Fig. 1. Basic principles. Schematics of (a) difference-frequency generation, (b) spectral focusing, and (c) optical sampling with asynchronous pulses. (d) Modulating the repetition frequency of signal pulses to improve the scan rate. CPLN, chirped-poling lithium niobate; $t$, the measurement time; $\tau$, the relative time delay between the pump and signal pulses; $f_r$, the repetition frequency of the pump laser; $\mathrm{\Delta }{f}_r$, the repetition frequency difference; $f_D$, the center frequency of the generated idler beam; $f_m$, the modulation frequency.

Fig. 2. Generation of broadband tunable mid-infrared light. (a) Experimental setup. CW, continuous-wave laser; IM, intensity modulator; PPG, picosecond pulse generator; DC-EDFA, double-clad Er-doped fiber amplifier; HNLF, highly nonlinear fiber; PMF, polarization maintaining fiber; YDFA, Yb-doped fiber amplifier; Col, fiber collimator; M, mirror; $\lambda /2$, half-wave plate; PCF, photonic crystal fiber; DM, dichroic mirror; CPLN, chirped-poling lithium niobate crystal; LPF, long-pass filter; MCT, HgCdTe photodetector. (b) Spectral characterization of the pump and the signal light. (c) Measured autocorrelation traces of the pump and the signal pulses. (d) Spectra of the narrowband mid-infrared light measured with a commercial spectrometer. The spectral intensity is normalized, and the blue lines depict the relative intensities of different spectral components.

Fig. 3. Results of temporal measurements and spectral reconstruction. (a) Pulse trains recorded in the time domain. (b) Enlarged view of the recoded pulses. The pulses separated by 16.5 ns were determined by the pump laser’s repetition frequency. (c) The normalized spectrum reconstructed from a single measurement and its comparison with the reference trace measured by a commercial spectrometer. (d) Signal-to-noise ratio (SNR) versus number of averages.

Fig. 4. Broadband spectral results. The reconstructed spectra of (a) dimethyl sulfoxide (DMSO) and (b) ethanol with 100-fold averaging. (c) Broadband absorption spectrum of flaxseed oil. In panel (c), the black curve is measured by an FTIR at a spectral resolution of $8{\mathrm{cm}}^{-1}$.

Fig. 5. Spectral tuning at high scan rates. Mid-IR pulse traces are recorded at (a) $2f_m=600\mathrm{kHz}$ and $\mathrm{\Delta }{f}_r=60\mathrm{kHz}$ and (b) $2f_m=2\mathrm{MHz}$ and $\mathrm{\Delta }{f}_r=200\mathrm{kHz}$. A comparison of spectral parameters between (c) unmodulated and (d) frequency-modulated asynchronous schemes. For the plots in panels (c) and (d), we assume a full spectral span of $1000{\mathrm{cm}}^{-1}$ and an effective time duration of 100 ps. The spectral width of a single pulse is fixed at $10{\mathrm{cm}}^{-1}$. The repetition frequencies of the fiber laser and the EO comb are $f_r=60.5\mathrm{MHz}$ and $5f_r+5\mathrm{\Delta }{f}_r=302.5\mathrm{MHz}+5\mathrm{\Delta }{f}_r$. Specifically, in panel (d), $\mathrm{\Delta }{f}_r$ is fixed at 18 kHz.