1. Introduction
Mid-infrared sources around 3~5 μm have been widely used in many fields, such as environment monitoring, medical diagnosis and Terahertz (THz) generation [
1,
2,
3,
4]. Recently, frequency conversion of single frequency (SF) lasers has attracted great attention [
5,
6,
7]. Two main methods have been used to achieve frequency down-conversion of SF lasers. The first is to utilize SF laser as the pump source of optical parametric oscillators (OPOs). OPOs are well-established mid-infrared sources, offering high power, tunable wavelength and broad bandwidth. However, OPOs require high power pump sources, and it is difficult for SF lasers to achieve high power output directly. The master oscillator power amplifier (MOPA) structure is an effective way to enhance power, but the system is relatively complicated, which includes a multi-amplifier stage. The fiber nonlinear effects such as stimulated Raman scattering (SRS), stimulated Brillouin scattering (SBS) and four-wave mixing are also easily induced during the amplification process. Besides, unexpected seed instability and back reflection from the output end may damage the MOPA system. In 2011, Liu et al. reported a mid-infrared singly-resonant OPO pumped by an SF pump source which was a MOPA system [
8]. When the pump power was 52.8 W, a 7.2 W idler power at 3.4 μm was achieved, indicating a conversion efficiency of 13.6%. In 2018, Shukla et al. demonstrated a singly-resonant OPO which was pumped by an SF Yb-fiber laser [
5]. When the pump power was 10.6 W, a 1.6 W idler power and the highest efficiency of 15% were obtained. However, a further increase in pump power led to a decrease in idler power, which limited the further improvement of conversion efficiency.
Another method is to use the difference frequency generation (DFG) process [
6], specifically, to mix the SF pump source around 1 μm and the signal source around 1.5 μm. DFG is a single-pass process with a relatively low efficiency. This method not only requires a high power SF pump source, but also a signal source with high enough power to realize the frequency down-conversion of the SF pump source. In 2018, Zhao et al. realized mid-infrared output using DFG between the SF pump source and signal source [
6]. Both sources were multi-stage amplified, which increased the damage risk and the complexity of the system. Meanwhile, the conversion efficiency was relatively low, where a 62.4 mW idler power was obtained under 23.2 W pump power and 4.31 W signal power.
Narrow linewidth mid-infrared sources have great potential in environment monitoring, gas trace sensing and medical diagnosis [
9,
10,
11]. The common method to achieve narrow linewidth mid-infrared radiation is to add etalons in the OPO cavity. In 2003, Herpen et al. demonstrated a continuous-wave (CW) singly resonant SF mid-infrared OPO which was pumped by a CW SF Neodymium-doped Yttrium Aluminium Garnet (Nd:YAG) source [
12]. A solid-state etalon was inserted in the resonant cavity to achieve a narrow linewidth output. However, it was found that the inserted etalon caused a reduction in idler power by approximately 50%. In 2018, Zhao et al. demonstrated a narrow linewidth mid-infrared output based on OPO [
7]. When adding an etalon to the cavity, the idler power was reduced from 1.3 W to 1.1 W. Undoubtedly, adding etalons in the cavity will cause additional losses and reduce the efficiency.
Our research group has made some achievements in utilizing a high-power intracavity signal beam to achieve mid-infrared output. In 2018, our group realized a tunable dual-wavelength mid-infrared output by using the 1060 and 1070–1090 nm tunable fiber laser as the pump source [
13]. By utilizing intracavity DFG, the 1070–1090 nm tunable fiber laser, whose power was insufficient to build the OPO process, was successfully converted to the mid-infrared region. However, the phase-matching conditions were not satisfied in the experiment, resulting in a relatively low conversion efficiency. Recently, we realized efficient frequency down-conversion of a low-power fiber source, based on the intracavity DFG process inside the periodically poled lithium niobate (PPLN) OPO. This method adopted a high-power assisted fiber laser to build a parametric oscillation and creatively took full advantage of the high-power intracavity resonant signal beam. For low-power fiber sources with different wavelengths, the signal wavelength was properly adjusted by changing the PPLN temperature or grating period to ensure that the DFG process between the pump source and signal beam satisfied the phase-matching conditions. In 2019, we reported this new scheme to achieve frequency conversion of the low-power source [
14]. The pump source consisted of a 1018 nm fiber laser and a 1080 nm fiber laser. Under 3 and 28 W of 1018 and 1080 nm simultaneously pumping and 40 and 2.4 W of 1018 and 1080 nm simultaneously pumping, respectively, intracavity DFG between the low-power pump beam and high-power resonant signal beam was achieved in both cases, thus verifying the scheme.
