Optical frequency combs (OFCs) exhibit a comb-like structure in the frequency domain, characterized by equally spaced frequency intervals. Due to the properties of this comb-like structure, OFCs have become ideal light sources for precision spectral measurements. The resolution of comb spectroscopy is determined by the line spacing of the comb, which corresponds to the pulse repetition rate. Mode-locked frequency combs typically operate in the range of hundreds of megahertz to gigahertz. A smaller repetition rate results in a reduced frequency interval between spectral sampling points, thereby enhancing spectral resolution. However, for mode-locked frequency combs, achieving a low repetition rate generally necessitates an increase in cavity length, which consequently enlarges the physical dimensions of the mode-locked comb and hampers its applicability for outdoor spectral measurements. On the other hand, microcombs typically exhibit repetition rates exceeding gigahertz, which limits the spectral resolution due to the large frequency spacing. Considering the advantages of both mode-locked combs and microcombs, obtaining low repetition rate outputs from high repetition rate combs with pulse picking is one method toward achieving high-resolution frequency combs. Dual-comb spectroscopy employs two frequency combs with slightly repetition rate difference to transfer spectral information from the optical frequency domain to the radio frequency domain, offering benefits such as high coherence and sensitivity. When combined with pulse picking technology, dual-comb spectroscopy can obtain high-resolution frequency comb spectra.

In this study, we conducted time-domain pulse picking using two mode-locked OFCs with repetition rates of 250 MHz and 249.92 MHz. An arbitrary waveform generator (AWG) generated rectangular wave signals at frequencies of 31.25 MHz and 31.24 MHz with a duty cycle of 12.5%, which were used to drive an electro-optic amplitude modulator (EOAM). By actively controlling the bias point of the EOAM, we selected one pulse from every eight pulses produced by the two OFCs, while suppressing the remaining pulses. This approach effectively increased the pulse period of the combs to eight times the original duration, resulting in a corresponding reduction in the repetition rate to one-eighth of the original, while the number of comb teeth increased eight-fold. Both the frequency combs and the AWG were synchronized to a rubidium atomic clock. Following pulse picking, the high-resolution OFCs passed through a gas absorption cell, and the resulting radio frequency spectrum, obtained through multi-heterodyne detection, contained information regarding the gas absorption characteristics. A photodetector recorded the time-domain interferograms (IGMs), which was then processed using a low-pass filter (LPF) with a cutoff frequency of 13 MHz. Subsequently, coherence was restored through a self-correction algorithm based on the cross-ambiguity function.

After pulse picking, the period of the IGMs increased from $\mathrm{12.5}\mathrm{\mu }\mathrm{s}$ to $\mathrm{100}\mathrm{\mu }\mathrm{s}$ (Fig. 4), corresponding to a spectral interval in the radio frequency spectrum that changed from 80 kHz to 10 kHz (Fig. 5), consistent with theoretical analysis. Since the optical bandpass filter used after the combination of the dual-comb exhibits a Gaussian line shape, we performed a Gaussian fit on the regions of the radio frequency spectrum outside absorption as a reference. The ratio of the measured value to the reference represents the spectral transmittance. The minimum value of the radio frequency spectrum was fixed at the H¹³C¹⁴N gas P10 absorption peak, specifically at 193.45 THz. By multiplying the frequency axis of the radio frequency spectrum by the conversion factor, we obtained the transmittance spectrum in the optical frequency domain, with a measurement point spacing of 31.25 MHz. Compared to the 250 MHz spacing before pulse picking, this represents an eight-fold improvement in resolution. Voigt fitting was performed on both sets of spectral lines, yielding fitted curves with residuals for both cases below 0.05 (Fig. 6), thereby demonstrating accurate measurement of the H¹³C¹⁴N P10 absorption peak spectrum.

We proposed a method for achieving high-resolution dual-comb spectra based on pulse picking, which actively controls the bias point of the EOAM. An AWG drives the EOAM to output rectangular wave signals with specific frequencies and duty cycles for time-domain pulse picking of the OFCs. This approach resulted in an eight-fold increase in both the number of radio frequency comb teeth and the spectral resolution. By measuring the P10 absorption peak of gas, we obtained accurate results with a resolution of 31.25 MHz, thereby validating the feasibility of this method. This approach provides a method and experimental evidence for obtaining high-resolution spectra from high repetition rate mode-locked microcombs, and the dual-comb system is not complex, facilitating outdoor spectral measurements. The subsequent work may focus on reducing the modulation frequency or applying this method to higher repetition rate comb sources to demonstrate its broad adaptability and explore the limits of resolution.