1 Introduction

A lot of the strongest fundamental vibrational absorption bands for various gaseous substances including hazardous (toxic, drug and explosive) ones are in the mid-IR range. Therefore, this range is often called as the “molecular-fingerprint region” (Picqué and Hänsch 2019). Accordingly, to solve a number of problems (for example, environmental monitoring and air control of a working area, analysis of processed gases and gas mixtures composition, detection of leaks in main gas and oil pipelines), the development of broadband mid-IR laser sources is required.

One of the most efficient amongst such laser sources is a CO laser. The cascade mechanism of its lasing leads to the fact that the CO laser has the highest efficiency among gas-discharge lasers, in the multiline mode reaching 50% for fundamental ro-vibrational transitions (Ionin 2007) and 16% for overtone ro-vibrational transitions (Ionin et al. 2006). This laser can generate on the large number (hundreds) of spectral lines in a broad wavelength range (Ionin 2007). For the fundamental vibrational bands, this range spans from 4.6 µm (McCord et al. 2000) up to 8.7 µm (Ionin et al. 2017a), and for overtone vibrational bands, from 2.5 µm up to 4.2 µm (Ionin et al. 2010). In addition, many spectral lines of the CO laser are within the spectral intervals of low atmospheric absorption, i.e. within the atmospheric transparency windows (Ionin et al. 2013, 2022a).

However, a number of problems require radiation with wavelengths that go beyond CO laser spectral bands. One of the methods for expanding and enriching the spectrum of a CO laser is frequency conversion of its emission in nonlinear crystals.

Due to the latter, the Gas Laser Laboratory of P. N. Lebedev Physical Institute of RAS has implemented broadband sum-frequency generation (SFG) in several nonlinear crystals: ZnGeP2 (Ionin et al. 2018), AgGaSe2 (Budilova et al. 2016), BaGa2GeSe6, GaSe (Badikov et al. 2018), and PbIn6Te10 (Ionin et al. 2016) with external conversion efficiency of 2.5%, 0.6%, 0.5%, 0.15% and 0.01%, respectively. It should be noted that these experiments were carried out with the same Q-switched CO laser and under the same conditions of focusing into a nonlinear crystal (by a lens with the focal length F = 20 cm). Therefore, after comparing the obtained results, we can claim that just the ZnGeP2 nonlinear crystal is the most efficient for this type of conversion. The CO laser SFG efficiency enhancement in nonlinear crystals ZnGeP2 is possible by several procedures.

Firstly, an increase of the SFG efficiency is possible due to a sharper focusing (Andreev et al. 2013). In this case, the maximum external efficiency of the SFG was achieved by focusing with a lens with F = 11.5 cm, which made it possible to operate at laser intensities on the crystal surface close to the optical damage threshold (for a ZnGeP2 crystal 78 MW/cm2 (Dmitriev et al. 1999)). At the moment, the external efficiency of the SFG is 3.4% that, taking into account Fresnel reflection losses from uncoated crystal faces, corresponds to the internal efficiency of 6.5%. However, this result was obtained nearly to the optical damage threshold of a ZnGeP2 crystal.

Secondly, it is potentially possible to increase the SFG efficiency by using an intracavity SFG scheme (Ionin et al. 2022b). In our intracavity SFG experiments, the efficiency of 2.9% (Ionin et al. 2022b) was obtained at laser intensity on the crystal surface of 0.9 MW/cm2 that is 85 times lower than the optical damage threshold for a ZnGeP2 crystal. Calculations performed in (Ionin et al. 2022b) demonstrated following results: if the intracavity laser intensity on the crystal surface increases up to 10 MW/cm2, (for example, due to an additional intracavity focusing lens (Petukhov et al. 2001)), there is some potential to achieve intracavity SFG efficiency up to 25%.