Based on the research results we have achieved, in this paper, we demonstrate the highly efficient frequency down-conversion of a low-power SF fiber source using phase-matched intracavity DFG. The SF pump source was fixed at 1064 nm and the maximum power was 340 mW. A high-power 1018 nm fiber source was used as the assisted laser to build a parametric oscillation and generate the intracavity signal beam. The PPLN temperature was also properly adjusted to ensure that the phase-matching conditions were met in the DFG process. At last, the 1064 nm SF pump beam was successfully converted to the 3.7 μm mid-infrared radiation. When the PPLN temperature was set at 20 °C, the maximum idler power reached 73.5 mW and the maximum pump-to-idler conversion efficiency reached 21.6%. The linewidth of 3.7 μm idler beam was as narrow as 271 pm (5.9 GHz) without adding any extra devices in the cavity, which also meant that the characteristics of the pump beam were successfully transferred to the idler beam. The experiment results prove that this method has great potential in the frequency down-conversion of low-power special fiber sources like frequency-swept lasers and generating special mid-infrared sources.
2. Experiment Setup
The experiment setup is shown in
Figure 1. The pump sources consisted of two CW linearly polarized fiber lasers, a low-power 1064 nm SF fiber laser and a high-power 1018 nm fiber laser, with a maximum power of 340 mW and 56 W, respectively. They were connected together using a polarization-maintaining 1018/1080 nm wavelength division multiplexer (WDM). As shown in
Figure 1, port 1 and 2 of the WDM were fused with the output fiber of two pump sources. Port 3 was the output port and was connected with a fiber collimator, and port 4 was a monitoring port used to monitor the working status of the two fiber sources. The OPO cavity configuration is similar to the setup described previously [
13], and the nonlinear crystal used in this experiment was a bulk of 50 mm × 5 mm × 1 mm 5% MgO-doped PPLN with a grating period of 29.8 μm. After the cavity was two dichroic mirrors, M5 and M6, which were used to separate the residual pump, signal and idler beams. In this experiment, an oven was used to control the temperature of the PPLN crystal for studying the frequency down-conversion of the 1064 nm SF pump beam under different phase-matching conditions.
3. Experiment Results and Discussion
The output characteristics of two fiber sources were measured firstly. The output dual-wavelength spectrum was measured using the Yokogawa AQ6370D Optical Spectrum Analyzer and is shown in
Figure 2a. Two spectral peaks appeared in the figure, located at 1018 nm and 1064 nm, respectively, indicating that the wavelength interval was 46 nm. The 1064 nm spectral peak was about 20 dB lower than the 1018 nm spectral peak.
Figure 2b reveals the measured interference result of 1064 nm pump laser using the Fabry–Perot interferometer. The blue and black lines represent the normalized transmissivity and scanning voltage, respectively. The figure indicates that the free spectral range (FSR) of the 1064 nm pump laser was about 4 GHz, which proved that the 1064 nm SF fiber source was in the steady operating state.
In this experiment, the 1064 nm pump power was too low to reach the oscillation threshold and build an independent parametric oscillation. Thus, in order to realize an effective frequency down-conversion, the phase-matching conditions need to be met, and the 1064 nm SF pump beam also needs to strongly interact with the intracavity signal beam. For confirming the most suitable PPLN temperature, the tuning curves corresponding to 1018 nm and 1064 nm pump beams at different temperatures were calculated based on the Sellmeier equations: [
15]
where a
1 = 5.756, a
2 = 0.0983, a
3 = 0.202, a
4 = 189.32, a
5 = 12.52, a
6 = 0.0132, b
1 = 2.86 × 10
−6, b
2 = 4.7 × 10
−8, b
3 = 6.113 × 10
−8, b
4 = 1.516 × 10
−4.
