Ultraviolet writing of channel waveguides in proton ... - ePrints Soton

3 downloads 0 Views 125KB Size Report
Sergey M. Kostritskii. Moscow Institute of Electronic Technology, .... Lett. 25, 458 1974. 3Yu. N. Korkishko and V. A. Fedorov, Ion Exchange in Single Crystals for.
JOURNAL OF APPLIED PHYSICS 101, 014110 共2007兲

Ultraviolet writing of channel waveguides in proton-exchanged LiNbO3 Katia Gallo,a兲 Corin B. E. Gawith, Iain T. Wellington, Sakellaris Mailis, Robert W. Eason, Peter G. R. Smith, and David J. Richardson Optoelectronics Research Centre, University of Southampton, Southampton SO17 1BJ, United Kingdom

Sergey M. Kostritskii Moscow Institute of Electronic Technology, Zelenograd 124498, Russia

共Received 8 August 2006; accepted 26 September 2006; published online 10 January 2007兲 We report on a direct ultraviolet 共UV兲 writing method for the fabrication of channel waveguides at 1.55 ␮m in LiNbO3 through UV irradiation of surface and buried planar waveguides made by annealed proton exchange and reverse proton exchange. A systematic study of the guidance properties as a function of the UV writing conditions is presented. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2400511兴 I. INTRODUCTION

II. THE TECHNOLOGY

LiNbO3-based integrated devices are key components for high-speed optical signal processing in optical telecommunications networks.1 Traditional waveguide fabrication techniques in LiNbO3 involve metal indiffusion2 or proton exchange 共in its multiple variations兲3 and rely on photolithography to define channel waveguide patterns. Recently, an alternative way to produce channel waveguides in congruent LiNbO3 has been demonstrated, based on direct ultraviolet 共UV兲 writing.4 In this letter we report on a hybrid approach for the photoinscription of guiding channels in LiNbO3 crystals, based on the combination of a traditional waveguide fabrication technology 共namely, proton exchange兲 with subsequent direct UV writing. Compared to UV writing in pure LiNbO3,4 the preliminary treatment of the surface layers via proton exchange can significantly improve the modal confinement and provide additional degrees of freedom to tailor the properties and depth of the resulting channels. The use of a direct-writing technique to define the guiding geometries on LiNbO3 completely eliminates the need for expensive and multistep lithographic processing. Our process involves firstly the fabrication of a 共surface or buried兲 planar waveguide by annealed proton exchange 共APE兲 or reverse proton exchange 共RPE兲,3 followed by irradiation with UV light to define guiding channels in the protonated layers. By means of systematic microscopic and optical characterizations of the UV channels we could identify the optimum fabrication conditions for high lateral confinement and single-mode operation around 1550 nm. The results presented in this letter were recorded over several months after UV exposure on z-cut APE/RPE LiNbO3 single-mode planar waveguides guiding only extraordinarily polarized light 共TM兲 at ⬃1550 nm and have proven the longevity of this improved process over the results obtained in Ref. 4. a兲

Electronic mail: [email protected]

0021-8979/2007/101共1兲/014110/3/$23.00

The surface APE planar waveguides were made through the sequence of proton exchange in pure benzoic acid 共BA兲 at 158 ° C for 15 h and annealing in air at 325 ° C for 26 h. The extraordinary refractive index profile of the APE slab 关Fig. 1共a兲兴 was reconstructed via Chiang’s algorithm5 from effective index measurements performed at 633 nm. The buried RPE slabs were fabricated by means of an initial 1.1 ␮m deep proton exchange in BA at 158 ° C, followed by annealing at 325 ° C for 8.5 h and reverse proton exchange in a eutectic mixture of LiNO3 : NaNbO3 : KNbO3 at 330 ° C for 10.5 h. This resulted in the simultaneous formation of a planar waveguide for the ordinary polarization in the top RPE layer 共guiding 633 nm but not infrared light兲 and of a buried slab beneath it, guiding extraordinarily polarized light at 1550 nm. The extraordinary refractive index profile 关Fig. 1共b兲, solid line兴 was inferred through a more elaborated procedure, combining effective index measurements on the surface layers before RPE and after a RPE-

FIG. 1. Schematic of the UV writing configuration. Depth 共z兲 distributions of the extraordinary refractive index increase 共⌬ne兲 at 633 nm in the top layers of the substrate for UV writing on 共a兲 APE: LiNbO3 and 共b兲 RPE: LiNbO3 planar waveguides.

