High-energy picosecond kHz optical parametric oscillator/amplifier tunable between 3 and 3.5 µm

When seeded with the maximum microchip oscillator output of 300 µJ, the pulse energy of the amplifier at 1 µm reaches 18 mJ after four passes through the first amplification module and 38 mJ after further two passes through the second amplification module (see Fig. 3a). Using low-doped Nd:YAG rods in the two amplifiers, a uniform gain is achieved and the thermal effects are kept low so that beam quality deterioration is minimal. The M² factor is measured with a silicon CCD camera and a lens with focal length of 150 mm and the estimated values are 1.05 × 1.02 for the oscillator and 1.5 × 1.2 for the output of the MOPA (Fig. 3b). To measure the duration of the laser pulses, a fast InGaAs photodiode and a 4-GHz oscilloscope were used. The estimated value of the 1-µm-MOPA pulse duration is 800 ps after deconvolution with the response time of the measuring system (160 ps).

Open image in new window

As mentioned before, the output of the MOPA is divided into two beams: one is used to pump the OPO and the other to pump the OPA. The measured OPO threshold is 1.2 mJ of pump energy, corresponding to 50 MW/cm2 average pump intensity which is in agreement with the Brosnan and Byer theory [20], for the case of a singly resonant OPO with pump reflection. A maximum output energy of 0.47 mJ is achieved when 4.4 mJ of pump energy is applied, thus the idler conversion efficiency is 11% and the quantum conversion efficiency is 31% (see Fig. 4a). Even though in our previous work on OPO [16] a much higher energy was presented from an identical oscillator this approach based on short-cavity design, single OPO has significant drawbacks due to the small number of roundtrips that are realizable in the OPO cavity for the duration of the pump pulse which is ≤ 1 ns. This requires the length of the resonator to be determined by the length of the PPSLT nonlinear crystal and the transverse beam diameter from the damage threshold of the nonlinear crystals or rear mirror. According to our experience, around 3-micron spectral range, the limited factor is the damage of the rear mirror coating which reflects pump at 1 micron and idler around 3 microns, simultaneously. In addition, the shortest cavity design leads to larger Fresnel numbers and the radiation field modes arise which cause a decline in beam quality. However, in the current setup, the output energy and efficiency of the whole system are essentially determined by the OPA, so the OPO can be operated with low-energy pump which is also beneficial for the output beam quality.

Open image in new window

Using a knife edge technique [21], the beam diameter of the idler was measured at various positions near the waist using CaF₂ focusing lens with focal length of 50 mm. The estimated beam quality of M² = 31 × 32 is shown on Fig. 4b. To measure the idler pulse duration, we frequency doubled the idler pulse using birefringent phase-matched SHG in a 4-mm thick KTP crystal and measured the pulse duration of the generated second harmonic signal (SH) which was 450 and 330 ps for the OPA and OPO output, respectively. The pulse duration measurements have been done using a InGaAs photodiode connected to a 4 GHz digital oscilloscope. The response time of the measurement system is 160 ps and it was measured with a signal-to-noise-ratio better than 100, and the statistical errors and noise contributions are not substantial (e.g., the RMS of the FWHM of the record response functions is not greater than ~ 5 ps). After deconvolution, the pulse duration (FWHM) of the frequency-doubled idler is found to be 288 and 420 ps for OPO and OPA, respectively, which corresponds to ~ 400 and ~ 600 ps idler duration. The accuracy of the measurement is estimated to be ± 25 ps.

The idler output from the OPO is imaged in the OPA with two CaF₂ lenses with focal lengths of 100 and 200 mm resulting in a beam diameter of 3 mm which is slightly larger than the pump beam diameter of 2.8 mm. The maximum of the OPA-seeded idler energy is measured to be ~ 0.39 mJ at 3.1 µm idler wavelength. It is a reduced OPO output by the Fresnel reflection of the both CaF2-uncoated lenses and transmittance from the dichroic mirror (DM on Fig. 1). When the OPA is pumped with the maximum energy of 29 mJ, the idler energy reaches 4.1 mJ, which corresponds to idler conversion efficiency of 13% and quantum conversion efficiency of 51% (Fig. 5). The setup was operated continuously for several months and no damage in any of the optical elements was observed which renders the system as a reliable source for high-energy mid-IR pulses.

