0.7 MW Yb:YAG pumped degenerate optical parametric oscillator at 2.06 μm

Highly sensitive and precise monitoring of short-lived climate pollutants in the atmosphere is critical for advanced studies on the carbon cycle, greenhouse gas balances, and disruptions. For example, methane is a key player among greenhouse gases for its substantial role in global warming and altering climatic patterns.^1,2 Although its presence in the atmosphere is considerably less than that of carbon dioxide—about 1.8 parts per million compared to carbon dioxide’s 400 ppm—methane’s effect on global warming is markedly more intense, with a warming potential 25 times greater than that of carbon dioxide, accounting for about 15% of future global warming scenarios. In addition, methane is a critical component in the feedback loops of atmospheric chemistry, acting as a signal of the Earth’s atmospheric dynamics on a large scale.^3,4

Time-domain broadband absorption spectroscopy offers great potential for on-site, highly precise, and sensitive detection of short-lived pollutants. It benefits from fast, self-calibrating measurements that do not require sample preparation, facilitating real-time atmospheric monitoring.^5–8 Atmospheric pollutants cover a broad range of fundamental vibration–rotation transitions in the mid-infrared and fingerprint region. Moreover, their overtone and combination bands at short-wavelength infrared (SWIR) range offer a sensitive detection window due to the lower absorption cross section of water in this spectral range. For instance, methane exhibits distinct absorption at 1.6 and 2.2 μm, where the water response is negligible.⁹ 

Based on the Beer–Lambert law, the detection signal-to-noise ratio in absorption spectroscopy is proportional to the brightness of the illumination source. Therefore, the detection sensitivity can be significantly enhanced by employing high-power sources operating at megahertz repetition rates. Moreover, when coupled with broadband, high-frequency modulation lock-in techniques, such frontends facilitate a further boost in detection signal-to-noise and sensitivity. Recent progress in frequency comb spectroscopy^10–13 and field-resolved spectroscopy^14–16 from the SWIR range up to ultraviolet spectral range has demonstrated an unparalleled ability to identify and classify molecular responses with exceptional sensitivity and accuracy. The detection sensitivity in both techniques benefits directly from the availability of bright sources with an intrinsic broadband modulation.

Femtosecond optical parametric oscillators (OPOs) synchronously pumped using Ti:Sa oscillators or fiber lasers have been a crucial tool for expanding the wavelength coverage of frequency combs, as the generated sub-harmonics are locked in frequency and phase to the input pump pulses.^17–22 Mainly, when operated at degeneracy, the signal and idler pulses with the same polarization become indistinguishable, exhibiting a self-phase-locked behavior through mutual injection, providing a single coherent broadband output centered at degeneracy.^23,24 On the other hand, diode-pumped Yb:YAG thin-disk oscillators are capable of delivering sub-100 fs pulses with tens of microjoule energy and hundreds of watts average power.^25–30 When used to pump OPOs, these advanced sources could potentially enable SWIR femtosecond sources to achieve peak powers in the mega-watt range and average powers in the tens of watts, operating at megahertz repetition rates. The spectral coverage of such powerful OPOs can be extended into fingerprint regions or even terahertz frequencies through downconversion processes. Furthermore, this scheme allows for high-bandwidth, dispersion-free modulation of output pulses at megahertz rates by precisely controlling the cavity length mismatch between the OPO and its pump laser.^31,32 This capability holds significant promise for enhancing detection sensitivity in spectroscopic applications, particularly in real-time atmospheric monitoring.

In this work, we present a degenerate OPO pumped using a Kerr-lens mode-locked Yb:YAG thin-disk oscillator delivering 114 fs, 2.8 W pulses at 30.9 MHz repetition rate, and 90 nJ averaged pulse energy. By operating the OPO cavity at twice the repetition rate of the pump cavity, broadband modulation of the output pulses is achieved.

The degenerate femtosecond OPO is pumped collinearly using a Kerr-lens mode-locked Yb:YAG oscillator, delivering 25 W, 15.45 MHz, 384 fs pulses at full-width at half maximum (FWHM), at 1030 nm [see Fig. 1(a)].³³ The pump mode is matched to the OPO cavity using a Kepler telescope with a ratio of 170%, resulting in a focus size of 74 μm at FWHM. This configuration corresponds to 61 GW/cm² pump peak intensity at the focus.