101, 014110-1

© 2007 American Institute of Physics

Downloaded 10 Nov 2009 to 152.78.208.72. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

014110-2

J. Appl. Phys. 101, 014110 共2007兲

Gallo et al.

equivalent annealing step, as detailed in Ref. 6. The estimated depth of the buried waveguide is z0 = 1.7 ␮m. To photoinscribe the channels in the APE/RPE slabs we employed the same setup used in Ref. 4: a 244 nm continuous wave UV beam from a FRED laser 共Coherent兲 was collimated and focused to a spot size of 7 ␮m at the surface 共−z兲 of the samples which were mounted on x-y nanopositioning stages. The channels were written by translating the crystal at a constant speed under the UV writing spot. The structure is sketched in Fig. 1. In all cases, the writing direction was parallel to the crystallographic x axis of the substrate. To identify the optimum writing parameters, we varied the UV powers and scanning speeds in the following ranges: P = 30– 80 mW and v = 10– 1000 mm/ min. For our fixed beamwidth 共w = 7 ␮m兲, the above conditions corresponded to UV energy fluences, F = Pw−1v−1 ⬃ 26 J / cm2 – 7 kJ/ cm2. III. RESULTS AND DISCUSSION

To a first approximation, the most relevant control parameter appeared to be the writing power. Good infraredguiding channels could be written with UV powers in the 30– 50 mW range. Below ⬃30 mW the refractive index increase induced by UV irradiation in the stripes was too low to induce optical confinement in the channel. On the other hand, too high an UV writing power also prevented guidance in the exposed channels. The latter effect could be reasonably ascribed to surface damage observed on these samples, likely to impair guidance in the original APE/RPE planar waveguides. Consistent with this, the upper threshold for the useful UV writing power range was slightly higher in the RPE samples, whose buried guiding layers are less sensitive to surface morphology. We carried out more detailed studies on the strongest channels by systematically analyzing their mode profiles at 1550 nm, under fiber butt coupling or free-space coupling excitation. We imaged the mode intensity distribution at the channel output via a 25⫻ objective lens and used a commercial mode profiler for accurate mode size measurements. In all cases, the optimum depth for incoupling corresponded to the position of the original APE or RPE planar waveguide, and only TM modes were guided. Figure 2 shows the measured lateral and vertical mode intensity distributions 共solid lines兲 of two well-guiding single-mode 共TM00兲 channels written on APE: LiNbO3 关共a兲 and 共b兲兴 and RPE: LiNbO3 关共c兲 and 共d兲兴. The vertical profiles of the TM00 modes matched well the depth distributions of the TM0 modes of the original APE/RPE planar waveguides 关cf. solid and dashed lines in Fig. 2共a兲兴. The refractive index increase induced by UV irradiation is responsible for the lateral mode confinement in the channels. We estimated empirically the index change associated with UV writing by assuming an index profile of the form ⌬n共z , y兲 = ⌬n共z兲⌬n共y兲, ⌬n共z兲 being the refractive index distribution of the original APE/RPE planar waveguide 关Figs. 1共a兲 and 1共b兲兴 and ⌬n共y兲 the contribution due to UV writing. The assumption made on ⌬n共z兲 corresponds to considering the UV-induced index changes as a small perturbation with respect to the ones associated with proton exchange.

FIG. 2. 共a兲 Vertical and 共b兲 lateral intensity distributions of the TM00 mode at 1.55 ␮m for a channel UV written with 40 mW, at 500 mm/ min, on the APE: LiNbO3 surface planar waveguide of Fig. 1共a兲. 共c兲 and 共d兲 are the same as 共a兲 and 共b兲 for a channel UV written with 40 mW, at 80 mm/ min, on the RPE: LiNbO3 buried planar waveguide of Fig. 1共b兲. The dashed line in 共a兲 is the TM0 mode profile of the APE planar waveguide before UV writing. The dotted lines in 共b兲 and 共d兲 are the numerical fits used in the evaluation of the refractive index increase induced by UV irradiation 共⌬nUV兲.

The good agreement between the mode distributions in depth for irradiated and nonirradiated areas 关cf. Fig. 2共a兲兴 supports this assumption. For ⌬n共y兲, we assumed a top-hat profile with a width equal to the spot size of the writing beam 共w兲, ⌬n共y兲 =



⌬nUV 1

for 兩y兩 艋 w/2

for 兩y兩 ⬎ w/2.



