February 15, 2009 / Vol. 34, No. 4 / OPTICS LETTERS
449
Tunable narrowband optical parametric oscillator using a transversely chirped Bragg grating Björn Jacobsson,1,* Valdas Pasiskevicius,1 Fredrik Laurell,1 Eugeniu Rotari,2 Vadim Smirnov,2 and Leonid Glebov2 1
Laser Physics, KTH–Royal Institute of Technology, 106 91 Stockholm, Sweden 2 Optigrate, 3267 Progress Drive, Orlando, Florida 32826, USA *Corresponding author:
[email protected]
Received November 12, 2008; revised December 16, 2008; accepted December 25, 2008; posted January 13, 2009 (Doc. ID 104051); published February 10, 2009 We demonstrate a novel technique for locking and tuning of a near-degenerate nanosecond optical parametric oscillator (OPO) using a transversely chirped volume Bragg grating. When the grating was translated, the OPO signal wavelength could be tuned from 1011 to 1023 nm 共3.5 THz兲. The OPO was based on a periodically poled KTiOPO4 as a nonlinear crystal and was pumped at 532 nm with up to 1.9 mJ of energy. The generated signal at an energy of 0.37 mJ had a bandwidth of 0.50 nm and a suppression of broadband background of ⬎30 dB. The demonstrated technique is widely applicable for the construction of narrowband tunable sources. © 2009 Optical Society of America OCIS codes: 050.7330, 190.4970.
Optical parametric oscillators (OPOs) find an important application as a spectral extension of conventional lasers. With the usage of quasi-phase-matched (QPM) nonlinear crystals, any wavelength in the transparency range of the crystal can be obtained for a given pump laser (or its harmonics). The QPM crystals provide high nonlinearity without walk-off and form the basis for efficient OPOs that can be built in a stable and simple design as singly resonant OPOs. As a pump source, nanosecond pulsed Nd lasers at ⬃1 m can be used, a mature technology that enables robust systems. The spectral characteristics of the generated OPO signal are determined by the spectral gain, i.e., the phase-matching conditions as well as the cavity feedback mirrors. Near degeneracy, where the signal and idler wavelengths are closely spaced, the OPO gain is broadband, since the difference in group velocity between signal and idler is small (for signal and idler in the same polarization). Generally, without intracavity spectral selection, the OPO signal has a broadband spectrum, which is undesirable for many applications. However, the bandwidth of the signal can be dramatically narrowed if a high resolution spectral filter restricts the oscillating wavelength. Then, the broadband gain is an asset, providing access to a broad tuning range. In previous publications [1–5], it was shown that volume Bragg gratings in photo-thermo-refractive (PTR) glass [6] are perfect elements for spectral selection in OPO cavities to obtain narrowband signal radiation, in particular near degeneracy. The cavity design is then very robust and simple; the reflective Bragg grating just replaces one of the cavity mirrors. Still, for equal reflectivity of a grating and conventional mirror, the efficiency of the OPO remains the same [1,4]. The gratings in PTR glass are ideal intracavity elements with spectral selectivity as narrow as 50 pm, high diffraction efficiency, and low loss. Also, the damage threshold is high, ⬃10 J / cm2 for ns pulses [7], which is needed to handle the high inten0146-9592/09/040449-3/$15.00
sity in an OPO. Volume Bragg gratings are manufactured in PTR glass in a two-stage process [6], first a recording stage with an interference pattern of UV light at 300– 350 nm (absorption band of Ce3+) forming a latent image and second a thermal development stage at temperatures above 500° C to induce a refractive index change of exposed parts. Here we demonstrate a novel technique to obtain a tunable and narrowband OPO in a very simple way, by usage of a transversely chirped Bragg grating. The grating has a fan-shaped structure with a transversely chirped grating period and thus a variation of the reflected wavelength over the grating’s transverse direction. This enables tuning by simple translation of the grating. These gratings can nowadays be written with almost linear chirp and a fairly constant grating peak reflectivity using cylindrical lenses to form the transversely varying exposure interference pattern. Previous demonstrations of laser tuning and locking with gratings of the same shape but in other materials are a distributed-feedback dye laser [8] and a semiconductor laser [9]. In addition, longitudinally chirped gratings in PTR glass have been used for pulse stretching and compression [10]. The OPO consisted of an input coupler, a nonlinear crystal, and the volume Bragg grating as a spectral selector and output coupler; see Fig. 1. The total cavity length was 21 mm. The OPO was pumped at 532 nm by a frequency-doubled, actively Q-switched, flashlamp-pumped Nd:YAG laser (Minilase I, New Wave) that delivered 5 ns pulses at 20 Hz. The pump 共M2 ⬃ 5兲 was focused in the OPO to a beam waist ra-
