When a material interacts with intense light its refractive index depends on the intensity of radiation. This dependency is described by a substance parameter called nonlinear refractive index ($n_2$). This parameter is responsible for many phenomena of nonlinear optics and in modeling such phenomena it is important to know the most precise value of $n_2$. Popular methods for measuring $n_2$, like Z-scan [1], are based on the assumption that the temporal and spatial intensity distributions of the laser radiation are in the form of a Gaussian function. In practice this is sometimes difficult to fulfill, especially in the infrared (IR) wavelength range between 2μm and 5μm, where there is a shortage of information about $n_2$. Therefore, a method for measuring $n_2$, which could include real spatial and temporal intensity distributions, is of great importance.
In this work it was chosen to measure $n_2$ by using an interferometer to detect a total on-axis nonlinear phase shift (B) after a sample and including measured spatial and temporal intensity distributions by numerically simulating the experiment. First, a theoretical model of a measurement was created. It was assumed that two identical linearly polarized waves with a plane wavefront are propagating at a small angle (as shown in Fig. 1). Wave1 propagates through a sample with $n_2$ and then is imaged on a charge-coupled device (CCD) matrix of a camera by two converging lenses of the same focal length, without any energy (E) loss and change in peak intensity. Both waves reach camera at the same time and an interference intensity pattern is formed in a form of fringes. Here B was found by fitting spatial intensity distribution (which can be any function) of wave1 to a spatial phase difference between two waves distribution (obtained from an interference pattern using Fourier transform method [2])

During numerical experimentation it was found that (because of temporal intensity modulation of a wave) this way obtained B value is less than the real value by some factor. This factor depends only on a shape of a laser pulse. Therefore, one can measure real pulse shape and then simulate the real experiment to correct the measured B values. This was demonstrated by building a real experimental setup, which realizes the model shown in Fig. 1, and measuring the $n_2$ of fused silica. The values of B were measured at different pulse energies. The results are shown in Fig. 2. Obtained B values were corrected according to the measured pulse shape and the $n_2$ was found from linearly fitted experimental data. Measured $n_2$ value was in good agreement with the ones given in the scientific literature.

In conclusion, a method to include real temporal and spatial intensity distributions of light in $n_2$ measurements was successfully prepared and tested. It could be used to measure $n_2$ in the IR wavelength range between 2μm and 5μm, where such measurements are needed.