Study Of In-Process Optical Measurment Of Wafer Surface Quality

Wafer is the basic material of the production of Integrate Circuit in the semiconductor process, where surface quality control plays an important role during back grinding. A wafer with poor surface quality has hidden risks of stress concentration and little cracks, which would cause a huge damage of bursting crack when dicing the wafer. Surface roughness is an important parameter for evaluating surface quality. It is a comprehensive evaluation of micro-geometric characteristics of all the micro-spacing and peak-valley roughness on the surface, reflecting stress distribution, hence used for surface quality evaluation. The wafer surface roughness parameter Ra (arithmetic average height) is required between 0.01μm and 0.8μm.An in-process wafer surface quality measurement method based on the light scattering principle was established, which is founded on a simple but effective machine vision system. By analyzing the images of the spatial distribution of the scattered light intensity, some parameters were selected to estimate Ra to achieve the evaluation of wafer surface quality. Elementary experiment results were shown by illustrating the relationships between characteristic parameters and surface roughness, and the feasibility of this method was demonstrated.Furthermore, the Support Vector Machine method based on the statistical learning theory was introduced to solve the limitation that using single parameter to evaluate surface roughness. A general approach was suggested for further study.