February 15, 2008 / Vol. 33, No. 4 / OPTICS LETTERS
345
Tunable, high-repetition-rate, femtosecond pulse generation in the ultraviolet M. Ghotbi,1 A. Esteban-Martin,1 and M. Ebrahim-Zadeh1,2,* 1
ICFO-Institut de Ciencies Fotoniques, Mediterranean Technology Park, 08860 Castelldefels, Barcelona, Spain 2 Institucio Catalana de Recerca i Estudis Avancats, Passeig Lluis Companys 23, Barcelona 08010, Spain *Corresponding author:
[email protected]
Received November 30, 2007; revised December 19, 2007; accepted December 23, 2007; posted January 17, 2008 (Doc. ID 90367); published February 11, 2008 We report efficient generation of tunable femtosecond pulses in the ultraviolet (UV) by intracavity doubling of a visible femtosecond optical parametric oscillator (OPO). The OPO, based on a 400 m BiB3O6 crystal and pumped at 415 nm in the blue, can provide visible femtosecond signal pulses across 500– 710 nm. Using a 500 m crystal of -BaB2O4 internal to the OPO cavity, efficient frequency doubling of the signal pulses into the UV is achieved, providing tunable femtosecond pulses across 250– 355 nm with up to 225 mW of average power at 76 MHz. Cross-correlation measurements result in UV pulses with durations down to 132 fs for 180 fs blue pump pulses. © 2008 Optical Society of America OCIS codes: 190.4970, 190.7220, 190.7710, 190.4400, 140.3610.
Tunable, high-repetition-rate, femtosecond pulses in the UV are of interest for a variety of applications in time-domain and photoelectron spectroscopy, biophotonics, waveguide fabrication, and nanotechnology. The most direct route to the generation of such pulses is external frequency doubling, tripling, or quadrupling of the Kerr-lens mode-locked (KLM) Ti: sapphire laser, which can in principle allow access to spectral regions from below ⬃200 to ⬃350 nm. While external frequency doubling can provide conversion efficiencies in excess of 50% over 350– 500 nm [1,2], the generation of practical average powers deeper in the UV using single-pass frequency tripling or quadrupling can be limited by low conversion efficiencies [3]. The attainment of tunable coverage in the UV also necessitates wavelength tuning of the KLM Ti: sapphire laser and, furthermore, requires simultaneous adjustment of three tuning parameters, the fundamental wavelength and phase-matching conditions in the second and third (or fourth) harmonic processes. In this Letter, we demonstrate an alternative approach to the generation of tunable high-repetitionrate femtosecond pulses in the UV, which circumvents the limitations of single-pass tripling of the Ti: sapphire laser. The technique, based on intracavity doubling of a visible femtosecond optical parametric oscillator (OPO) [4], provides practical average powers and wide tuning in the UV. Moreover, tuning is achieved at a fixed Ti: sapphire wavelength by adjustment of only two parameters, the phase-match angle in the OPO and the doubling crystals. The approach offers major advantages over, for example, external enhancement doubling of the OPO output [5]. These include the use of only one synchronous cavity and no need for active length control or input coupling optimization of the enhancement cavity, resulting in reduced complexity. Our approach also provides 83% efficiency from usable OPO output to the UV compared with 53% in external enhancement doubling [5]. 0146-9592/08/040345-3/$15.00
Intracavity frequency doubling of near-IR femtosecond OPOs has been previously used for access to visible wavelengths [6]. The approach enabled the generation of femtosecond pulses over 580– 657 nm at up to 240 mW average power. More recently, we reported femtosecond pulse generation across the entire 480– 710 nm range in the visible at up to 270 mW average powers by using a synchronously pumped OPO based on BiB3O6 (BIBO) pumped in the blue [4]. Here, we demonstrate intracavity frequency doubling of this OPO to generate femtosecond pulses across 250– 355 nm in the UV at up to 225 mW average power. The blue-pumped BIBO OPO is a uniquely versatile source of femtosecond pulses with wide and continuous tunability across the visible using collinear phase matching. Similar attempts based on BBO 共-BaB2O4兲 using noncollinear pumping have resulted in substantially lower output power and limited tuning range 共76 nm兲 in the visible, as well as beam pointing variation across the tuning range [7]. These, together with the lower nonlinearity of BBO, render this approach inefficient, offering limited output power and wavelength coverage in the UV. The configuration of the intracavity-doubled BIBO femtosecond OPO is shown in Fig. 1. The OPO is synchronously pumped by the second harmonic of a KLM Ti: sapphire laser at 415 nm. The laser provides 2.1 W of average power at 830 nm in ⬃150 fs pulses at 76 MHz. Single-pass doubling in a 1 mm long BIBO crystal 共 = 152° , = 90° 兲 with type I 共e + e → o兲 phase matching provides 1.15 W of blue average power at ⬃55% efficiency [1,2]. The pump pulses at 415 nm have durations of ⬃180 fs. The OPO cavity is a bifocal ring, comprising four concave 共r = 100 mm兲 and two plane mirrors. The concave mirrors M1 and M2 provide the focus for the OPO crystal, whereas M5 and M6 allow focusing into the doubling crystal. The angle of incidence on the curved mirrors is kept ⬍7.5° to minimize astigmatism. All mirrors are ⬎99% reflecting over 500– 700 nm, while M1 and M2 © 2008 Optical Society of America
346
OPTICS LETTERS / Vol. 33, No. 4 / February 15, 2008
Fig. 1. (Color online) Configuration of the intracavity frequency-doubled visible BIBO femtosecond OPO synchronously pumped by the second harmonic of KLM Ti: sapphire laser in the blue. L1 and L3, focusing lenses; L2 and L4, collimating lenses.