We then determined the value of ⌬nUV by fitting the calculated lateral mode profiles 关dashed lines in Figs. 2共b兲 and 2共d兲兴 to the measured ones 共solid lines兲. For the UV written channels in APE and RPE shown in Fig. 2, the fits yielded ⌬nUV = 9 ⫻ 10−4 and 5 ⫻ 10−4, respectively. The effect of the UV scanning speeds on the modal confinement is illustrated in Fig. 3, in which the TM00 mode lateral width of UV channels written at 40 mW is plotted as a function of the writing speed for APE: LiNbO3 共empty circles兲 and RPE: LiNbO3 共filled circles兲. For the channels in APE: LiNbO3, the lateral mode size is almost constant 关9 – 10 ␮m full width at half maximum 共FWHM兲兴 over a broad range of writing speeds and degrades only when these decrease below 100 mm/ min. As previously discussed, this effect is associated with surface damage occurring at higher

FIG. 3. FWHM of the lateral intensity distribution of the TM00 mode at 1.55 ␮m in channels written at 40 mW as a function of the scanning speed for APE: LiNbO3 共empty circles兲 and RPE: LiNbO3 共filled circles兲.

Downloaded 10 Nov 2009 to 152.78.208.72. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

014110-3

J. Appl. Phys. 101, 014110 共2007兲

Gallo et al.

writing fluences. For the RPE case, the trend is slightly different: the optical confinement is still poor at high writing speeds 共as a higher fluence is needed to induce a significant refractive index change in the buried layers兲, but it improves significantly when the writing speed is reduced below 100 mm/ min. The mechanisms responsible for waveguide formation cannot be conclusively stated at this stage. Preliminary micro-Raman measurements on UV irradiated stripes in APE: LiNbO3 give evidence for amorphization of the crystal in regions irradiated at higher fluences, when no channel waveguides are produced. In those cases, the Raman spectra exhibit new phonon lines associated with the formation of LiNb3O8 micrograins, as observed in LiNbO3 crystals after thermally induced decomposition leading to Li2O outdiffusion.3 By contrast, Raman measurements on guiding channels written in APE: LiNbO3 show no characteristic signature for LiNb3O8 micrograin formation, but exhibit new phonon lines with widths comparable to the main phonon bands in virgin crystals. We are further investigating the nature of these additional lines 共e.g., at 782 cm−1兲, normally not present in LiNbO3 Raman spectra 共i.e., neither in APE nor in congruent LiNbO3兲. Another important aspect concerns waveguide stability and longevity. Although lasting longer and exhibiting a higher resilience to photorefractive damage than waveguides UV written in pure LiNbO3, all of our channels in APE: LiNbO3 and RPE: LiNbO3 were found to disappear after two years. We also discovered that some of the waveguides could be “refreshed” through a low-temperature

annealing followed by fast cooling, but this issue will be specifically addressed in a future publication. In conclusion, we demonstrated a hybrid technique to fabricate surface and buried guiding channels in LiNbO3, based on the combination of proton exchange and direct UV writing. Extensive optical measurements allowed us to identify the optimum range of fabrication conditions for the production of well-guiding single-mode channel waveguides at telecommunications wavelengths. A deeper understanding of the physics involved in waveguide formation and aging should enable the exploitation of this technology to produce complex planar light wave components in LiNbO3, as routinely achieved in other materials.7 ACKNOWLEDGMENTS

This work is supported by INTAS 共Project No. 03-516562兲. One of the authors 共K.G.兲 also gratefully acknowledges support from the Leverhulme Trust through an Early Career Fellowship 共ECF/2004/0401兲. 1

F. J. Leonberger, IEEE/LEOS 13th Annual Meeting Conf. Proceedings 共IEEE, New York, 2000兲, Vol. 1, p. 5. R. V. Schmidt and I. P. Kaminow, Appl. Phys. Lett. 25, 458 共1974兲. 3 Yu. N. Korkishko and V. A. Fedorov, Ion Exchange in Single Crystals for Integrated Optics and Optoelectronics, 共Cambridge International Science, Cambridge, 1999兲. 4 S. Mailis, C. Riziotis, I. T. Wellington, P. G. R. Smith, and C. B. E. Gawith, R. W. Eason, Opt. Lett. 28, 1433 共2003兲. 5 K. S. Chiang, J. Lightwave Technol. 3, 385 共1985兲. 6 K. Gallo, C. Codemard, C. B. E. Gawith, P. G. R. Smith, J. Nilsson, N. G. Broderick, and D. J. Richardson, 4th IEEE/LEOS Workshop on Fibres and Optical Passive Components (WFOPC) Proceedings 共IEEE, New York, 2005兲, Vol. 1, p. 36. 7 G. D. Emmerson, S. P. Watts, C. B. E. Gawith, V. Albanis, M. Ibsen, R. B. Williams, and P. G. R. Smith, Electron. Lett. 38, 1531 共2002兲. 2

Downloaded 10 Nov 2009 to 152.78.208.72. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Suggest Documents