The optical ring cavity of the OPO operates at twice the repetition rate of the pump with a total length of 10 m [Fig. 1(b)]. The ring cavity allows for easy in- and outcoupling of the pump beam. Two concave sliver mirrors [M2 and M3 in Fig. 1(c)] with a radius of curvature of 700 mm are used to focus the cavity mode to the beam waist of 83 μm (FWHM). By having the cavity mode 1.1 times larger than the pump beam, an efficient energy transfer from the pump to the signal and idler pulses is ensured. Two dielectric plane mirrors with a low group delay dispersion are used for in-coupling and outcoupling of the pump beam [M1 and M4 in Fig. 1(c)]. A dielectric mirror with 30% transmission is used to couple out the OPO pulses. Several highly reflecting silver mirrors with a protective coating fold the cavity. A 1-mm thick 5% magnesium-doped periodically poled LiNbO₃ (PPLN) crystal at type-0 phase matching is used for degenerate downconversion of the pump pulses to 2.06 μm. The PPLN has a period of 30.8 μm and is placed in an oven at the temperature of 116 °C. The end-faces of the crystal are coated by a broadband anti-reflection coating with less than 0.5% reflectivity. The PPLN is placed behind the focus of the pump beam to avoid optical damage in the crystal.

Due to the large cavity length and environmental effects, including thermal effects, air fluctuations, and mechanical disturbances, active stabilization of the cavity is necessary. The stable operation of the OPO cavity is achieved by using a “dither-and-lock” technique, which follows the Pound–Drever–Hall stabilization technique widely used for locking lasers to a center of resonance of a cavity.^34,35 The schematic of the stabilization system is illustrated in Fig. 2(a). The system comprises a Red Pitaya, a voltage controller, two piezoelectric transducers (PZTs), and a slow photodiode. Red Pitaya is an affordable lock-in device based on a microcontroller board (STEMlab 125-14 board), integrating a lock-in amplifier and a PID controller. The feedback signal for cavity stabilization is measured at the OPO output using the photodiode (PDA100DT-EC, Thorlabs). One of the folding cavity mirrors [mirror M8 in Fig. 1(c)] is mounted on a slow PZT with an approximate speed of ∼0.2 μm/s (T-108-01, Piezosystem Jena) to scan the cavity length in the ramp mode. The fast PZT (PC4R10M, Thorlabs) is glued at the back of the M8 to generate an error signal in a dither mode.

The dither of the fast PZT induces a slight intensity modulation on the OPO output, monitored using the photodiode. The output signal of the photodiode is mixed with a sinusoidal local oscillator signal generated in Red Pitaya. The mixed signal is then passed through a low-pass filter (LPF) and the PID controller to generate a predicted value. This value is used as a feedback signal to the slow PZT to adjust the OPO cavity length on a micrometer scale. The proportional and integral functions of the PID controller are used to generate the feedback signal, with the derivative function deactivated to prevent the possible amplification of undesirable noise. The OPO operation was achieved in two modes, “dither mode” and “continuous mode.” In the dither mode, a periodic triangular signal (at a few Hz) is used to drive the PZT of M8. This mode is usually used for the alignment of the OPO. In the continuous mode, the OPO is locked to one of the resonance peaks.

The cavity supports degenerate and non-degenerate modes of operation due to the cavity dispersion.^19,37,38 Figure 2(b) shows the excited cavity modes when the OPO was operated in a “ramp mode.” Here, the intra-cavity piezo-actuator was driven with a linear ramp. The first three modes indicated by the green-shaded area are degenerate and near-degenerate modes, while the subsequent non-degenerate modes are shown in the violet-shaded area. For continuous operation, the cavity length of the OPO cavity can be locked to one of the modes. Figure 2(c) shows the spectra of a resonance mode (green area) and two non-resonant modes (purple area). The transition of the cavity from degeneracy to the maximum achievable spectral bandwidth of 1900–2200 nm is constrained by the reflectivity of the cavity’s optics. The phase-matching condition in the PPLN supports the wavelength tunability of the cavity. Figure 2(d) shows the output wavelength-tunability of the cavity in a PPLN with a period of 30.8 μm.

At degeneracy, the cavity operates at negative dispersion with an estimated total group delay dispersion of −88 fs². For intra-cavity compression of the degenerate OPO pulses, a 0.9 mm-thick Borosilicate plate is inserted into the cavity at a Brewster angle. The output pulses of the OPO are characterized by using second-harmonic generation frequency-resolved optical gating (SHG-FROG) comprising a 20 μm-thick BBO crystal. Figures 3(a) and 3(b) show the measured and retrieved spectrograms along the temporal profile of the OPO pulses corresponding to 114 fs at FWHM when the cavity operates at degeneracy. The spectral bandwidth of the OPO pulse supports 110 fs Fourier transform-limited pulses. Figure 3(c) shows the output power of the OPO vs the input pump power. The output power of the OPO is measured after two long-pass filters to exclude the contribution of the other nonlinear parasitic effects in the crystal, which are co-propagating with the OPO pulses. With a 30% output coupler, the OPO has 13% optical-to-optical efficiency and a lasing threshold at 4 W pump power. The OPO output average power has a standard deviation of 2%, while the pump laser has a standard deviation of 1%. The overall stability of the system can be improved by employing an air-tight cover for the cavity.

