Taguchi Analysis of Relation Between Tensile Strength and Interfacial Phases Quantified via Image Processing

Quantitative analysis was performed via image processing to identify the relationship between the tensile strength and the thickness of the Cr x B y phase layer at the interface of brazed 304 stainless steel with Ni-based filler metal (MBF20). The experimental design was based on the Taguchi method to determine the relative contributions of the processing conditions, including the brazing temperature, heating rate, holding time, and filler metal thickness. The Cr x B y phase at the brazement interface was extracted by an image-processing method from backscattered electron imaging; the numerically defined thicknesses of the Cr x B y phase layers developed under varied processing conditions were calculated. The tensile strengths at temperatures of 25 °C and 650 °C were measured for specimens brazed under identical experimental conditions based on the Taguchi method and the post-tensile testing fracture surfaces were analyzed. Regarding the relationship between the thickness of the Cr x B y phase layer, as determined through the image processing of the microstructure, and the tensile strength at 25 °C, thicker Cr x B y layers deteriorated the tensile strength of the brazement interfaces. Although a slight discrepancy occurred in the brazement tensile strengths between the testing temperatures of 25 °C and 650 °C, the elevated temperature during tensile testing affected the brazed interface microstructure; with this consideration, the overall results for both tensile strength tests corresponded to the quantitatively analyzed Cr x B y phase layer thicknesses. From the relationship between Cr x B y layer thickness and tensile strength, the heating rate is the most effective processing condition to achieve high bonding strength, because changes in heating rate compared to those of other processing conditions have the greatest effect in changing the Cr x B y phase layer thicknesses and tensile strength. Therefore, the results confirmed that the image-processing method enabled accurate quantitative analysis of the microstructure, permitting prediction of the mechanical strength of the joint.