Fig. 1. OPO setup. © 2009 Optical Society of America
450
OPTICS LETTERS / Vol. 34, No. 4 / February 15, 2009
dius 共e−2兲 of 220 m. The pump energy was controlled by a half-wave–plate polarizer arrangement, and kept below 2 mJ to avoid damage to coatings and the nonlinear crystal. The input coupler was a flat dielectric mirror coated for high reflectivity around 1.06 m and high transmission at 532 nm. The nonlinear crystal was a periodically poled KTiOPO4 (PPKTP) with a period of 9.01 m, a total (poled) length of 12 mm 共10 mm兲, 1 ⫻ 5 mm aperture, and antireflection (AR) coated for all interacting waves. In addition, to avoid a subcavity formation due to the crystal faces, it was rotated by ⬃5°, which results in a slightly noncollinear QPM interaction. Enforced by the cavity mirrors, the signal was collinear to the pump with a slightly noncollinear idler. Still, at our small angles, the efficiency of the OPO is similar to a collinear interaction. The crystal was placed on a temperature-controlled copper block for tuning of the phase matching; 40° C was used for grating 1 and 32° C for grating 2. All interacting waves were linearly polarized in the PPKTP c direction. Two different transversely chirped Bragg gratings were used. Grating 1 had a peak reflectivity of ⬃50% for 997– 1016 nm, a transverse chirp of 1.1 nm/ mm, and a FWHM spectral bandwidth of 0.46 nm at any point of the grating for a beam with ⬃200 m radius 共e−2兲 in the chirp direction. The surfaces were polished parallel to the grating planes and AR coated for all interacting waves (R ⬍ 0.2% at 1020 nm). Grating 2 had ⬃35% reflectivity for 1010– 1023 nm, a chirp of 0.88 nm/ mm, and a bandwidth of 0.31 nm FWHM. To avoid parasitic oscillations (see below), the surfaces were wedged orthogonal to the chirp direction, with an angle of 5° relative to the grating planes, and AR coated for all the interacting waves (R ⬍ 0.2% at 1020 nm). The gratings were mounted onto a translation stage for tuning of the resonated wavelength. Note that the angle between the grating planes due to the chirp is small enough 共⬃0.3 rad兲 so that it has no practical impact on the OPO cavity alignment. The resonating OPO signal was locked to the Bragg grating and could easily be tuned by translation of the grating. The tuning curves are shown in Fig. 2 for both gratings. Having the grating facets parallel to the grating planes, as for grating 1, was
not optimal. Parasitic broadband OPO oscillation was then observed due to the facets despite the AR coating, due to the OPO’s high gain. To amend the problem, grating 2 with wedged surfaces was used instead. This removed the parasitic broadband oscillation, and hereafter grating 2 was used. The spectral purity of the signal and the idler was measured at the maximum pump energy of 1.9 mJ. A typical spectrum is shown in Fig. 3. The locked spectra are very clean with a suppression to broadband background of ⬎30 dB. This strong suppression demonstrates that there is very little grating reflectivity off the Bragg wavelength since it would otherwise cause parasitic oscillations as seen for grating 1. The signal bandwidth was 0.50 nm FWHM at full pump energy (inset of Fig. 3) and 0.37 nm near threshold (resolution ⬍0.05 nm). To visualize the available gain in the PPKTP crystal, a broadband uncoated flat glass substrate was instead used as the output coupler. This OPO had a bandwidth FWHM of 34 nm for the signal (see Fig. 3), demonstrating a bandwidth reduction due to the grating of ⬃70 times, compared to a broadband mirror. In Fig. 3, some additional subpeaks can also been seen, although the energy content in them is very small. The spacing to the first subpeak was 7.4 THz and remained constant when tuning the OPO. The strength was sensitive to the pump energy; at 1.9 mJ of pump the suppression was 19 dB, increasing to ⬎35 dB for ⬍1.3 mJ of pump. The fixed spacing of this peak corresponds fairly well to the strongest Raman peak in KTP at 8.07 THz [11]. Thus, we attribute this subpeak to Raman scattering followed by optical parametric amplification (OPA). In addition to the peaks shown in Fig. 3, weak peaks at the conjugate Stokes/anti-Stokes positions were also present. The second subpeak in Fig. 3 is 26 THz off the main signal/idler peak, which corresponds to the spacing between the signal and the idler waves themselves. When tuning the OPO, the subpeak spacing changed, keeping this relation fixed. We thus at-
Fig. 2. (Color online) Tuning of the OPO signal wavelength with grating position, experimental points, and linear fit.
Fig. 3. (Color online) Bragg OPO logarithmic spectrum for grating 2 and comparison with OPO gain. The inset shows the signal spectrum in linear scale.
