LAUSR

Abstract White light scanning interferometry is a very important technique for nanometrology. In this paper, a robust white light interference signal processing algorithm using random phase noise correction method is proposed. By analyzing the phase response of the correlgram mathematically, the phase noise is modelized by the combination of random noise and systematic deviation. The random noise can be corrected by assembling the least square result of phase information. As a result, the relationship between phase distribution and surface height is established in a more accurate way. The simulations show that the novel algorithm is capable of achieving robust measurements with standard deviation one tenth comparing to the scanning step length errors, and has an extremely high noise tolerance 14.25 nm with scanning step length 20 nm. In the experiments, a step height standard (VLSI, 182.7 ± 2.0 nm), a protected aluminum mirror and a drive shaft cover of aircraft engine were tested, where the repeatability error for the step height standard is 0.5 nm, the roughness repeatability of the mirror is only 0.03 nm, and the roughness repeatability of the cover is 5.14 nm, which proves the accuracy and robust of the algorithm.