2460
OPTICS LETTERS / Vol. 30, No. 18 / September 15, 2005
LiInSe2 nanosecond optical parametric oscillator Jean-Jacques Zondy Institut National de Métrologie, CNAM, 292 Rue Saint-Martin, F-75003 Paris, France
Vitaliy Vedenyapin, Alexander Yelisseyev, Sergei Lobanov, and Ludmila Isaenko Branch of the Institute of Mineralogy and Petrography, SB RAS, 43 Russkaya Street, 630058 Novosibirsk, Russia
Valentin Petrov Max-Born-Institute for Nonlinear Optics and Ultrafast Spectroscopy, 2A Max-Born-Strasse, D-12489 Berlin, Germany Received March 15, 2005; revised manuscript received April 29, 2005; accepted May 18, 2005 Optical parametric oscillation using the new lithium selenoindate nonlinear crystal is reported for what is to our knowledge the first time. A 17 mm long, type II phase-matched sample is pumped by a 10 ns Nd:YAG laser. The minimum pump energy threshold is 3 mJ for a signal-resonant configuration. The signal and idler waves are tunable between 1.47 and 1.57 m, and 3.3 and 3.78 m, with a total output energy of 170 J corresponding to a 2.4% energy conversion at 8 mJ pump, only limited by the AR coatings damage. With optimized crystal quality and coatings, lithium selenoindate should show superior performance as compared with AgGaS共e兲2 crystals, owing to its 4⫻ larger thermal conductivity. © 2005 Optical Society of America OCIS codes: 190.4970, 190.2620, 190.4400.
For a long time the only available nonlinear crystal capable of downconverting 1 m radiation from highpower Nd lasers to the deep mid-IR has been the chalcopyrite AgGaS2 (AGS). Earlier optical parametric oscillators (OPOs) using compounds such as proustite1 共Ag3AsS3兲 were abandoned because of growth difficulties. More recently, a 1064 nm pumped HgGa2S4 OPO was reported,2 but this material is currently available only as small-size single crystals. The other chalcopyrites with higher nonlinearities, such as AgGaSe2 (AGSe) or ZnGeP2 (ZGP), are not phase matchable for pump wavelengths near 1 m or exhibit severe two-photon absorption due to their lower energy bandgap. After the first demonstration of an AGS OPO,3 little progress followed, especially in tunability,4 until recently when Vodopyanov et al.5 demonstrated tuning from 3.9 to 11.3 m with a Nd:YAG-pumped OPO. AGS is also the only mid-IR material with which a true cw OPO could be demonstrated, owing to its low 共␣ 艋 0.01 cm−1兲 absorption loss.6 Since two-photon absorption is not detrimental in the cw regime, pumping at yet shorter wavelengths (845 nm) was possible, although output mid-IR power was clamped to 1–2 mW by dramatic thermal lensing effects originating from the poor thermal conductivity K ⬃ 1.5 W / m / K.7 In this Letter we introduce what is to our knowledge the first OPO based on LiInSe2 (LISe)8 pumped by a nanosecond Nd:YAG laser at repetition rates ranging from 10 to 100 Hz. Like AGS(e), LISe belongs to the AIBIIIC2VI ternary chalcogenide semiconductor group: LISe is hence the third compound of this group for which parametric oscillation has been demonstrated. Unlike the Ag compounds that crystallize ¯ 2m point group), the in the chalcopyrite structure (4 8 9 biaxial LISe and LIS crystallize with the  -NaFeO2 structure (orthorhombic mm2 symmetry group). Our interest in LISe is motivated by its supe0146-9592/05/182460-3/$15.00
rior thermomechanical properties10 (isotropic expansion, K ⬃ 5 W / m / K, smaller thermo-optic coefficients) in comparison with AGS. Such properties, also common to LIS, would favor cw OPO operation in the deep mid-IR 共⬎5 – 6 m兲 if their residual absorption and scattering loss, presently at the level of ␣ ⯝ 0.04 cm−1, could be reduced to the level of that of AGS. The dark-red annealed LISe sample used in this experiment (dimensions 5 mm⫻ 6 mm⫻ 17 mm) was grown from an oriented seed by the vertical two-zone furnace Bridgman technique. It was cut at = 72° in the XY plane for type II 共eoe兲 downconversion (p = 1064 nm→ s = 1450– 1650 nm; i = 3300– 3800 nm). The effective nonlinear coefficient, deoe = d24 sin2 + d15 cos2 ,8 was reestimated by us to be ⬃8.5 pm/ V from the measured OPO thresholds. The sample was broadband antireflection (AR) coated for the pump, signal, and idler ranges. The spectrum of the ARcoated crystal displays a transmittance of Tp,s,i ⬃ 92%, with a loss 共␣ ⬃ 0.04 cm−1兲 due to absorption