are also ⬎90% transmitting over 380– 450 nm. The ring cavity allows the generation of UV output in one direction through M6. For maximum UV extraction, M6 also has high transmission 共⬃70% – 90% 兲 over 250– 350 nm. The pump beam is focused to a beam waist w0 ⬃ 25 m inside the OPO crystal. To maximize input pump power to the OPO, the collimating and focusing lenses, L1 and L2, are antireflection (AR) coated 共R ⬍ 1 % 兲 in the blue. The nonlinear crystal for the OPO is BIBO, cut for type I 共o → e + e兲 interaction at = 159° 共 = 90° 兲. The crystal end faces are AR coated 共R ⬍ 0.5% 兲 over 500– 700 nm and have ⬎95% transmission over 375– 435 nm. From considerations of group-velocity mismatch [1], we used a crystal length of 400 m. For frequency doubling, we employed BBO, because of its deeper UV transparency 共⬃180 nm兲 compared with BIBO 共⬃280 nm兲 and higher nonlinear efficiency for UV generation [8]. The BBO crystal was 500 m thick, cut at = 42° for type I 共o + o → e兲 interaction, providing an effective nonlinear coefficient, deff ⬃ 1.4– 1.8 pm/ V, across the fundamental signal tuning range. The crystal faces were AR coated over 500– 700 nm 共R ⬍ 1 % 兲 and over 250– 350 nm 共R ⬍ 8 % 兲. Wavelength tuning in the UV was achieved by continuous tuning of the OPO signal across the visible through angular rotation of the BIBO crystal and simultaneous angular tuning of the BBO secondharmonic crystal. At each wavelength, the cavity was optimized to compensate for any lateral shifts due to angular tuning or changes in synchronous length, to optimize the UV output. The OPO signal could be tuned over 500– 710 nm by changing the internal BIBO crystal angle from = 171.5° to 154.5°, with the corresponding UV wavelength range of 250– 355 nm generated for a change in the internal angle of the BBO crystal from = 52.3° to 33.1°. The limit to the obtained UV tuning range was set by the overall reflectivity of the cavity mirrors at the signal wave-
length. By using mirrors with broader reflectivity and shorter pump wavelengths near 400 nm, full coverage across 230– 360 nm should be readily attainable. Figure 2 shows recorded spectra of the visible signal and the UV second harmonic. As is evident in Fig. 2(a), the signal spectral widths vary from ⬃3 to ⬃3.5 nm. At shorter signal wavelengths in the green, however, the bandwidth is increased significantly to ⬃8 nm, which we attribute to the net OPO cavity dispersion conditions in this range in the absence of intracavity dispersion compensation. Combined with self-phase-modulation, this leads to spectral broadening and chirping of signal pulses in the green. Nevertheless, the generated UV spectra exhibit consistent behavior, with bandwidths ranging from ⬃0.5 nm to ⬃1 nm. The bandwidth reduction from the visible to the UV is attributed to the spectral acceptance in BBO, which results in gain narrowing, thus constraining the second-harmonic conversion bandwidth for fundamental pulses. The calculated secondharmonic spectral acceptance in the 500 m BBO crystal varies from ⬃1 to ⬃5 nm over the fundamental range of 500– 700 nm, implying stronger spectral narrowing at shorter signal wavelengths. Accordingly, one would expect narrower UV spectra toward the shorter wavelengths. This is qualitatively supported by the recorded spectra in Fig. 2(b). Further spectral narrowing from the visible to the UV is also
Fig. 2. Typical spectra of (a) the visible signal across the OPO tuning range and (b) the corresponding generated second harmonic in the UV.
