Abstract

In precision optical measurements, detection linearity often constitutes a challenge, in particular for signals with a broad range of intensities. A prominent example is broadband molecular vibrational spectroscopy in the mid-infrared (MIR) spectral region (2.5-25μm), which has traditionally been limited by the sensitivity and dynamic range of HgCdTe detectors. Recently, electro-optic sampling (EOS) has set sensitivity and dynamic-range records for detection in the MIR range [1,2]. The most sensitive EOS measurements to date employ phase-matched sum-frequency generation between MIR wave and near-infrared gate pulse in a thick EOS crystal. They reach a fielddetection sensitivity within a factor of 4 from the ultimate limit of detecting all MIR photons in the temporal gate, while at the same time achieving a detection dynamic range of >14 orders of magnitude in intensity [2]. However, in this “phase-matched EOS” configuration [3], the MIR signal is strongly depleted upon interaction with the gate pulse through the EOS crystal [2]. This is in contrast to most previous EOS implementations, where both sum-and difference-frequency generation are simultaneously phase-matched, which requires thin nonlinear crystals [1,4]. Here, we study the linearity of high-sensitivity phase-matched EOS experimentally and numerically. The latter uses a 3+1-dimensional solver of the nonlinear wave equation, considering 2^nd and 3^rd order nonlinear processes [5]. We find that a high gate-pulse energy benefits both sensitivity and linearity of EOS detection, and that linearity is preserved to within a few percent deviation even for MIR power depletions reaching ≫ 10%.

© 2023 IEEE