Thirdly, conversion efficiency can be increased by an application of a double-pass scheme. The double-pass scheme provided about 30% higher efficiency of CO laser SFG in a ZnGeP2 crystal as compared to a traditional single-pass one (Kinyaevskiy et al. 2023), while the spectral bandwidth of frequency converted radiation was the same. Moreover, a feasibility of slight spectral bandwidth extension by a variation of a reverse beam direction was demonstrated for the double-pass scheme. It must be noted that the double-pass SFG scheme automatically solves the problem of SFG radiation walk-off effect (Bhar et al. 1990) in contrast to an optical scheme with two crystal samples (Antipov et al. 2017).

The fourth option for increasing the conversion efficiency is application of a nonlinear crystal AR-coating which is the objective of the given paper research. This research became possible due to the development of new broadband multirange AR-coatings (Zinovev et al. 2022) for the ZnGeP2 crystal with the high optical damage threshold both in the CO laser spectral range (5–7 µm) and in the sum-frequencies range (2.5–3.5 µm). Thus, in this paper the broadband SFG of multiline Q-switched CO laser radiation was studied in AR-coated ZnGeP2 crystal samples under single-pass and double-pass schemes.

2 Experimental setup

The experiments were carried out with a low-pressure cryogenic Q-switched CO laser pumped by a DC discharge. Figure 1 shows the general optical scheme of the experiments. The active medium of the CO laser was mixture He:N2:CO:Air = 140:11:2:1 at gas pressure of 7.7 Torr. The laser cavity was formed by high reflecting spherical mirror 1 (the radius of curvature r = 9 m) and a flat output mirror with ~ 25% transmittance. A high reflecting rotating mirror was placed inside the laser cavity, which provided the laser operation in the Q-switched mode with repetition rate of ~ 60–130 Hz.

Fig. 1
figure 1

General optical scheme of the experiment

Full size image

The CO laser spectrum consisted of ~ 60 spectral lines in the wavelength range from 4.9 to 6.2 μm, its maximum being near the wavelength of 5.2 μm (see Fig. 2). The CO laser total peak power integrated over its spectrum was up to ~ 4.5 kW. The peak power of the CO laser individual spectral lines in Fig. 2 was calibrated to the integrated peak power.

Fig. 2
figure 2

The CO laser spectrum

Full size image

To control the CO laser average output power, a part of its emission (~ 6%) was directed into Ophir-3A power meter by a BaF2 plate and spherical mirror 2 (r = 1 m). The second BaF2 plate was used to control the laser pulse temporal shape with PEMI-10.6 photodetector (time resolution 1 ns).

In the experiment, both the single-pass and double-pass SFG schemes were used. In the first case, the main part (~ 94%) of the CO laser emission was focused into a nonlinear AR-coated ZnGeP2 crystal with a CaF2 lens (the focal distance F = 200 mm). After the crystal, the radiation was directed by spherical mirror 3 (r = 50 cm) to the SFG characteristics measuring system.

In the second case, the main part of the radiation was focused into the ZnGeP2 crystal with the same lens (F = 200 mm) through a flat dichroic mirror (reflection coefficient R ~ 99% in the SFG wavelength range and ~ 10% in the CO laser wavelength range). After the crystal, the CO laser beam was reversed into the crystal by spherical mirror 4 (r = 30 cm) either straight back (the deflecting angle β = 0°) or deflected by a small angle in the horizontal plane (β = 1.0°), (see Fig. 3). In Fig. 3, the angle α denotes the CO laser beam angle of incidence on the nonlinear crystal during the first pass.

