1 kW peak power passively Q-switched Nd3�-dopedglass integrated waveguide laser

B. Charlet,* L. Bastard, and J. E. BroquinInstitut de Microélectronique, Electromagnétisme et Photonique—Laboratoire d'Hyperfréquences et Caractérisation,

3 Parvis Louis Néel, BP 257, 38016 Grenoble Cedex 1, France*Corresponding author: charletb@minatec.grenoble‑inp.fr

Received March 24, 2011; revised April 21, 2011; accepted April 21, 2011;posted April 22, 2011 (Doc. ID 144621); published May 23, 2011

Embedded optical sensors always require more compact, stable, and powerful laser sources. In this Letter, wepresent a fully integrated passively Q-switched laser, which has been realized by a Agþ=Naþ ion exchange on aNd3þ-doped phosphate glass. A BDN-doped cellulose acetate thick film is deposited on the waveguide, actingas an upper cladding and providing a distributed saturable absorption. At λ ¼ 1054nm, the device emits pulsesof 1:3 ns FWHM with a repetition rate of 28kHz. These performances, coupled with the 1 kW peak power, arepromising for applications such as supercontinuum generation. © 2011 Optical Society of AmericaOCIS codes: 140.3540, 140.3530, 130.2755, 230.3120.

Because of their compactness and simplicity, miniatur-ized Q-switched lasers are interesting for use in em-bedded optical sensors, where low power consumptionand robustness to environmental changes are highly de-sirable. Waveguide-based lasers present, on one hand,excellent stability of the output beam and, on the otherhand, an excellent optical energy confinement, which iscompatible with a low-power laser diode pumping.Therefore, using a waveguide approach to realize minia-turized Q-switched lasers is of high interest. Waveguidelasers can be realized using either fiber optic [1,2] orintegrated optic [3] technologies. Nonetheless, pulsedoptical fiber lasers require an external element such asa semiconductor saturable absorber mirror (SESAM)[1] or an accousto-optic modulator [2]. Because these sa-turable absorbers cannot be monolithically integratedwithin the optical fiber, fiber lasers must contain bulkelements such as lenses or circulators, yielding to a lossof compactness and stability. This problem can be solvedby using integrated optics, which adds to the advantagesof fiber optics (single mode operation, strong confine-ment) those of planar integration (compacity, reliability,self alignment, function integration). Among the differenttechnologies available for fabricating integrated optic de-vices, ion exchange on glass is a well mastered one.Indeed, it has been used for more than 20 years to realizenumerous passive devices with applications ranging fromoptical telecommunication to sensors and astronomy [4].Moreover, the use of rare-earth-doped glasses has al-lowed realization of efficient active devices such as am-plifiers [5] and continuous-wave [6–8] and mode-locked[9] lasers. However, there are only a few studies dealingwith integrated optic Q-switched lasers: a first realizationwas reported in 1991 using a thermo-optic modulator anda quite complex cavity [10]. One of the first passively Q-switched lasers based on ion-exchanged waveguides ac-tually used this method,only as a gain medium inside aclassical Fabry–Perot cavity [11]. In 2008, a fully inte-grated passively Q-switched waveguide laser wasreported by Salas-Montiel et al. using a saturable absor-ber hybridized on the surface of the ion-exchangedwaveguide [12]. Based on a Nd3þ-doped phosphate glasscavity closed by two dielectric mirrors, it emitted pulses

of 10 ns with a peak power of 1W at a rate of 350 kHz.This low peak power was mainly due to propagationlosses greater than 1 dB=cm and a nonoptimized satur-able absorber. Moreover, the presence of a mirror onthe output facet prevented a direct coupling of the laserbeam into an optical fiber. In this Letter, we show that asimilar integrated structure can be used to achieve a highpeak power when the technological process and the sa-turable absorber are optimized. Moreover, we demon-strate the possibility of efficiently coupling the laseroutput in an optical fiber without bulk optics. The firstpart of this Letter deals with the realization process ofthe hybrid passively Q-switched laser. The second partis devoted to the presentation of the passive and activecharacteristics of the realized device, which are then in-terpreted. Finally, the perspectives opened by these re-sults are presented before the conclusion.

