Residual stress measurement on TiN thin films by combing nanoindentation and average X-ray strain (AXS) method

Abstract

In our previous study, an average X-ray strain (AXS) method was proposed to improve the measurement accuracy of X-ray residual stress (XRS) and X-ray elastic constants (XECs) by increasing the sampling volume in X-ray diffraction (XRD). Using AXS and elastic constant determined by nanoindentation (ENI) to calculate XRS, the stress deviation from that measured by optical curvature technique (ORS) could be reduced to 3% for TiN coatings with thickness larger than 1.55μm. The purposes of this study were to further explore the lower limit of film thickness where the AXS method could be employed to accurately determine the residual stress using lab-XRD facilities, and to assess the feasibility to calculate XRS of thin films by combining AXS and ENI from thicker coatings (>1μm). TiN thin films on Si substrate with thickness ranging from 85nm to 2.5μm were selected as the model system. AXS was determined using cos2αsin2ψ XRD technique at several rotational (ϕ) angles ranging from 0 to 180°, and the XRS was calculated by multiplying ENI with the AXS. The deviation of XRS from ORS was served as an indicator to validate the effectiveness of the AXS method. The results showed that AXS could be accurately measured down to a thickness of 350nm by using lab X-ray source; however, the resolution of AXS was not sufficient to differentiate the strain for a specimen with a thickness less than 160nm, which was mainly due to insufficient sampling volume. Moreover, the ENI measured following ISO 14577 standard procedures could be used for calculating XRS, if the film thickness was sufficiently large, 680nm in this study, such that substrate effect could be avoided. Two approaches were proposed for measuring the Young's modulus and Poisson's ratio of TiN thin films. The results also confirmed that the thickness limit of AXS method was about 350nm by comparing the resultant elastic constants with literature and experimental data.