Fig. 1. DRO with double-pass pump (DPP): (a) total signal+ idler energy versus pump energy at f = 30 Hz; (b) total energy versus repetition rate for DPP and single-pass pump (SPP). © 2005 Optical Society of America
September 15, 2005 / Vol. 30, No. 18 / OPTICS LETTERS
Fig. 2. IRO total signal+ idler output with pump recycling. The filter F is a long-wave-pass filter that blocks the pump wave.
and scattering by residual point or extended defects. The pump laser source was a flash-pumped, electro-optically Q-switched Nd:YAG laser (Minilite, Continuum), delivering 10 ns pulses with variable repetition rates (10–100 Hz) and a spectral linewidth of 1 cm−1 共⬃0.1 nm兲. Its output beam had a Gaussian diameter 2wp = 2 mm, resulting in a walkoff 共 = 0.7° 兲 length equal to ⬃4⫻ the physical crystal length. A half-wave plate and a polarizer were used to attenuate the laser and adjust the single pulse pump energy. For a separate AR-coated LISe plate, AR coatings damage was found to occur for Ep ⬎ 8 mJ, i.e., a pump fluence of 0.25 J / cm2, which is far less (10⫻ at least) than the surface–bulk damage threshold. Hence, during the experiments, the pump laser was clamped to 8 mJ (impinging on the crystal) to avoid surface damage. We started first with a true doubly resonant OPO cavity (DRO) with identical CaF2 plane mirrors (M1, M2) with Tp = 0.9, Rs,i 艌 0.995; see the inset of Fig. 1(a). The distance between M1 and M2 was set to Lcav = 25 mm, corresponding to an average number of 30 round trips. The transmitted pump was further recycled back by a Ag-coated mirror via a dichroic beam splitter (Rp = 1, Ts,i 艌 0.9) that transmitted the signal and idler waves leaking through M2. Figure 1(a) shows the net signal+ idler energy leaking from M1 and M2 at a repetition rate f = 30 Hz, as measured by a calibrated pyroelectric detector. The crystal was set at normal incidence, corresponding to a generated signal at s = 1537.1 nm 共i = 3456.9 nm兲, as measured with a multichannel spectrum analyzer. The signal linewidth was narrow, less than the 0.6 nm resolution (5 cm−1 or ⬃150 GHz) of the spectrum analyzer, as expected from the type II 共eoe兲 coupling.5 The measured pump energy threshold was Epth共10 Hz兲 = 3.1 mJ and increased slightly to 3.4 mJ at 30 Hz. At 6 mJ of pump energy incident on the crystal and f = 10 Hz, the total downconverted output power, Es + Ei, was 12 J, corresponding to an energy conversion efficiency of 0.2% for the DRO case. Figure 1(b) displays the output energy as the laser repetition rate is increased. In spite of the nearly constant pump pulse energy over the whole repetition rate range, the OPO output energy experiences first a smooth decrease up to f = 40 Hz and then dramatically drops beyond 50 Hz. This behavior is appar-
2461
ently due to the thermal lensing effects (due to the residual bulk absorption) that increase with the repetition rate. From the leakage transmittance of M2, one can infer that the maximum downconverted circulating energy incident on M2 amounts to ⬃8 mJ 共f = 10 Hz兲. With the backward pump blocked, the double-pass pump enhancement is found to lie between 2 and 6 in the range 10⬍ f ⬍ 40 Hz. The DRO configuration resulted in less than 10% pump energy depletion. To increase the signal outcoupling efficiency, a CaF2 plate (0.5 mm thick) was inserted inside the DRO cavity (Lcav = 35 mm; Fig. 2, inset), yielding a signal outcoupling efficiency of ⬃7% per plate