The device that has been realized is sketched in Fig. 1.It is composed of a 4 cm long ion-exchanged single-modewaveguide realized on a 1:5 × 1026 m−3 neodymium—

doped phosphate glass (IOG-1 from Schott). The lasercavity is closed on one side by a dielectric mirror, witha reflectivity of 99.9% at λ ¼ 1054 nm and a 99% transmis-sion at λ ¼ 800 nm (pump wavelength), stuck on thewaveguide’s input facet. On the other side, the outputfacet has been carefully polished at an angle of 90° withrespect to the waveguide to provide a Fresnel reflectionof 4%. The quality factor of the cavity is passively modu-lated thanks to a saturable absorber. For the absorber,we used a bis(4-dimethylaminodithiobenzil)nickel (BDN)

Fig. 1. (Color online) Schematic view of the realized device.The output facet provides laser feedback due to Fresnelreflection and allows direct fiber coupling.

June 1, 2011 / Vol. 36, No. 11 / OPTICS LETTERS 1987

0146-9592/11/111987-03$15.00/0 © 2011 Optical Society of America

dye dispersed in a cellulose acetate (CA) thick film de-posited on the waveguide’s surface.The amplifying waveguide was realized using a

Agþ=Naþ ion exchange through a 120 nm thick siliconmask, deposited by sputtering and etched by a SF6 reac-tive ion etching. Indeed, this process prevents the altera-tion of the glass surface that has been observed whenusing a standard aluminum mask. For a molten salt bathof 20mol:% AgNO3–80mol:% NaNO3 at 320 °C, a diffu-sion time of 4 min through an aperture of 4 μm was used.These parameters were determined to obtain the maxi-mum effective volume of the laser cavity while maintain-ing a single-mode operation. The calculations werecarried out using a modeling of the ion-exchange process[13,14], taking into account the mixed-alkali effect [15]and a scalar mode solver (Optiwave). The saturableabsorber layer was realized by mixing the CA powderwith a BDN-doped acetone solution. Stirring the solutionfor 20 min at 60 °C allowed us to obtain a homogeneousgel-like material, which was then deposited on the sur-face of the waveguide and dried under rough vacuumat 150 °C for 10 min. The film prepared this way showedgood adhesion on the glass substrate.The mode profiles of the waveguide were measured at

λ ¼ 1054 nm by injecting light with a HI1060 single-modeoptical fiber while imaging the waveguide output fieldthrough a microscope objective on a silicon CCD camera.The mode shape, depicted on Fig. 2, was fitted by Gaus-sians along the horizontal and vertical axes showing 1=e2

diameters of 7:6� 0:2 μm and 5:0� 0:2 μm, respectively.These dimensions allow a 90% theoretical coupling effi-ciency with a HI1060 optical fiber. Replacing the camerawith a calibrated photodetector, propagation losses low-er than 0:1dB=cm were determined with and without anundoped CA cladding, which proves the good opticalquality of the CA layer. These low propagation losses alsoconfirm the improvement of the waveguide surface qual-ity obtained due to the use of the silicon mask process.A systematic study of the influence of the BDN con-

centration in the cellulose acetate film on the laserpeak power has been carried out. For this experi-ment, the device was pumped by a continuous-wave

titanium–sapphire laser emitting ð430� 1ÞmW atλ ¼ 800 nm coupled into the waveguide with anf ¼ 15:3mm lens. The laser output beam was collectedby microscope objective and launched into an integratingsphere to measure the average output power. Then, thepulse width and the repetition frequency were deter-mined with a 25GHz bandwidth photodetector con-nected to a 6GHz oscilloscope that provided thetemporal evolution of the laser output. Finally, the peakpower was calculated by dividing the average power bythe repetition frequency and by the pulse width.

Measuring the laser characteristics for different satur-able absorber concentrations, we observed that the peakpower increased steadily with the doping level of thesaturable absorber, until BDN concentration reached2:4 × 1024 m−3, where a sharp decrease of the peak poweroccurred, as can be seen on Fig. 3. This behavior is due tothe formation of BDN clusters in the CA at high concen-trations, which entails a dramatic increase in the unsatur-able losses of the laser cavity. For an optimal BDNconcentration of 2:4 × 1024 m−3, an average output powerof ð38:8� 0:5ÞmW was measured. Figure 4 presentsthe temporal evolution of the emitted pulses, whichpresent a FWHM of ð1:3� 0:1Þns and are spaced by

Fig. 2. (Color online) Laser output beam intensity picture withvertical and horizontal intensity profiles (λ ¼ 1054nm).

Fig. 3. (Color online) Measured laser peak power as a functionof the BDN concentration.

Fig. 4. (Color online) Temporal evolution of the optimizedlaser (NBDN ¼ 2:4 × 1024 m−3) when pumped with 430mW. Insetshows the shape of a single pulse.