February 15, 2009 / Vol. 34, No. 4 / OPTICS LETTERS
tribute the second peak to four-wave mixing (FWM) between the signal and the idler, subsequently enhanced by OPA pumped at 532 nm. Close to degeneracy, the FWM is partly phase matched by the QPM structure due to cascaded 共2兲 : 共2兲 interactions, previously shown in [12]. The FWM is determined for the short wavelength peak at frequency s⬘ by the equation s + s = i + s⬘. First the signal frequency s is doubled to generate 2s. This radiation acts as a new pump wave for OPA of the idler i, where the new frequency s⬘ is generated as the conjugate wave. The cascaded 共2兲 : 共2兲 interaction was corroborated experimentally by the presence of second-harmonic waves of the signal and the idler. Close to degeneracy, in addition to better FWM phase matching, more gain is also available for the OPA process pumped at 532 nm, which should result in a stronger FWM signal. In the experiments, this was also seen; when tuning the OPO, the FWM peak was strongest for the closely spaced signal and idler. The FWM peak strength was also sensitive to the pump energy. The suppression was 18 dB at 1.9 mJ of pump, 25 dB at 1.3 mJ of pump, and 32 dB at 0.85 mJ of pump. The energy dependence of the system is shown in Fig. 4, measured at a signal wavelength of 1017.6 nm. For the maximum pump energy of 1.9 mJ (after the input coupler), the total output energy was 0.70 mJ. The Manley–Rowe relations yield a distribution of 0.37 mJ of signal and 0.33 mJ of idler energy. When tuning the OPO, the maximum signal energy remained fairly constant above 0.35 mJ from 1011– 1023 nm. The depletion of the pump reached 65%. It was measured by comparing the transmitted pump under OPO action with the transmitted pump below OPO threshold and scaling appropriately. The maximum efficiency of the OPO was ⬃36%, given by the conversion from incident pump to signal and idler. The discrepancy between depletion and efficiency, common to this type of OPO [1,4], indicates some unaccounted for loss mechanisms. Previously, we have demonstrated that the loss is not due to the volume Bragg grating [1,4]. Instead, we potentially attribute it to resonator loss of the OPO signal.
Fig. 4. (Color online) Energy properties of the OPO with grating 2 at 1017.6 nm.
451
By comparing the tuning curves of Fig. 2 to a linear fit, the linearity of the grating chirp can be evaluated. For grating 2 from 1011 to 1023 nm, the maximum local deviation from a linear behavior is ±0.01%, relative to the grating’s period, with an overall parabolic shape. It should be noted that our system provides two output beams, the signal and the idler, where the distance between the two can be freely tuned. This could be used as a pump source for a tunable differencefrequency generation (DFG) process. Here, the difference is ⬃30 THz and tunable, showing potential for a tunable terahertz-wave source, by placing a suitable nonlinear medium for DFG in the output beam. In conclusion, we have demonstrated a neardegenerate OPO, where the wavelength is locked and narrowed by a fan-shaped volume Bragg grating with a transversely chirped period. By translation of the grating, the signal wavelength could be tuned from 1011 to 1023 nm 共3.5 THz兲 with a signal bandwidth of 0.50 nm 共140 GHz兲, a suppression of broadband background of ⬎30 dB, and a signal energy of 0.37 mJ. The reduction in bandwidth due to the grating is almost 2 orders of magnitude compared to a conventional broadband mirror. Since the tuning is done by simple translation of the Bragg grating, the OPO can be made very compact with a minimum of components. The demonstrated method is widely applicable to any wavelength region within the components’ transparency ranges. We acknowledge Carlota Canalias for manufacturing the PPKTP crystal. Partial financial support was received from Carl Trygger’s foundation, Göran Gustafsson’s foundation, and the Swedish Research Council. References 1. B. Jacobsson, M. Tiihonen, V. Pasiskevicius, and F. Laurell, Opt. Lett. 30, 2281 (2005). 2. M. Henriksson, M. Tiihonen, V. Pasiskevicius, and F. Laurell, Opt. Lett. 31, 1878 (2006). 3. J. Saikawa, M. Fujii, H. Ishizuki, and T. Taira, Opt. Lett. 32, 2996 (2007). 4. B. Jacobsson, C. Canalias, V. Pasiskevicius, and F. Laurell, Opt. Lett. 32, 3278 (2007). 5. P. Blau, S. Pearl, S. Fastig, and R. Lavi, IEEE J. Quantum Electron. 44, 867 (2008). 6. O. Efimov, L. Glebov, L. Glebova, K. Richardson, and V. Smirnov, Appl. Opt. 38, 619 (1999). 7. O. M. Efimov, L. B. Glebov, S. Papernov, and A. W. Schmid, Proc. SPIE 3578, 564 (1999). 8. A. Matsuda and S. Iizima, Appl. Phys. Lett. 31, 104 (1977). 9. E. L. Portnoi, Czech. J. Phys. Sect. B 34, 469 (1984). 10. K. Liao, M. Cheng, E. Flecher, V. I. Smirnov, L. B. Glebov, and A. Galvanauskas, Opt. Express 15, 4876 (2007). 11. V. Pasiskevicius, A. Fragemann, F. Laurell, R. Butkus, V. Smilgevicius, and A. Piskarskas, Appl. Phys. Lett. 82, 325 (2003). 12. A. Varanavicius, A. Dubietis, A. Berzanskis, R. Danielius, and A. Piskarskas, Opt. Lett. 22, 1603 (1997).
