The two dimensional microstructure characterization of cemented carbides with an automatic image analysis process

Abstract

Abstract The traditional two dimensional microstructure characterization of cemented carbide, based on stereology of linear intercept method, requires tedious and subjective manual measurements. In this study, an automatic image analysis procedure with two key techniques, i.e. maximum class square error method and watershed transformation method, has been successfully developed. The image analysis for WC-16Co cemented carbides with this procedure easily acquires consistent microstructure parameters. The analysis for area weighted WC grain size, as well as the subsequent mean free path of Co binder show quite different results compared with the conventional number weighted data. It is found that for both number weighted and area weighted data, the contiguity of WC/WC grains is insensitive to the variation of either mean WC grain size or mean free path of Co binder. The mean WC grain size is linearly related to the mean free path of Co binder. The hardness of cemented carbide, having a linear relationship with the inverse square root of mean WC grain size, conforms to Chermont and Osterstock's model. Although it is too early to conclude whether number weighted or area weighted WC grain size (and subsequent mean free path of Co binder) is better, this study shows that area weighted WC grain size and the corresponding mean free path of Co binder are more suitable for Chermont and Osterstock's hardness modeling compared with number weighted WC grain size. The area weighted WC grain size and subsequent mean free path of Co binder, which have rarely been considered for microstructure characterization of cemented carbide previously, could be the key parameters for a better understanding of the microstructure evolution, as well as a better mechanical behavior modeling for cemented carbide.