Open image in new window

Pulse durations of the incident pump and the second harmonic of the idler pulse from the OPO and OPA are shown in Fig. 6. The estimated pulse duration of the idler pulses after parametric amplification is 600 ps, which indicates the onset of parametric fluorescence; however, the spectral broadening in OPA stage is not substantial (see Fig. 7a). The Idler spectrum is measured with a monochromator and a PbSe detector and the signal spectrum is measured with a commercial NIR spectrometer (Photon Control SPM-002). At 3.1 µm idler canter wavelength and maximum energy output from the OPA, the idler spectral width is ~ 45 nm (FWHM). The bandwidth of the idler spectrum is determined mostly by the spectral bandwidth of the phase matching for collinear interaction as well as by the divergence of the parametric emission which enables noncollinear phase matching, generating spectral components different from those of collinear phase matching [22]. Parametric gain bandwidth can be evaluated simply in the case of plane-wave collinear interaction [23] for large gain, i.e., for parametric gain coefficient Г, Γ >> 1/L and Γ >> Δk/2 gain-half-maximum is approximately given by \(\Delta {k_{1/2}}\):

Open image in new window

Open image in new window

$$\Delta {k_{1/2}}=2\sqrt {(\Gamma \ln 2)/L}$$

$$\Delta k=\left| {{k_{\text{p}}} - {\text{ }}{k_{\text{s}}} - {\text{ }}{k_{\text{i}}}} \right|,~\Gamma ={\left( {\frac{{2{\omega _{\text{s}}}{\omega _{\text{i}}}{d_{{\text{eff}}}}}}{{{n_{\text{i}}}{n_{\text{s}}}{n_{\text{p}}}{c^3}{e_0}}}~{I_{\text{p}}}} \right)^{1/2}}$$

L is the nonlinear crystal length, n_i, n_s and n_p are the refractive indices for idler, signal and the pump waves, respectively; k_p, k_s and k_i are the pump, signal and idler wave vectors, respectively, L is the nonlinear crystal length, I_p is the input pump intensity.

In the presented frequency conversion setup, the input pump intensity used for the OPA is higher than the threshold for parametric fluorescence (single-pass optical parametric generation) and much higher than the one used for OPO pumping. The higher pump intensity determines the broader idler OPA spectrum (at least ~ 1.4 times) compared to the OPO spectrum despite its longer crystal length.

By changing the temperature of the PPSLT crystal from room temperature up to 265 °C, we are able to achieve continuous tuning from 3.5 to 3 µm employing the three domain inversion periods (Fig. 7b). The experimental results of OPA wavelengths(solid curves) are in very good agreement with the theoretically calculated curves (solid lines) for the domain inversion periods (Fig. 7b). In the above calculations, we used the Sellmayer coefficients and the expression for thermal expansion of SLT derived by Dolev et al. [24]. In the used temperature range, the chosen PPSLT design provides the wavelength being detuned along the right slope of the water absorption curve at 3 microns, whereby the absorption coefficient changes with few orders of magnitude.

We present table-top 1-µm-master oscillator power amplifier system with further frequency conversion to mid-IR. The mid-IR picosecond output (~ 600 ps) is characterized with high energy − 4.1 mJ at 0.5 kHz which results in a high-average power as well. The generated wavelength is tuned near the water absorption peak at 3 µm which is of fundamental interest for medical applications and specifically for minimally invasive surgery. Although the two-stage frequency conversion is more complex than a single OPO, the utilized approach offers more reliability and further scaling of the energy and average power. The obtained results show that the presented laser system is a viable alternative of the mid-IR FELs for potential medical and material science applications as a table-top mid-IR laser.