February 15, 2008 / Vol. 33, No. 4 / OPTICS LETTERS
to be expected from the nonlinearity of gain in the second-harmonic process. We obtained UV average powers of ⬎175 mW over ⬃70% of the tuning range 共275– 350 nm兲 and ⬎100 mW over ⬃80% of the tuning range 共270– 355 nm兲. In the wavelength range of 255– 270 nm, practical powers of 25– 100 mW were still generated, with 5 mW available at 250 nm. We believe the decline in the generated UV power below 270 nm is due mainly to the spectral broadening and chirping of visible signal pulses in the green as described above. Together with reduced spectral acceptance in BBO, this results in lower UV power toward shorter wavelengths. Figure 3 shows the generated power and conversion efficiency at 323 nm, where the highest UV power was obtained, versus pump power at 415 nm. The UV output can be seen to increase almost linearly, reaching 225 mW at the maximum blue power of 1.15 W, representing a conversion efficiency of 19.7%. The UV power of 225 mW is close to the maximum visible signal power of 270 mW extracted directly from the OPO previously [4], implying that the intracavity doubling acts as an almost optimized nonlinear coupler for the signal. Given the linear rise in power and no evidence of saturation, we expect increased UV output powers in excess of 225 mW with higher pump powers. The short lengths of BIBO and BBO crystals resulted in minimum beam distortion arising from spatial walkoff, tight focusing, or cavity astigmatism, so that the UV output beam had close to a circular profile with a M2 ⬍ 1.1. The blue pump power threshold for the OPO was 150 mW, equivalent to a fundamental Ti: sapphire laser power of 600 mW. Temporal measurements of the UV output were performed by using the cross-correlation technique. The UV pulses were mixed with 150 fs Ti: sapphire fundamental pulses at 830 nm in a 500 m thick BBO crystal cut at = 26° for type I 共o + o → e兲 phase matching, and background-free intensity profiles were obtained by using a GaAsP detector. A typical cross-correlation profile and spectrum at 323 nm are shown Fig. 4, confirming a near-transform-limited pulse 共⌬⌬ ⬃ 0.34兲, assuming a sech2 pulse shape. Across the UV tuning range, pulse durations of
Fig. 3. (Color online) Variation of the generated UV average power and conversion efficiency at 323 nm with the blue pump power at 415 nm.
347
Fig. 4. Cross-correlation trace, and (inset) spectrum of the generated UV pulses at 323 nm. The time duration of ⌬ ⬃ 132 fs and the spectral bandwidth of ⌬ ⬃ 0.9 nm result in near-transform-limited pulses with a time–bandwidth product of ⌬⌬ ⬃ 0.34.
132– 250 fs were measured, with time–bandwidth products varying from ⌬⌬ ⬃ 0.34 to ⬃0.6. In conclusion, we have demonstrated a new source of femtosecond pulses with wide tunability and practical powers in the UV. Based on intracavity doubling of a visible femtosecond OPO, using collinear phase matching and angle tuning at room temperature, a single set of optics, and only two tuning parameters, we have generated femtosecond pulses across 250– 355 nm at 225 mW of average power. The wavelength coverage can be further extended down to 230 nm by improving the mirror coatings and by using shorter pump wavelengths near 400 nm. With intracavity dispersion compensation, we also expect further enhancements in the UV power at shorter wavelengths and transform-limited pulses throughout the tuning range. The wide spectral coverage, practical powers, near-transform-limited pulses, and room-temperature operation will make this device an attractive source of femtosecond UV pulses for a variety of applications. This work was supported, in part, by the Ministry of Education and Science of Spain through the Consolider project SAUUL under contract CSD2007–013. References 1. M. Ghotbi and M. Ebrahim-Zadeh, Opt. Express 12, 6002 (2004). 2. M. Ghotbi, M. Ebrahim-Zadeh, A. Majchrowski, E. Michalski, and I. V. Kityk, Opt. Lett. 29, 2530 (2004). 3. F. Rotermund and V. Petrov, Opt. Lett. 23, 1040 (1998). 4. M. Ghotbi, A. Esteban-Martin, and M. Ebrahim-Zadeh, Opt. Lett. 31, 3128 (2006). 5. V. P. Yanovsky and F. W. Wise, Opt. Lett. 19, 1952 (1994). 6. R. J. Ellingson and C. L. Tang, Opt. Lett. 18, 438 (1993). 7. G. M. Gale, M. Cavallari, T. J. Driscoll, and F. Hache, Opt. Lett. 20, 1562 (1995). 8. M. Ghotbi and M. Ebrahim-Zadeh, Opt. Lett. 30, 3395 (2005).