facet (the intracavity loss for the e-polarized idler wave was negligible). The pump was recycled with a dielectric highly reflective mirror at 1064 nm. The two outputs from the CaF2 plate were recombined by the Ag mirror. We refer to this cavity as an idler-resonant OPO (IRO). The idler wave output content was only about 5% of the total energy. Figure 2 displays the maximum downconverted output occurring for a tilted LISe 共 = 68.5° 兲 at several repetition rates. The corresponding signal wavelength (1510 nm) actually lies near the lower edge of the mirror reflectance band. The IRO threshold Epth = 4 mJ 共f = 10 Hz兲 was slightly higher than the DRO one, but the extracted energy was Es + Ei = 170 J (17⫻ the DRO output, i.e., an energy conversion efficiency of 2.4%). Given their quadratic slopes, the data points in Fig. 2 actually lie in the vicinity of the threshold zone (hence the obviously faster decrease with the repetition rate as compared with the DRO case). If the pump energy could be increased without AR coatings damage, a much higher output efficiency would be expected. The mid-IR pulse-to-pulse fluctuations 共⬃ ± 15% 兲 were in conformity with the ⬃ ± 10% pump pulse fluctuations. The maximum pump energy in Figure 2 corresponded to a single-pass pump operation at just above threshold for f = 10 Hz. The angular tuning characteristics of the IRO are displayed in Fig. 3. Tuning ranges were s 苸 关1474.8; 1561.6兴 nm and i 苸 关3339.1; 3819.8兴 nm, limited only by the mirror coatings and crystal aperture. The solid curves in Fig. 3 are calculated from new updated Sellmeier coefficients (not yet published; the relations in Ref. 10 display tens of nanom-
Fig. 3. IRO angular tuning characteristics.
2462
OPTICS LETTERS / Vol. 30, No. 18 / September 15, 2005
eters of discrepancies from the experimental wavelengths). From the slope of the tuning curves, type II phase-matching does not offer a wide mid-IR tunability for the range of plotted in spite of its narrower linewidth characteristics. However at smaller angles (down to ⬃ 36°) LISe is type II phase-matchable up to its IR transmission edge 共i ⬃ 12 m兲. By orienting the intracavity CaF2 plate so as to extract the idler wave, instead of the signal, about Ei = 43 J of idler energy 共i = 3602 nm兲 could be extracted from the OPO. Given the i / s = 2.4 times signal photon energy, the photon conversion efficiency in either configuration is nearly comparable. Finally, we have tried a last genuinely singly resonant OPO (SRO) configuration that allows us to extract more idler energy (Fig. 4, inset). The pump laser was coupled to this SRO cavity via a 45°incidence sapphire dichroic beam splitter (D) with Rp ⯝ 1, Ti ⯝ 1. M1 (plano fused silica with Rs ⬎ 0.995) transmitted 98% at 1064 nm and 90% at the idler range, while the Ag mirror 共Rp,s,i 艌 0.985兲 acted as the rear OPO mirror. This cavity 共Lcav = 23 mm兲 hence recycles back both the pump and the idler waves. Thresholds of 3, 4, and 5 mJ at 10, 30, and 50 Hz, respectively, and a maximum idler output of Ei = 92 J (conversion efficiency ⬃1.15%) were achieved. The threshold increase versus the repetition rate is possibly related to slow thermal-lensing-induced spatial mode-mismatching effects and also to thermal dephasing effects caused by the radial temperature gradient as internal average power is increasing11 (the sample exhibits the same absorption at the three wavelengths). However, no roll off in output energy versus pump pulse energy is observed up to 8 mJ of pump input (0.4 W average power at f = 50 Hz), which supports the assumption of slow heating effects instead of peak power effects. The identical slopes of the curves indicate that output energy scaling is expected at larger pump energies. This final setup was found to be the least sensitive to thermal effects (Fig. 5), since at f = 50 Hz up to 50 J of power are obtained.
