Phase retrieval via regularization in self-diffraction based spectral interferometry

TL;DR

This paper introduces a new spectral phase interferometry method using self-diffraction for direct electric-field reconstruction, enabling characterization of ultrashort pulses beyond current techniques.

Contribution

The paper presents a novel SD SPIDER technique based on third-order nonlinear effects and regularization for phase retrieval, expanding spectral measurement capabilities.

Findings

Successfully demonstrated SD SPIDER with 40-fs pulses

Regularization effectively reconstructs fundamental spectral phase

Potential to measure bandwidths above two optical octaves

Abstract

A novel variant of spectral phase interferometry for direct electric-field reconstruction (SPIDER) is introduced and experimentally demonstrated. Other than most previously demonstrated variants of SPIDER, our method is based on a third-order nonlinear optical effect, namely self-diffraction, rather than the second-order effect of sum-frequency generation. On one hand, self-diffraction (SD) substantially simplifies phase-matching capabilities for multi-octave spectra that cannot be hosted by second-order processes, given manufacturing limitations of crystal lengths in the few-micrometer range. On the other hand, however, SD SPIDER imposes an additional constraint as it effectively measures the spectral phase of a self-convolved spectrum rather than immediately measuring the fundamental phase. Reconstruction of the latter from the measured phase and the spectral amplitude of the fundamental turns out to be an ill-posed problem, which we address by a regularization approach. We discuss the numerical implementation in detail and apply it to measured data from a Ti:sapphire amplifier system. Our experimental demonstration used 40-fs pulses and a 500 $\mu$m thick BaF${}_2$ crystal to show that the SD SPIDER signal is sufficiently strong to be separable from stray light. Extrapolating these measurements to the thinnest conceivable nonlinear media, we predict that bandwidths well above two optical octaves can be measured by a suitably adapted SD SPIDER apparatus, enabling the direct characterization of pulses down to single-femtosecond pulse durations. Such characteristics appear out of range for any currently established pulse measurement technique.