Fig. 3
figure 3

Scheme of a double pass through the AR-coated ZnGeP2 crystal: α is the incidence angle of the laser beam on the crystal; β is the deflecting angle of the laser beam in the horizontal plane

Full size image

The SFG radiation obtained as a result of two passes of the CO laser emission through the ZnGeP2 crystal and a part of the CO laser radiation were reflected by the dichroic mirror to spherical mirror 5 (r = 60 cm), and, then, directed to the SFG characteristics measuring system. A quartz plate was used as a spectral filter to separate SFG radiation from the CO laser emission. The following SFG characteristics were measured: a temporal pulse shape, average power, and spectrum. With spherical mirror 6, the SFG radiation was focused onto Ophir-3A power meter to measure the average SFG power. To determine the pulse shape, it was directed onto PEM-L-3 photodetector (time resolution 0.5 ns) with a plane-parallel BaF2 plate. To measure the SFG spectrum, its radiation was directed to the input slit of IR spectrometer IKS-31, LOMO Ltd (spectral resolution ~ 1 nm). Photodetector Thorlabs PDA20H was installed at the IR spectrometer output slit. A He–Ne laser beam was used to align the optical scheme.

The experimental sample of nonlinear ZnGeP2 crystal of 8 × 8 × 8 mm3 volume was grown and AR-coated on both sides at «Laboratory of optical crystals LLC, Russia». Transmittance of the sample was ~ 85–90% within the SFG wavelength range and ~ 92–95% within the CO laser wavelength range (see Fig. 4).

Fig. 4
figure 4

Transmittance and photos of AR-coated and uncoated samples of the nonlinear ZnGeP2 crystal

Full size image

Figure 5 demonstrates a dependence of a phase-matching angle for second harmonic generation (a particular case of SFG) by type e + e → o in ZnGeP2 crystal on a pump (CO laser) wavelength. Also, CO laser spectrum in a logarithmic scale transferred from Fig. 2 is indicated in Fig. 5. Since the CO laser spectrum is located near the dependence minimum, the noncritical spectral phase-matching for SFG of multiline CO-laser radiation takes place (Andreev et al. 2013). Thus, broadband SFG with a great number of frequency conversion lines is possible in the ZnGeP2 under the crystal fixed angle.

Fig. 5
figure 5

Dependence of a phase-matching angle for second harmonic generation (a particular case of SFG) by type e + e → o in ZnGeP2 crystal on a pump (CO laser) wavelength and CO multiline laser spectrum in a logarithmic scale

Full size image

3 Experimental results and discussion

3.1 Single-pass SFG scheme

3.1.1 SFG time characteristics

Figure 6 shows the temporal pulse shapes of the CO laser and SFG pulses at various mirror rotation frequencies (Q-switching frequencies). These curves are the waveforms of the CO laser pulse incident on the crystal and the SFG pulse coming out of the crystal representing the time dependence of the respective powers. The peak power of the CO laser \({{P}}_{{{\text{peak}}}}^{{{\text{CO}}}}\), i.e. the maximum radiation power per pulse, increases with Q-switching frequency rise reaching its maximum at frequency f = 86 Hz, and then begins to decline. This decline is associated with a sharp decrease in the pulse energy, because as the mirror rotation speed increases, the time the resonator remains in the adjusted state declines and lasing does not have time to be developed effectively. A shape of the CO laser pulse changed from “triangular” to “Gaussian” with the Q-switching frequency increase.

Fig. 6
figure 6

CO laser (left axes) and SFG (right axes) temporal pulse shapes at different mirror rotation frequencies: a 63 Hz, b 86 Hz, c 132 Hz

Full size image

Measured FWHM duration of the CO laser pulse (\({{t}}_{1/2}^{{{\text{CO}}}}\)) and the SFG pulse (\({{t}}_{1/2}^{{{\text{SFG}}}}\)) at different Q-switching frequencies (f) are presented in Fig. 7. They had a general trend of declining with Q-switching frequency increase. But CO laser and SFG pulse durations were the same within measurement error bars at low f ~ 60 Hz and notably different at high f ~ 130 Hz.