1988 OPTICS LETTERS / Vol. 36, No. 11 / June 1, 2011

ð35� 2Þ μs, corresponding to a repetition frequency ofð28� 2ÞkHz. A peak power of ð1:0� 0:15ÞkW wasdeduced from these measurements. This result showsthat a peak power three orders of magnitudes higher thanwhat has been reported previously for similar devices[11,12] has been obtained. Moreover, since the interac-tion of the laser pulses with the saturable absorber isquite weak and distributed along the entire laser cavity,no photobleaching has been observed after more thanone day of continuous operation.Finally, the mirror-free output facet allows direct

coupling of the emitted pulses in an optical fiber byapproaching it close to the waveguide. The fiber can thenbe glued to the waveguide, provided that an air gap is leftaround the fiber core to maintain the 4% Fresnel reflec-tion. Doing so, we succeeded in coupling ð650� 100ÞWpeak power pulses into an HI1060 Corning optical fiber,which corresponds to a 65% coupling efficiency. Thisvalue can be increased by a tapering of the waveguide’soutput. Taper could also be used to increase the effectivevolume of the cavity and allow pumping by large area la-ser diode, as proposed by Madasamy et al. [16]. Appliedto Q-switched laser operation, this approach could alsoincrease the pulse energy. However, a 650W fiber-coupled peak power corresponds to a power densityof 21:5MW⋅m−2, which already allows obtaining non-linear effects in optical fibers. Thus, if coupled with aphotonic crystal fiber, our device can be used as a pumpfor compact and self-aligned supercontinuum sources.To conclude, we presented an integrated Q-switched

laser, which has been realized by a Agþ=Naþ ion ex-change on a neodymium-doped phosphate glass. Satur-able absorption is provided by a BDN-doped CA filmhybridized on the device surface. This device has shownperformances 3 orders of magnitude higher than the pre-viously published devices reaching a peak power ofð1:0� 0:15Þ kW. Moreover, the presence of a mirror-free

output facet allows stable coupling into optical fiber withan efficiency of 65% without requiring any coupling op-tics. It could therefore be used in embedded optical sen-sors or to create compact supercontinuum sources.

References

1. R. Paschotta, R. Häring, E. Gini, H. Melchior, U. Keller, H. L.Offerhaus, and D. J. Richardson, Opt. Lett. 24, 388 (1999).

2. M. Delgado-Pinar, D. Zalvidea, A. Diez, P. Perez-Millan, andM. Andres, Opt. Express 14, 1106 (2006).

3. T. Kitagawa, K. Hattori, M. Shimizu, Y. Ohmori, and M.Kobayashi, Electron. Lett. 27, 334 (1991).

4. J. E. Broquin, Proc. SPIE 6475, 647507 (2007).5. D. Barbier, M. Rattay, F. Saint André, G. Clauss, M.

Trouillon, A. Kevorkian, J. M. P. Delavaux, and E. Murphy,IEEE Photon. Technol. Lett. 9, 315 (2002).

6. A. Yeniay, J. M. P. Delavaux, J. Toulouse, D. Barbier, T. A.Strasser, and J. R. Pedrazanni, IEEE Photon. Technol. Lett.9, 1099 (1997).

7. S. Blaize, L. Bastard, C. Cassagnetes, and J. E. Broquin,IEEE Photon. Technol. Lett. 15, 516 (2003).

8. D. L. Veasey, D. S. Funk, N. A. Sanford, and J. S. Hayden,Appl. Phys. Lett. 74, 789 (1999).

9. J. B. Schlager, B. E. Callicoatt, R. P. Mirin, and N. A.Sanford, IEEE Photon. Technol. Lett. 14, 1351 (2002).

10. J. S. Wilkinson and E. K. Mwarania, in Proceedings of the

Third Microoptics Conference (1991), pp. 114–117.11. J. A. Aust, K. J. Malone, D. L. Veasey, N. A. Sanford, and

A. Roshko, Opt. Lett. 19, 1849 (1994).12. R. Salas-Montiel, L. Bastard, G. Grosa, and J. E. Broquin,

Mater. Sci. Eng. B 149, 181 (2008).13. D. Bucci, J. Grelin, E. Ghibaudo, and J. E. Broquin,

Electron. Lett. 42, 920 (2006).14. D. L. Veasey, J. M. Gary, and J. Amin, Proc. SPIE 2996,

109 (1997).15. A. Lupascu, A. Kevorkian, T. Boudet, F. Saint-André, D.

Persegol, and M. Levy, Opt. Eng. 35, 1603 (1996).16. P. Madasamy, S. Honkanen, D. F. Geraghty, and N.

Peyghambarian, Appl. Phys. Lett. 82, 1332 (2003).

June 1, 2011 / Vol. 36, No. 11 / OPTICS LETTERS 1989