摘要： Electro-optic sampling (EOS) allows for field-sensitive detection of optical waveforms in a broad spectral range, extending from the far-infrared (THz) all the way to the near-infrared. Most recently, EOS has been employed for highly sensitive spectroscopy of liquids and gases in the mid-infrared (MIR) molecular fingerprint region, overcoming the notorious limitations of broadband detectors in this spectral range and providing a glance at the potential of this coherent detection scheme for molecular spectroscopy. In this paper, we investigate – in theory and experiment – the trade-off between quantum-efficiency/dynamic range (DR) and detection-bandwidth in EOS. The theoretical model describes the two-step-process of the EOS detection in one dimension, where the nonlinear up-conversion is calculated with the first order propagation equation, followed by heterodyne detection of the generated sum-frequency field with a strong gate pulse. In a first experiment, we sample an octave-spanning spectrum centred at 8.6 μm, using high-power, 16-fs pulses spectrally centred at 1030 nm. We show a frequency up-conversion-efficiency of 2%, which is unprecedented in EOS, measured as the maximum delay-dependent MIR power decrease behind the EOS crystal (500-μm thick GaSe). On the other hand, the maximum MIR depletion is only 0.3% for a 85-μm thick crystal, which is due to the shorter interaction length. However, this thinner crystal greatly benefits the detection bandwidth, as can be seen from the simulations. As our heterodyne detection with unprecedented local oscillator powers (450 mW at 28 MHz) is shot-noise-limited, this conversion efficiency results in an unprecedented amplitude DR of 6.5×106 (for 20 s of measurement time). This DR directly benefits spectroscopic measurements with unprecedented sensitivities. Our simulations show beneficial phase matching for sampling pulses centred at 2 μm. Therefore, the bandwidth-DR trade-off is expected to be less severe and we are currently investigating the 2 μm-based detection experimentally.