Fig. 7
figure 7

FWHM duration of the CO laser (triangles) and SFG (circles) pulses versus Q-switching frequency f

Full size image

To explain this fact, we should consider the SFG theory taking into account a peculiarity of CO lasing time behavior on different spectral lines. The SFG power P3 and the conversion efficiency K can be represented by expressions (1, 2), see, for example, the Handbook of Nonlinear Optical Crystals (Dmitriev et al. 1999):

$$P_{3} = \frac{{8\pi^{2} d_{{{\text{eff}}}}^{2} L^{2} P_{1} P_{2} }}{{\varepsilon_{0} cn_{1} n_{2} n_{3} {\uplambda }_{3}^{2} A}}{\text{sinc}}^{2} \left( {\frac{{\left| {\Delta k} \right|L_{{{\text{eff}}}} }}{2}} \right),{ }$$
(1)
$$K = \frac{{8\pi^{2} d_{{{\text{eff}}}}^{2} L^{2} P_{1} }}{{\varepsilon_{0} cn_{1} n_{2} n_{3} {\uplambda }_{3}^{2} A}}{\text{sinc}}^{2} \left( {\frac{{\left| {\Delta k} \right|L_{{{\text{eff}}}} }}{2}} \right),{ }$$
(2)

where deff is effective nonlinear coefficient; ni are the refractive indices at λi; λ3 is SFG wavelength; ε0 is vacuum permittivity; P1 и P2 are power of CO laser lines; P3 is SFG power; Δk is wavelength mismatch; Leff is effective crystal length; A is cross-sectional area of the laser beam; c is speed of light in vacuum.

According to expression (1), the SFG power is proportional to the product of the pump lines peak power. Therefore, at exact (ideal) temporal matching of the pump pulses (as well as for the second harmonic generation), the SFG pulse duration is \(\sqrt 2\) shorter than the pump pulse duration. In our case, this is clearly not so, especially, at low Q-switching frequency (see Fig. 7). This is due to the fact that the CO lasing time behavior on individual ro-vibrational transitions is not the same even in the Q-switching mode (Ionin et al. 2017b). Such a behavior resulted in the “triangular” shape of the integrated over their spectrum CO laser and SFG pulses (see Fig. 6a) with similar duration at the low Q-switching frequency. At the high Q-switching frequency of 130 Hz the SFG pulse duration was 1.25 times shorter than that of the CO laser (tending to the \(\sqrt 2\) ratio) because lasing dynamics on different CO laser transitions has a better temporal matching which was demonstrated in (Ionin et al. 2017b).

3.1.2 SFG power and conversion efficiency

Figure 8a demonstrates the dependences of the CO laser (PCO) and SFG (PSFG) average power on the Q-switching frequency f. Figure 8b shows the CO laser peak power (\({{P}}_{{{\text{peak}}}}^{{{\text{CO}}}}\)) and SFG peak power (\({{P}}_{{{\text{peak}}}}^{{{\text{SFG}}}}\)) on the frequency f. Figure 8c demonstrates the average-power conversion efficiency (Kav = PSFG/PCO) and peak-power conversion efficiency (Kpeak = \({{P}}_{{{\text{peak}}}}^{{{\text{SFG}}}}\)/\({{P}}_{{{\text{peak}}}}^{{{\text{CO}}}}\)) as functions of the frequency f.

Fig. 8
figure 8

a Average power of the CO laser radiation incident on the crystal PCO and the SFG PSFG average power versus Q-switching frequency f; b peak power of the CO laser \({{P}}_{{{\text{peak}}}}^{{{\text{CO}}}}\) and SFG radiation \({{P}}_{{{\text{peak}}}}^{{{\text{SFG}}}}\) versus f; c average conversion efficiency Kav and peak conversion efficiency Kpeak versus f

Full size image

The maximum values of both the average power and peak power of the CO laser emission incident on the crystal (it will be called just the CO laser emission or beam further on) and SFG radiation, as well as the average power and peak power conversion efficiency, were achieved at frequency f = 86 Hz. It can be seen that after passing the maximum, the average power conversion efficiency decreases from Kav = 4.3% down to Kav = 3.4%, while the peak power conversion efficiency changes just a little remaining in the range of 4.25–4.5%. This can be explained by the fact that the SFG peak power does not go down as sharply as that of CO laser due to a better SFG conversion caused by the better temporal matching of CO laser pulses on different ro-vibrational transitions.

