Fracture strength of Czochralski silicon wafers: Statistically rigorous measurement and dopant effects

Abstract

The fracture of silicon wafers is a critical concern in the manufacturing of microelectronic devices, solar cells, and micro-electro-mechanical systems, making the study of silicon wafer fracture strength both technologically and scientifically significant. In this work, we have employed a custom-designed, computer-controlled fracture strength testing system based on the “ball-on-ring” biaxial stress loading method to investigate the fracture strengths of six types of 300 mm-diameter, 〈100〉-oriented, double-sided polished Czochralski (Cz) silicon wafers, which are lightly boron-doped, lightly phosphorus-doped, heavily boron-doped, heavily phosphorus-doped, heavily antimony-doped, and heavily arsenic-doped, respectively. It is found that the statistical sample size is crucial for obtaining reliable Weibull characteristic fracture strength. Concretely, a minimum sample size of 50 is required for the testing method adopted in this work. Mann–Whitney U tests conducted on the 50-sample sets confirm no statistically significant differences in the fracture strengths of the aforementioned six types of silicon wafers, indicating that the dopant type and concentration have negligible effects on the silicon wafer fracture strength. Moreover, we have analyzed the fracture modes under both low and high fracture strengths. Based on the mechanism of covalent bond rupture thus forming the fracture surfaces, we have gained understanding on the negligible dopant effects on the silicon wafer fracture strength. This study offers new insights into the rigorous measurement of fracture strength and the role of dopants in determining fracture properties of silicon wafers.