Fig. 4. SRO idler output with pump+ idler recycling. D, dichoic beam splitter; crystal at normal incidence.
Fig. 5. SRO idler output versus repetition rate.
In conclusion, parametric oscillation using the new ternary chalcogenide LiInSe2 has been reported for the first time to our knowledge, for various cavity configurations. Because LISe can be pumped by widespread 1 m lasers and can now be grown with long dimensions, this result represents an important step in the development of tunable 共2 – 12 m兲 mid-IR sources. Given its larger thermal conductivity and high bulk damage threshold, with the design of damage-resistant AR coatings and further reduction of its residual absorption, in the future LISe should show improved thermal performance in comparison with Ag chalcopyrites. This work was partly supported by the Russian Foundation for Basic Research (grant 04-02-16334). J.-J. Zondy is also affiliated with NLO-Technologies, Ltd. (
[email protected]). References 1. D. C. Hanna, B. Luther-Davies, and R. C. Smith, Appl. Phys. Lett. 22, 440 (1973). 2. V. V. Badikov, A. K. Don, K. V. Mitin, A. M. Seregin, V. V. Sinaiskii, and N. I. Schebetova, Quantum Electron. 33, 831 (2003) [Kvantovaya Elektron. (Moscow) 33, 831 (2003)]. 3. Y. X. Fan, R. C. Eckardt, R. L. Byer, R. K. Route, and R. S. Feigelson, Appl. Phys. Lett. 45, 313 (1984). 4. P. P. Boon, W. R. Fen, C. T. Chong, and X. B. Xi, J. Appl. Phys. 36, L1661 (1997). 5. K. L. Vodopyanov, J. P. Maffetone, I. Zwieback, and W. Ruderman, Appl. Phys. Lett. 75, 1204 (1999). 6. A. Douillet and J.-J. Zondy, Opt. Lett. 23, 1259 (1998). 7. A. Douillet, J.-J. Zondy, A. Yelisseyev, S. Lobanov, and L. Isaenko, J. Opt. Soc. Am. B 16, 1481 (1999). 8. L. Isaenko, A. Yelisseyev, S. Lobanov, V. Petrov, F. Rotermund, G. Slekys, and J.-J. Zondy, J. Appl. Phys. 91, 9475 (2002). 9. S. Fossier, S. Salaün, J. Mangin, O. Bidault, I. Thénot, J.-J. Zondy, W. Chen, F. Rotermund, V. Petrov, P. Petrov, J. Henningsen, A. Yelisseyev, L. Isaenko, S. Lobanov, O. Balachninaite, G. Slekys, and V. Sirutkaitis, J. Opt. Soc. Am. B 21, 1981 (2004). 10. A. P. Yelisseyev, V. A. Drebushchak, A. S. Titov, L. I. Isaenko, S. I. Lobanov, K. M. Lyapunov, V. A. Gruzdev, S. G. Komarov, V. Petrov, and J.-J. Zondy, J. Appl. Phys. 96, 3659 (2004). 11. C. L. Marquardt, D. G. Cooper, P. A. Budni, M. G. Knights, K. L. Schepler, R. DeDomenico, and G. C. Catella, Appl. Opt. 33, 3192 (1994).