3.1.3 SFG spectrum

Figure 9 shows the SFG spectrum for the case of a single-pass obtained under SFG conversion efficiency Kav = 3.7 ± 0.3% (the incidence angle of the CO laser beam α = 0°). It contained ~ 40 detected spectral lines in the wavelength range from 2.52 µm up to 2.65 µm with the maximum of the SFG spectrum at λ ~ 2.58 μm. The spectrum bandwidth was ~ 0.1 μm at 0.1 level.

Fig. 9
figure 9

SFG spectrum observed in the single-pass scheme

Full size image

3.2 Double-pass SFG scheme

3.2.1 SFG average power and conversion efficiency

To increase SFG efficiency the double-pass scheme was implemented. Figure 10 demonstrates the dependences of the SFG average power and the average power conversion efficiency Kav on the CO laser average power for one and two passes through the ZnGeP2 nonlinear crystal (f = 103 Hz, α = 0°, β = 0°). The CO laser average power was controlled by a variable attenuator based on a mid-IR polarizer. In both cases, the SFG average power grew up quadratically, and the conversion efficiency Kav grew up linearly with the CO laser average power increase, as it was supposed by the expressions (1) and (2). Thus, these expressions adequately describe the experiment even in the case of broadband SFG of multiline CO laser under noncritical spectral phase-matching conditions.

Fig. 10
figure 10

Dependencies of the SFG average power PSFG a and the conversion efficiency Kav b on the CO laser average power for one and two passes

Full size image

Following the obtained dependencies, we calculated the gain factor for the SFG in the second pass through the ZnGeP2 sample:

$$G = P_{{{\text{SFG}}}}^{{\text{second pass}}} /P_{{{\text{SFG}}}}^{{\text{one pass}}} ,$$
(3)

where \(P_{{{\text{SFG}}}}^{{\text{second pass}}}\) — SFG power in the second pass through the ZnGeP2 sample, \(P_{{{\text{SFG}}}}^{{\text{one pass}}}\) — SFG power in the one pass through the ZnGeP2 sample. Figure 11 demonstrates the gain factor G for the SFG in the second pass through the ZnGeP2 crystal. The gain factor was from 2 to 3.5 depending on the CO laser average power. The reason for the gain factor to be higher than 2 means that the second pass provided not just linear addition to the single-pass SFG power, but did provide the nonlinear increase of SFG power. However, the gain factor went down with the CO laser power enhancement. At the moment, a reason for the SFG gain factor to go down in the second pass is yet unclear for us; this decrease may be related with frequency conversion saturation for the strongest SFG spectral lines.

Fig. 11
figure 11

Gain factor of the SFG radiation in the second pass versus CO laser average power

Full size image

In the next experiments, we studied an effect of CO laser beam incidence angle (α) on the SFG power, conversion efficiency, and SFG spectrum. This angle was changed by rotating the nonlinear crystal. Figure 12 shows the dependences of the average SFG power (PSFG) and conversion efficiency Kav on the angle of incidence α (two passes, β = 0°). It can be seen that the maxima of both PSFG and Kav take place at the angle of incidence ~ 0.7°. The maximum SFG average power was 13.9 ± 0.7 mW, that corresponded to Kav = 9.7 ± 0.5%. For the normal incidence, the conversion efficiency was slightly lower, Kav = 8.2 ± 0.5%. This indicates that the crystal cut angle does not perfectly match the optimal one for the broadband SFG. It can be noted that PSFG and Kav are above the level of 0.5 relative to their maxima in a wide range of incidence angles — from –3 to 0 degrees. This fact allows us to hope that under the intracavity conversion of CO laser emission in the AR-coated nonlinear ZnGeP2 crystal, it is possible to tune the SFG spectrum not only by changing the crystal temperature (Ionin et al. 2016), but also by rotating the crystal relative to the cavity axis.

