A regression technique to analyze the second-order nonlinear optical response of thin films

We present a new technique, based on regression analysis, to determine the second-order nonlinear optical susceptibility tensor of thin films. The technique does not require the absolute levels or phases of measured signals to be mutually calibrated. In addition it yields indicators that address the quality of theoretical models describing the sample. We use the technique to determine the susceptibility tensor of samples of a nonracemic chiral material which have very low symmetry (both chiral and anisotropic) and have many independent tensor components. The results show the importance of using detailed theoretical models that account for the linear optical properties of the sample.