1. Introduction

Recently we investigated cascaded schemes for down-conversion to the mid-IR spectral range above 5 µm by implementing the second stage based on a non-oxide nonlinear crystal inside the cavity of a first-stage nanosecond optical parametric oscillator (OPO) based on Rb-doped periodically-poled (PP) KTiOPO₄ (Rb:PPKTP), pumped by a Q-switched Nd:YAG laser at 1064 nm [1–3]. Intracavity difference-frequency generation (DFG) [2,3] turned out to be more efficient compared to intracavity pumped OPO [1], however, wavelength tuning was restricted and a broadband dichroic half-wave plate was needed for either type-I or type-II phase-matching in the studied crystals of AgGaSe₂ and BaGa₄Se₇. Low-symmetry cubic semiconductors, such as orientation-patterned GaAs offer more options for phase-matching in which intracavity polarization rotation becomes redundant. With a similar spectral coverage of up to ~18 µm and 2 to 3 times higher effective nonlinearity, OPGaAs exhibits 50 to 100 times higher thermal conductivity, better damage resistivity, and offers an unique possibility for convenient temperature tuning in both stages in non-critical configurations. These advantages point at possible extensions to much higher repetition rates (1-100 kHz). On one hand such low energy regimes will utilize perfectly the limited aperture of the presently available OPGaAs samples but on the other hand will enable the generation of unprecedented average powers in the mid-IR spectral range. In this work we study intracavity DFG in OPGaAs inside (i) the same Rb:PPKTP OPO operated at 100 Hz [1–3] but using tighter focusing, and (ii) a temperature tuned MgO-doped PP LiNbO₃ (MgO:PPLN) OPO operated at 1-3 kHz which resulted in a substantial increase of the average power.

2. Experimental set-up

The epitaxial growth of orientation-patterned structures of the cubic (point group $\overline{4}3m$) GaAs revolutionized the nonlinear frequency conversion technology in the mid-IR. This is related to the extended transparency range (0.85–18.5 µm at the 3 cm⁻¹ absorption level), high thermal conductivity (55 W/mK), and high nonlinear coefficient (d₁₄ = 83-86 pm/V for second-harmonic (SH) generation at 10.6 µm) of GaAs [4]. The main limitation of OPGaAs related to the aperture of the periodic structure, prompted the study of frequency conversion schemes based on tight focusing: continuous-wave (CW) DFG and OPOs, high-repetition rate (~100 MHz) DFG or synchronously pumped OPOs, as well as nanosecond OPOs with repetition rates in the 1-100 kHz range [4]. For instance, such an OPO pumped by 150 ns pulses at 1.95 µm from a Tm-fiber laser system, delivered an average idler power exceeding 800 mW at 50 kHz for a wavelength of 10.6 µm [5]. It should be noted that when pumped by ultrashort (fs and ps) or short (ns) pulses, according to its band-gap, OPGaAs requires pump wavelengths exceeding ~1.75 µm in order to avoid two-photon absorption (TPA).

Apart from ~2-µm laser systems based on Tm- and Ho-ion doped materials, such frequency down-conversion schemes can be based on the cascaded approach starting with the more mature laser systems operating in the ~1-µm spectral range [4]. Here we implement this approach with the OPGaAs inside the OPO cavity.

The experimental set-up was similar to the one described in [2,3] with the wave plate removed but the pump beam down collimated. Two versions were used as shown in Fig. 1, with different optical elements as dictated by the different pump sources and nonlinear crystals in the first OPO stage. In the first version (a), the beam of the 100 Hz Nd:YAG pump laser (1.0642 µm, ~8 ns, M²~2, linewidth ~1 cm⁻¹) was down-collimated by a lens telescope to a Gaussian diameter of ~0.8 mm in the PPKTP crystal. The PPKTP crystal employed was 12-mm long (x-axis, propagation), 8-mm wide (y-axis) and 5-mm thick (z-axis, poling). The poled region (Λ₁ = 38.5 µm) was 8 mm long by 5 mm wide giving a useful aperture of the uncoated sample of ~5 × 5 mm². The OPO cavity was formed by two concave total reflectors, TR1 (Ag) and TR2 (Au), cf. Figure 1(a).

 

Fig. 1 Experimental set-up of the cascaded OPO-DFG pumped: (a) by a 100 Hz Q-switched diode-pumped Nd:YAG laser and (b) by a 0.1-5 kHz Q-switched diode-pumped Nd:YLF laser.

