David M. Rines, Glen A. Rines and Peter F. Moulton
Schwartz Electro-Optics, Inc., Research Division
Abstract
We report greater than 15 mJ/pulse at 10 Hz and tuning over the range 3.58 -4.18 um from a CdSe OPO pumped by an amplified 2.79-um Cr,Er:YSGG laser.
Summary
Many tunable mid-infrared sources are currently under development for a wide variety of applications. Using an appropriate pump laser and nonlinear material to construct an optical parametric oscillator is the most efficient means of obtaining both high energy and tunablility in the mid-infrared. Our primary goal for the work reported here was to develop an efficient, high-energy, 10-Hz, 4-um source. That involved engineering a high-energy, 10-Hz, Cr,Er:YSGG pump laser as well as a CdSe OPO.
CdSe is an attractive nonlinear material that can operate as an OPO in the 3.58-4.2 um (signal) and the 8.3-12.6 um (idler) regions. When pumped near 3 um, CdSe has an effective nonlinear coefficient, deff, of 17-18 pm/V, and operates in a Type II phase-matched orientation with the angles ranging from 67 to 90 degrees. Energy conservation dictates that the efficiency of converting pump energy to signal energy increases as the pump wavelength approaches the signal wavelength. Therefore pumping near 3 um can potentially yield high overall efficiency.
Historically the absorption losses in CdSe near 3 um have been in the range 0.05-0.07 cm-1 [1]. Though crystals as long as 6 cm can be grown routinely, the losses make high-energy operation difficult. With high absorption the susceptibility to damage and the inefficiency of the system become prohibitive. As part of our effort we supported development of high-quality CdSe at Cleveland Crystals, Inc. By modifying the processes used to obtain single-crystal CdSe they were able to reduce the absorption by nearly an order of magnitude. The absorption coefficient of the crystal used in our experiments was nominally constant at 0.007-0.01 cm-1 over the 1-10 um region [1].
The Cr,Er:YSGG laser we used as a pump source can be operated at 2.79 um with reasonable efficiency under flashlamp pumping [2]. Low oscillator thresholds allow operation at high (10 Hz) pulse repetition frequencies while minimizing the heat load in the rod. However the excess heat load associated with chromium sensitization combined with the relatively poor thermo-optic properties of YSGG result in severe thermal lensing and thermally-induced birefringence.
Given the above considerations we performed thermal lensing measurements on a 5x75mm Cr,Er:YSGG rod. The data we obtained indicated a short focal-length (10-15 cm at 500 W), astigmatic thermal lens. We used this data to design the folded resonator shown schematically in Figure 1. The angle (49 degrees), radius of curvature (75 cm), and placement (5 cm from the rod end) of the convex, spherical mirrors, M2 and M3, provide the proper amount of astigmatic, negative lensing to compensate the positive astigmatic lensing induced in the laser rod by 300W (30 Joules, 10 Hz) lamp pulses. They are coated to be highly reflective at 45 degrees for the p-polarization at 2.8 um. Using a model for Gaussian beam propagation, we traced a beam from the center of the rod off the convex mirror and into space calculating the curvature of the wavefront at various positions. We then chose the positions of the high reflector and the output coupler such that their radii (10 meter concave and flat, respectively) matched those of the wavefronts of the theoretical fundamental mode to form the desired resonator. The Q-switch is LiNbO3 cut at Brewster's angle to eliminate the need for anti-reflective (AR) coatings and to polarize the resonating beam.
Figure 1. Schematic layout of CdSe OPO system.
This resonator reliably produced 26-mJ, linearly polarized, near-diffraction-limited, 35-ns pulses at 10 Hz. The output beam was elliptical with 1/e2 diameters of 2.3 and 1.9 mm at the output coupler, and the M2 values were measured to be M2x = 2.8 and M2y = 1.4. Higher pulse energies were not possible for extended periods of time without sustaining optical damage to the Q-switch and/or any of the resonator mirrors.
We constructed a Type II critically phasematched CdSe OPO employing the laser described above as the pump. The CdSe crystal had low-loss (alpha= 0.007 cm-1) at the pump wavelength and both optical faces were AR-coated for the pump and signal wavelengths. It had the dimensions 10x11x51mm and was cut for nominal 73 degrees phase-matching. The mirror though which the pump beam entered the OPO was coated for maximum transmission of the pump and minimum transmission of the signal. The OPO output coupler was placed 6 cm from the input mirror and was coated for maximum reflection of the pump and nominally 65% transmission of the signal. These coating designs allow high transmission of pump radiation into the OPO cavity and retro-reflection of the pump for a second pass through the crystal, thereby doubling the effective interaction length. No particular reflectivities were specified for the idler wave. Both mirrors were nominally 80% transmissive in the 8-12 um range. Pumping with the maximum available energy from the Er oscillator we obtained signal pulse energies of 4 mJ at 4.13 um with no damage to any optics. We measured the pump, signal and idler energies individually by passing the output beam through a dispersing prism. The signal wave was tuned from 3.585 um to 4.135 um. The data was corrected for the losses in the prism and is represented by the diamonds in Figure 2, below.
To increase the available pump energy we constructed a 5x75mm Cr,Er:YSGG amplifier. The layout is shown in Figure 1 with the oscillator and the OPO. Since the gain cross-section of this Er transition is low, it was necessary to pump with 50-60 Joules (i.e. 500-600W at 10 Hz) and double-pass the amplifier to realize significant gain. As mentioned earlier the thermal aberrations induced in Cr,Er:YSGG rods are severe at these lamp powers. In this case we compensated the thermal lens by employing a 12-cm convex mirror to reflect the beam at a slight angle for the second pass through the amplifier. We did not compensate the depolarization caused by thermal birefringence, but we placed a silicon plate in the amplified beam at Brewster's angle to reflect the majority of unwanted s-polarized light. Of the 83 mJ of output from the amplifier 63 mJ was transmitted through the silicon plate and used to pump the CdSe OPO. This yielded more than 15 mJ (150 mW) of 3.87-um output from the OPO. A small amount of idler energy (~ 0.5 mJ) was measured, but the exact wavelength was not determined. Since both OPO mirrors were highly-transmissive at the idler wavelengths, a direct correlation of signal and idler energy was not possible. The tuning data for the signal wave is represented by the circles in Figure 2. The temporal pulse width of the OPO was measured to be 23 ns (FWHM) and the spectral width was nominally 5 nm (FWHM). The slope efficiency of the pump energy, incident on the crystal face, to signal energy was 28%.
Figure 2. Tuning data for the signal wave of the CdSe OPO for the two cases discussed: osc only (diamonds) and osc/amp (circles). The output coupler transmission is plotted for reference.
In summary, we have demonstrated greater than 15 mJ of 3.87-um signal output at 10 Hz from a CdSe OPO. We obtained tuning of the signal wave over the range 3.58-4.18 um. We have also demonstrated a 26-mJ, 10-Hz, near-diffraction-limited, flashlamp-pumped Cr,Er:YSGG oscillator and greater than 80 mJ at 10 Hz from a 2.79-um, Cr,Er:YSGG oscillator/amplifier.
References
1. Data from Gary Catella and Michael R. Panfil, Cleveland Crystals, Inc., Cleveland OH.
2. P.F. Moulton, J.G. Manni and G.A. Rines, "Spectroscopic and Laser Characteristics of Er,Cr:YSGG," IEEE J. Quantum Electron. 24, 960 (1988).
If you got here from a search engine, click the icon to go
to Recent Technical Papers at Q-Peak