MODAL BISTABILITY IN A GaAlAs LEAKY WAVEGUIDE J. Valera, J. Aitchison, D. Goodwill, A. Walker, I. Henning, S. Ritchie To cite this version: J. Valera, J. Aitchison, D. Goodwill, A. Walker, I. Henning, et al.. MODAL BISTABILITY IN A GaAlAs LEAKY WAVEGUIDE. Journal de Physique Colloques, 1988, 49 (C2), pp.c2-307-c2-310. <10.1051/jphyscol:1988272>. <jpa-00227689> HAL Id: jpa-00227689 https://hal.archives-ouvertes.fr/jpa-00227689 Submitted on 1 Jan 1988 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Colloque C2, Suppl6ment au n06, Tome 49, juin 1988 MODAL BISTABILITY IN A GaAlAs LEAKY WAVEGUIDE J.D. VALERA, J.S. AITCHISON(~), D.J. GOODWILL, A.C. WALKER, I.D. HENNING* and S. RITCHIE* Department of Physics. Heriot-Watt University, Riccarton, GB-Edinburgh EX14 4AS, Scotland, Great-Britain *BTRL Martlesham Heath, GB-Ipswich IP5 7RE, Great-Britain Abstract - We report the observation of power dependent mode changes leading to a nonlinear transmission response and optical bistability in a GaAlAs leaky slab waveguide. The dominant mechanism appears to be an optothermal nonlinearity.. Previously reported observations of optical bistability in a GaAs/GaAlAs waveguide cavity included whole-sample thermally induced absorptive and refractive bistability /1/ and localised (ps) thermally induced absorptive bistability /2/. Other groups have observed refractive bistability due to an optoelectronic mechanism in MQW GaAs/GaAlAs waveguides /3,4/. These experiments were all conducted in a proper waveguide geometry with confined modes. In this paper we report the observation of bistability due to changes in the modes propagating in a leaky GaAs/GaAlAs waveguide. In a normal planar waveguide - a high refractive index layer between two lower index layers - total internal reflection permits the propagation of the allowed modes with low loss. A leaky planar waveguide, such as those used here, has the reverse structure: a low index layer bounded by two higher index layers. Although there is then no total internal reflection, grazing incidence reflections still support discrete modes. These suffer radiative losses, where for the nth order mode the attenuation coefficient is proportional to (n+li2 /5/. In this investigation the waveguides used were short enough that a significant fraction of the incident light passed through the waveguide without being radiated. They were also thick enough to support several modes. The output of an argon-ion pumped styryl-9 infra-red dye laser, operating in the 0.8-0.85 p range, was focussed by a X40 microscope objective to a spot diameter of 2 p and end-fire coupled into the waveguide. The output face of the guide was imaged by a X20 objective onto a slit and then by a further relay lens onto a Si-Vidicon camera. A Hamamatsu digitiser/frame-store and a Hewlett-Packard HP-85 computer were used to record the output mode profiles. The input and output powers were monitored with Si photodiodes. 0-60 mw of optical power was available into the input objective - adjustable via an acousto-optic modulator. The plane of polarization of the input light could be changed by the insertion of a h/2 plate. The sample consisted of a 3-layer planar structure grown by molecular beam epitaxy (MBE). On a GaAs substrate were deposited a 3.6 pm thick layer of Ga,,A1,,As, a 1.7 p layer of Ga,,A1,,As and finally a 0.75 pm cladding layer of Ga,,A1,,As. At 825 nm the thin 10% A1 layer forms a well-confined single-mode asymmetric guiding layer. The thick 13% A1 layer, bounded by two layers of higher refractive index, forms a multimode leaky waveguide. The sample was cleaved into short lengths; the 32% reflectivity end-faces creating a low finesse Fabry-Perot cavity. Two lengths of sample were used: 200 pm and 400 pm, Each was mounted in wax on an aluminium holder and so was poorly heatsunk. Both guides showed power dependent modal effects when probed by 820-836 nm radiation. (''present address : Bell Comunicationc Research, Red Bank, NJ 07701. U.S.A. Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1988272
(b) Fig. 1 - Transmission characteristics and mode profiles for maximum coupling at low power. Figures la, 2a, 3a show a set of input/output characteristics for TE modes in the 200 pm long leaky guide at 828 nm. In Fig. la, the overlap between the Gaussian input beam and the zero order mode is maximised so that most light is coupled into the guide. For inputs up to 35 mw, there are small bistable loops due to whole-sample heating leading to refractive index changes. These are expected in a short Fabry-Perot waveguide. From 35 mw to 40 mw input power, there is a large drop in transmission with a 3:l contrast ratio which is apparently bistable. At higher powers (not shown) the transmitted power increases until there is a fast (ps), almost total, switch-off due to localized thermally-induced absorptive switching. Associated with the first drop in transmission are slight changes in the output mode profiles (Fig. lb). These are more apparent in the second case (Figs. 2a and 2b), where the input spot was shifted slightly closer to the substrate. The overlap between the input Gaussian and the low power zero order mode is less in this case, so the initial transmission is lower. Up to 23 mw input, there are again small bistable loops. The drop in transmission between 23 mw and 33 mw input is more gradual and clearly not bistable. However it is associated with major changes in the mode profile. At low power there is a single peak in the centre of the leaky layer. As the input power rises, this moves towards the upper layers and, when the transmission drops, a second peak appears near the substrate. As the input power increases further this second peak becomes dominant and moves towards the centre of the guide. Finally there is the fast switch-off due to localized heating. Figs. 3a and 3b show the case where the input spot is shifted in the other direction: towards the 10% A1 layer. This indicates similar effects to those in Fig. 1, in that there is a sharp and apparently bistable drop in transmission around 33 mw input. However it now shows a greater contrast ratio of 4:l and the mode changes are more evident. The initial single peak mode moves towards the substrate as power is increased then jumps back towards the centre of the leaky layer as the drop in transmission occurs. No double-peaked distribution is observed. When the plane of polarization of the laser light is rotated so that TM modes are excited, similar results are obtained. At longer wavelengths the double-peaked mode is more apparent and the drop in transmission occurs at higher input powers, while theaswitch contrast increases.