Download Full Size | PDF

In the second version (b), the 0.1-5 kHz Nd:YLF pump laser (1.053 µm, 17 ns, M²~1.3, linewidth ~1.8 cm⁻¹) was focused by a single lens to a Gaussian diameter of ~0.45 mm in the PPLN crystal. The PPLN crystal employed was 20-mm long, 10-mm wide and 3-mm thick (z-axis, poling). It had three sections with periods Λ₁ = 31.3, 31.4 and 31.5 µm, each of them 3-mm wide. The thickness of the poled region along the z-axis was 2 mm. The 10 × 3 mm² input and output faces were single-layer AR-coated for the pump wave. Optimum conditions for DFG in this case were provided by a curved TR1 (Ag) and a plane TR2 (Au), cf. Figure 1(b).

In both versions the pump beam was in- and out-coupled by two dichroic Infrasil mirrors DM1 and DM2, which were high transmission (HT) for the signal (S, T = 96%) and the idler (I, T = 91%), and high reflection (HR) for the pump (P) and its SH. DM3 was HR for the pump P but transmitted its SH. It was positioned at 75 mm from TR1 when the PPKTP crystal was used and at 95 mm from TR1 when the longer PPLN crystal was used, so that a double pump pass in the OPO crystal was realized, essential to reduce the OPO threshold, while the parasitic SH generated did not reach the DFG stage of the set-up.

The 45° dichroic ZnSe mirror DM4 was HR for signal S and idler I, 6 mm thick and AR-coated on the rear side for the DFG spectral range (T~97% for the resulting wavelengths). It served to extract the DFG radiation after a double pass through the OPGaAs crystal. Due to imperfect characteristics of DM1, DM2 and DM4, the OPO cavity had 6 parasitic outputs for signal S and idler I, shown in Fig. 1 by yellow arrows. In fact these parasitic outputs help to avoid damage in the absence of DFG as a loss mechanism to the OPO.

Two OPGaAs samples with different dimensions and grating periods were available for the present experiment. The dimensions of sample #1 were 12.2(long) × 5.8(wide) × 1.8(thick) mm³ and the thickness of the periodic structure (Λ₂ = 55.6 µm) was ~1.2 mm. The second sample #2 had dimensions of 25.7(long) × 8(wide) × 1.2(thick) mm³, with a thickness of the periodic structure (Λ₂ = 63.3 µm) of 0.6 mm. The input and output faces of these samples were AR-coated for the 2 µm spectral range and in the mid-IR up to ~5 µm. Only OPGaAs #1 was available for the experiments with the PPKTP OPO while both OPGaAs #1 and #2 were studied inside the PPLN OPO.

The signal S and idler I polarizations were both vertical and parallel to the pump P. Linear polarization of the DFG output can be obtained for parallel S and I polarizations (i.e. without a dichroic half-wave plate in the OPO cavity) when the latter are along the [110] or $\left[\overline{\text{1}}\text{1}0\right]$ crystallographic direction, cf. Fig. 2 [6]. In this configuration adopted on the present work the DFG beam is polarized in orthogonal (i.e. horizontal) direction along [001] and the effective nonlinearity is d_eff = d₁₄. The effective nonlinearity will be in fact slightly higher with all polarizations along [111], however, this advantage is partly canceled in our case by the characteristics of the dichroic ZnSe mirror DM4.

 

Fig. 2 Photograph of the AR-coated OPGaAs sample #2 with designation of the crystallographic directions.

Download Full Size | PDF

3. Results and discussion

Figure 3(a) shows the DFG pulse energy E_DFG obtained with the 100 Hz PPKTP OPO at ~7 µm, with both crystals at room temperature and for two physical cavity lengths, a short one (108 mm, minimum possible) and a long one (140 mm, to accommodate both heating units: a Peltier element for PPKTP and an oven for OPGaAs #1). The signal and idler were at 1.85 and 2.51 µm, respectively. As could be expected for the relatively short pump pulses, the longer cavity results in a smaller number of round trips and the output characteristics are accordingly inferior. Figure 3(b) shows the DFG tuning range obtained with the long cavity when heating both crystals (PPKTP up to 50°C and OPGaAs up to 140°C) in non-critical configurations.

 

Fig. 3 Input-output dependence obtained with OPGaAs #1 in the 100 Hz PPKTP OPO for two cavity lengths (a), and temperature tuning using the longer cavity (b).

Download Full Size | PDF

Calculations indicate that for a PPKTP temperature of 103°C the signal and idler wavelengths correspond to a DFG wavelength of 16.7 µm. The DFG process can be phase-matched at this wavelength at room temperature for an OPGaAs period of Λ₂ = 85 µm. Thus, by using a multiple grating (Λ₂ = 55-85 µm) OPGaAs sample at room temperature, one can expect to cover the 7 to 16.7 µm DFG spectral range while heating the present PPKTP sample (Λ₁ = 38.5 µm) only up to 103°C.

In the case of the PPLN OPO, both crystals (PPLN and OPGaAs) were placed in thermo-stabilized ovens. The OPO physical cavity length amounted to 135 and 148 mm with OPGaAs #1 and OPGaAs #2, respectively.