As the cavity length of the OPO is half that of the pump laser, the OPO pulse meets the pump pulse and obtains parametric gain after traveling two round trips in the OPO cavity. Therefore, the output pulse train of the OPO has a repetition rate of 30.9 MHz determined by the OPO cavity length. At this regime, the OPO output is modulated at 15.45 MHz with a broadband modulation depth of 41% [Fig. 1(b)]. The modulation depth between the two pulses can be tuned by varying the cavity loss in the OPO cavity. The modulation depth will be only a few percent for small cavity losses. The inherent capability for adjustable broadband modulation makes such cavities an ideal frontend for highly sensitive spectroscopic applications.

The availability of broadband, bright sources from the ultraviolet to terahertz range is crucial for sensitive, precise, and live monitoring of short-lived climate pollutants in the atmosphere.⁴⁷ The fundamental resonances of molecules have a unique fingerprint in the mid-infrared spectral range, allowing for precise spectroscopic selectivity. Overtone and combination bands of molecules at SWIR provide similar information with the advantage of lower water absorption cross section. Among various downconversion schemes aimed at generating broadband pulses for absorption spectroscopy across overtone, fundamental, and rotational resonances, OPO stands out for its higher conversion efficiency.^33,48–51 Yb:YAG thin-disk oscillators offer great potential to pump OPOs and to elevate the brightness, repetition rate, and peak power of the femtosecond pulses in these spectral ranges. Moreover, enhancing detection sensitivity in absorption spectroscopy requires implementing lock-in systems, necessitating high-rate signal modulation. Existing techniques, such as acousto-optic⁵² or mechanical modulation, suffer from limitations such as insufficient bandwidth, low-frequency modulation rate, high spectral dispersion, and low throughput.

This work presented the first Yb:YAG thin-disk pumped OPO containing a broadband, dispersion-free modulator. The OPO comprised a ring cavity and a 1 mm-thick PPLN crystal operating at 30.9 MHz repetition rate, twice the repetition rate of the Yb:YAG pump laser. The frontend delivers 114 fs, 2.8 W pulses at 2.06 μm with an average energy of 90 nJ. The wavelength tunability of the cavity from 1900 to 2200 nm is demonstrated by adjusting the cavity length, overlapping with several crucial atmospheric gasses resonances in SWIR, particularly with the water-free window of methane. Figure 4 shows a comparison of the performance of the frontend with the state-of-the-art sources at MHz repetition rates at overtone and combination band spectral regions. The peak power in this scheme can be scaled to a megawatt level by employing Yb oscillators with a higher peak and average power. Increasing the pump average power necessitates improved thermal management to maintain stable phase-matching conditions in the nonlinear crystal and effective cavity stabilization for a stable operation.^25–30 

The intrinsic broadband, dispersion-free signal modulation at 15.45 MHz offers a great opportunity for enhancing detection sensitivity in field-resolved spectroscopy. The system holds potential for further wavelength tunability by optimizing the intra-cavity dispersion or employing other nonlinear crystals. At the same time, the average power and peak power of the OPO pulses can be scaled by implementing higher energy Yb:YAG oscillators at MHz repetition rates, allowing for the downconversion of the OPO pulses to terahertz and mid-infrared spectral range.^53,54 With an intrinsic modulator, this scalable scheme holds promise to pave the way for atmospheric monitoring of short-lived pollutants and novel spectroscopic techniques.¹⁴ 

We thank Kilian Scheffter for the insightful discussion. This work was supported by research funding from the Max Planck Society.

The authors have no conflicts to disclose.

Anni Li: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Mehran Bahri: Data curation (supporting); Investigation (equal); Software (supporting); Writing – review & editing (supporting). Robert M. Gray: Software (equal); Writing – review & editing (supporting). Seowon Choi: Investigation (supporting); Software (supporting). Sajjad Hoseinkhani: Investigation (supporting). Anchit Srivastava: Methodology (supporting); Supervision (supporting). Alireza Marandi: Conceptualization (supporting); Software (supporting). Hanieh Fattahi: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Visualization (equal); Writing – original draft (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.