the sample and is believed to produce a more precise and accurate result.