The performance of the PPLN OPO at 1 kHz without the OPGaAs crystal is shown in Fig. 4 for a temperature of 165°C which results in a signal wavelength of 1.84 µm and an idler wavelength of 2.462 µm. The energy is measured at the output (1), cf. Figure 1(b), and the distribution between signal and idler depends on the characteristics of DM1, DM2, and DM4.

 

Fig. 4 Input-output (1) characteristics of the PPLN OPO (Λ₁ = 31.5 µm) at 1 kHz without the OPGaAs crystal, for a cavity length of 135 mm.

Download Full Size | PDF

From Fig. 4, one can calculate an intracavity signal S energy of 182.5 µJ and an idler I energy of ~69 µJ at a pump energy of 1 mJ. These values decreased by roughly 1/3 with the OPGaAs #1 crystal in the cavity with the reduction approximately equally due to insertion losses and DFG conversion efficiency.

The DFG pulse energy E_DFG was measured for both OPGaAs samples versus repetition rate. It is shown together with the average DFG power P_DFG and the overall conversion efficiency η = E_DFG/E_p in Table 1. The DFG wavelength obtained with OPGaAs #1 at room temperature was 7.3 µm for the above specified signal and idler wavelengths at a PPLN temperature of 165°C (Λ₂ = 31.5 µm). With OPGaAs #2 at room temperature, the DFG wavelength was 9.2 µm obtained at PPLN temperature of 172.2°C (Λ₂ = 31.5 µm) which resulted in signal and idler wavelengths of 1.888 and 2.381 µm, respectively. The maximum overall conversion efficiencies in Table 1 correspond to quantum conversion efficiency of 4.2% for OPGaAs #1 at 2 kHz and 3.4% for OPGaAs #2 at 2.5 kHz. The lower efficiency for OPGaAs #2 is attributed to the thinner useful aperture of this sample. In general, the overall conversion efficiency from the pump to the DFG pulses is affected by the depletion of the signal pulses by the intracavity DFG but to a greater extent determined by insertion losses. Substantial improvement in terms of efficiency can be expected with optimized characteristics of the nonlinear crystals in both stages.

Table 1. Performance of the two OPGaAs DFG crystals in the PPLN OPO at the maximum incident pump energy E_p available for repetition rates f_p between 1 and 3 kHz.

View Table

Figure 5(a) shows the temporal pulse shapes measured. The pump P and signal S pulses were measured by a 70 ps InGaAs photodiode but the mid-IR pulse measurement at ~9.2 µm has to be corrected for the 2-ns rise time of the (HgCdZn)Te detector used which yields 11.3 ns for the FWHM. The results were very similar for the two OPGaAs samples.

 

Fig. 5 (a) Temporal shapes of the pump (P) pulse at 1.053 µm, the signal (S) pulse at 1.89 µm and the DFG pulse at 9.2 µm from the PPLN OPO, and (b) M² measurements of the DFG beam in the horizontal (h) and vertical (v) planes with 2D and 3D images of the spatial profiles recorded at 200 mm from the f = 50 mm BaF₂ focusing lens used (insets). For both measurements OPGaAs #2 was employed at 1 kHz.

Download Full Size | PDF

Figure 5(b) shows the DFG beam profiles recorded at 1 mJ pump energy (1.053 µm) and the results of the M² fits by Gaussian diameters D_g evaluated at the 1∕e² intensity level. Measured M² values for both OPGaAs crystals were also very similar and represent an improvement of about an order of magnitude compared to the results at low repetition rates with large beam diameters [2,3].

4. Conclusion

In conclusion, we studied intracavity DFG in a doubly-resonant OPO achieving an average power of >10 mW (2 kHz) at ~7.3 µm equivalent to an overall conversion efficiency from the 1.053 µm pump of ~0.6% and average power of ~7 mW (3 kHz) at ~9.2 µm equivalent to an overall conversion efficiency of ~0.37%. The results in terms of idler energy are comparable to the best reported with OPGaAs OPOs at similar thickness [6]. In terms of average power, they are comparable with the best results achieved with OPGaAs at similar repetition rates: e.g., 14 mW of combined signal plus idler output at degeneracy (~6 µm) was reported for an OPGaAs OPO pumped in a cascaded scheme by the idler pulses from a 2 kHz PPLN OPO [7] and an average power of 2.8 mW at ~8.5 µm can be calculated for the nanosecond OPGaAs based optical parametric amplifier described in [8] which also operated at a repetition rate of 2 kHz.

Somewhat thicker OPGaAs samples with optimized AR-coating characteristics and operation at ~100 kHz rate [5] are expected to boost the average power of the intracavity DFG source described in the present work to the 1 W level while multiple gratings or fan-out structures will enable continuous tuning of the wavelength in the entire clear transparency spectral range of OPGaAs in the mid-IR up to ~18 µm.