Fig. 12
figure 12

SFG average power PSFG (rhombs) and conversion efficiency Kav (circles) versus the CO laser beam angle of incidence α for two passes (f = 106 Hz, β = 0°)

Full size image

3.2.2 SFG spectrum

Figures 13(a–c) show the SFG spectra in the case of two passes through the crystal with the reverse laser beam directed exactly backwards (the deflecting angle β = 0°). For the angles of incidence α = 0° and α = –1.0°, it contained ~ 80 spectral lines in the wavelength range from 2.52 µm up to 2.75 µm. For α = –1.4°, it contained ~ 130 spectral lines in the wavelength range from 2.52 μm up to 2.8 μm, the maximum of the SFG spectrum being shifted to 2.65 μm.

Fig. 13
figure 13

SFG spectrum at different angles of incidence (α) of the CO laser beam on the ZnGeP2 crystal. Second pass back at the deflecting angle β = 0°, a α = 0°, b α = –1.0°, c α = –1.4°; second pass back at the deflecting angle β = 1.0°, d α = 0°, e α = –2.0°, f α = –4.0°

Full size image

Figures 13(d–f) demonstrate the SFG spectra for the case of two passes, while the laser beam traveled back at the deflecting angle β = 1.0°, i.e. the phase-matching angle in the ZnGeP2 crystal was slightly different in the second pass. The results show for the spectrum at the angle of incidence α = 0° to be the narrowest one, the number of SFG lines being ~ 55 in the wavelength range from 2.52 µm to 2.66 µm. Spectral width on level 0.1 in this case was 0.11 µm. The narrowest SFG spectrum obtained under these conditions was due to the fact that phase-matching conditions were shifted to the shortwave edge of the CO laser spectrum with higher steep slope in the SFG dependence in Fig. 5, and a small number of CO laser lines (near 5.1 μm) were converted.

At the angle of incidence α = –2.0°, the spectrum expanded to the long wavelength region up to 2.82 µm, the number of SFG spectral lines increasing up to ~ 185. At the angle of incidence α = –4.0°, the spectrum expanded up to 2.92 µm, the number of SFG spectral lines being ~ 190. Spectral width on level 0.1 in this case was 0.35 µm that was three times higher than one for the normal incidence (α = 0°, β = 0°) and for the single-pass scheme (see Fig. 9). Thus, this geometry (α = –4°, β = 1°) provided the best overlapping of phase-matching conditions of the first and second passes through the crystal to obtain the widest SFG spectrum involving almost all CO laser spectral lines.

4 Conclusions

A broadband sum-frequency conversion of multiline Q-switched CO laser emission under its single-pass and double-pass through an AR-coated ZnGeP2 crystal was experimentally studied, CO laser radiation being focused back into the crystal by a spherical mirror in the latter case. The maximum conversion efficiency in the double-pass scheme came up to ~ 10% that was 2.5 times higher than one in an uncoated nonlinear ZnGeP2 crystal for the same optical scheme (Kinyaevskiy et al. 2023). The control over the SFG spectrum was demonstrated by changing the CO laser radiation angle of incidence on the ZnGeP2 crystal within the wavelength range from 2.52 μm up to 2.92 μm. The broadest SFG spectrum was observed with the double-pass scheme when the CO laser beam was not directed straight back, but was deflected by the angle of 1.0° and passed through the crystal, the angle of incidence of the laser beam on the crystal being –4.0°. The number of SFG spectral lines came up to ~ 190. The SFG spectrum width on level 0.1 was three times broader as compared to the narrowest spectrum obtained in our experiments (the deflecting and incidence angles 1.0° and 0°, respectively).

The research was partially supported from the Russian Science Foundation grant № 22–22-20103, https://rscf.ru/project/22-22-20103/ and Tomsk region administration.