Fig. 2 - Transmission characteristics and mode profiles for input coupling near substrate. Fig. 3 - Transmission characteristics and mode profiles for input coupling near 10% A1 layer. The longer (400 pun) sample has greater attenuation and so its overall transmission is lower. Figure 4 shows the input/output characteristics for this guide (TE polarization) at 828 nm, when the input spot is centralised on the leaky layer. The small bistable loops are narrower than for the shorter guide as the Fabry-Perot cavity is longer, and show lower contrast as the greater attenuation reduces the finesse. The drop in transmission shows up to 5:l contrast ratio. The single-peaked mode is much broader than for the shorter sample but again it does not change much as the power is scanned. Even when the input coupling is near the substrate, no double-peaked mode is observed. A fully quantitative explanation of these phenomena has not yet been attempted. However we can come to some general conclusions at this stage. Time constants of > 100 ms imply that all the effects described here are thermal. 828 m radiation lies within the bandedge of the substrate and in the bandtail of the 10% and 12% A1 layers so there is significant absorption. This heats the structure and the resultant shift in the
Fig. 4 - Transmission characteristics for the 400 pm sample. Fig. 5 - Transmission characteristics for the 200 pm sample showing multistability. band-edge (de /dt = - 4 x lo-' ev K-' /6/) causes refractive index changes. As the refractive insex steps between the layers are small and the modes depend critically upon these steps, there will clearly be changes in the mode profiles. If the original TE, mode moves away from the centre of the leaky layer, it becomes more leaky and has a larger component propagating in the more absorptive substrate or 10% A1 layer. This causes a further increase in temperature and further induced absorption. Thus a sharp drop in transmission can occur, giving the possibility of bistability or, in a cavity with refractive nonlinearities, multistability (Fig. 5). The double-peaked distribution observed may be identified as a TE, mode. Such a mode has an attenuation coefficient, due to radiative losses alone, four times that of a TE, mode and, as the radiated power is coupled directly into absorbing outer layers, it also has a much higher absorption coefficient. Thus, while both TE, and TE, modes can propagate significantly through the shorter sample (Fig. 2), only the TE, mode is observed at the output of the longer sample (Fig. 4). A further consideration is the changes in the input coupling efficiencies. These are calculated from the overlap integrals between the input Gaussian and the modes of the guide. The overlap with the TE, mode is largest when the input spot is away from the centre of the guide and hence the double-peaked distribution is most clearly seen when the input spot is offset from the guide centre. A quantitative assessment of the modal changes requires a very precise knowledge of the aluminium content of the layers and the temperature variation of the refractive index near the band edge to a precision of better than 0.1%. (The index step between the leaky guide and the 10% A1 :Layer is only % 0.02). These data are not available to sufficient accuracy. In addition, most modelling of waveguides has been for high index, non-leaky guides and further theoretical work is needed to achieve a more complete understanding of these modal effects. Thanks are due t:o Dr D.A. Andrews for the samples used in this investigation and to the Director of Research and Technology, British Telecom for permission to publish this paper. References /1/ Walker A.C., Aitchison J.S., Ritchie S. and Rodgers P.M., Electron Lett., 22, 366 (1986). /2/ Aitchison.l.S., Valera J.D., Walker A.C., Ritchie S., Rodgers P.M., McIlroy P. and Stegeman G-I., Appl. Phys. Lett., Sl, 561 (1987). /3/ Li Karn Wa 1'. and Robson P.N., IEEE J. Quantum Electron., QE-23, 1962 (1987). /4/ Warren M., Gibbons W., Komatsu K., Sarid D., Hendricks D., Gibbs H.M. and Sugimoto M., Appl. Phys. Lett., 5l. 1209 (1987). /5/ Yariv A. "Optical Electronics" Holt, Rinehart and Winston, 452 (1985). /6/ Adachi S., J. Appl. Phys., 58, R1 (1985).