The simulation results were shown in
Figure 3a. The tuning curves at the same temperature were displayed in the same color. The x-coordinate of the dotted blue line was 29.82 μm, which was the adopted period in the experiment. The figure reveals that there is always an intersection between the 1018 nm and 1064 nm tuning curve, and the intersection wavelength changes with the temperature. Specifically, as the temperature changed from 10 °C to 60 °C in steps of 5 °C, the grating period of the crystal changed from 29.65 to 29.85 μm and the intersection signal wavelength changed from 1489.3 to 1490.9 nm. At the intersection, the frequency conversion of two pump beams could share the same signal beam and the phase-matching conditions were simultaneously met. The simulation results also indicate that the PPLN temperature should be adjusted to 20 °C to meet the phase-matching conditions. The tuning curves of 1018 and 1064 nm were further simulated and shown in
Figure 3b. The intersection of the two tuning curves corresponds to the signal wavelength of 1490.6 nm at 20 °C. The simulated idler wavelengths at the intersection were 3210 nm and 3718 nm separately.
After the PPLN temperature was set at 20 °C, the signal and idler spectrum were measured when the 1064 nm pump power was 340 mW and the 1018 nm pump power was 56 W. The center signal wavelength was 1492 nm with a full width at half maximum (FWHM) of 42 pm (3.5 GHz), as shown in
Figure 4a. The corresponding idler spectrum was measured using a Bristol 721B Optical Spectrum Analyzer, shown in
Figure 4b. The wavelengths of the two idler beams were 3206.3 nm and 3711.3 nm, which for convenience, were abbreviated as 3.2 μm and 3.7 μm, respectively. Clearly, the power of the 1064 nm SF laser was too low to build its own independent parametric oscillation, since the OPO threshold was about 13 W. Therefore, it could be concluded that the 3.2 μm beam was generated by the parametric oscillation of 1018 nm beam, while the 3.7 μm beam was generated by intracavity DFG between the low-power 1064 nm beam and the high-power 1492 nm intracavity signal beam. The generated wavelengths agreed well with the simulation in
Figure 3b, considering the thermal effect of the crystal [
16] and slight possible deviation of the temperature control system. Meanwhile, the FWHM of 3.2 μm and 3.7 μm idler beams were 777.98 pm (22.7 GHz) and 270.74 pm (5.9 GHz) separately. The narrow linewidth of 271 pm (5.9 GHz) corresponded to the frequency down-conversion of SF fiber laser, which proved that characteristics of the SF pump source were successfully transferred to the mid-infrared region.
Then the PPLN temperature was adjusted from 20 to 60 °C in steps of 5 °C to investigate the frequency conversion of the 1064 nm SF pump wave under different phase-matching conditions. The obtained 3.7 μm mid-infrared output power as well as the conversion efficiency was shown in
Figure 5a. As can be seen from the figure, the idler power firstly increased rapidly from 16.15 mW to 73.5 mW and then decreased to 3.35 mW, and the corresponding pump-to-idler conversion efficiency increased from 4.75% to 21.6% and then decreased to 0.99%. The maximum output idler power and conversion efficiency were achieved when the temperature was 20 °C. The experiment results clearly shown that the deviation in temperature could cause an obvious decrease in generated power and conversion efficiency. Taking the results under 15 °C and 20 °C as an example, the temperature only changed by 5 °C, but the 3.7μm idler power and conversion efficiency decreased from 73.5 to 18.8 mW and 21.6% to 5.5%, respectively. The results also proved that the importance of meeting the phase-matching conditions was highlighted and the selected temperature in this experiment was suitable.
Then we used a dichroic mirror (anti-reflection coated for 3.47–4 μm, T
> 95%, high reflectivity for 2.5~3.47μm, T < 1%) to separate the dual-wavelength mid-infrared beam. Under the phase-matching conditions, the output idler power and conversion efficiency at different 1064 nm pump powers is shown in
Figure 5b. The 3.7 μm idler power monotonically increased from 8.7 mW to 73.5 mW when the 1064 nm pump power increased from 73 mW to 340 mW. The corresponding conversion efficiency increased from 12% to 21.6%, and the slope efficiency of the whole parametric process reached 20.186%. Benefiting from the high intracavity signal power, the record conversion efficiency far exceeded the efficiency of other traditional single-pass DFG processes and was almost comparable to that of OPO processes. Some other near-infrared low-power sources, such as ultra-short pulsed lasers or frequency-swept lasers, would be of significant application values if they could be converted to the mid-infrared region. However, it is difficult for them to pump the OPO independently because of their low-power and limitations by fiber nonlinear effects during the amplification stages. The highly efficient frequency conversion of the low-power SF source in this paper paves the way for other low-power special fiber lasers to realize efficient frequency down-conversion by intracavity difference mixing.