Low-loss shallow-rib waveguides were fabricated using As2Se3 chalcogenide glass and polyamide-imide polymer. Waveguides were patterned directly in the As2Se3 layer by photodarkening followed by selective wet etching. Theory predicted a modal effective area of 3.5–4 µm2, and this was supported by near-field modal measurements. The Fabry-Perot technique was used to estimate propagation losses as low as ~0.25 dB/cm. First-order Bragg gratings near 1550 nm were holographically patterned in some waveguides. The Bragg gratings exhibited an index modulation on the order of 0.004. They were used as a means to assess the modal effective indices of the waveguides. Small core As2Se3 waveguides with embedded Bragg gratings have potential for realization of all-optical Kerr effect devices.
© 2004 Optical Society of America
There has been considerable interest in the third-order nonlinear properties of chalcogenide glasses in the past decade [1–5]. In particular, As2Se3 glass has a bandgap energy (Eg ~1.75 eV) that is slightly larger than twice the photon energy at λ=1550 nm. As a result, it is one of the more promising alloys for 3rd-order nonlinear devices in fiber networks. For example, it exhibits an ultrafast non-resonant Kerr nonlinearity ~3 orders of magnitude larger than that of SiO2 without excessive two-photon absorption [3–5]. Recently, the Raman  and Brillouin  gain properties of As2Se3 and closely related alloys have been studied.
Since nonlinear effects scale with the optical intensity, a common goal of nonlinear guided wave optics is to minimize the effective modal area (Aeff). For example, Spalter et al. [2,7] have analyzed the potential of realizing integrated Kerr effect nonlinear switching devices in chalcogenide glass. They predicted that, for waveguides with length of a few centimeters and effective area on the order of ~1 µm2, nonlinear phase shifts in excess of π are possible for pulses of 1 ps duration and 1 pJ energy (~1 W peak power). An implicit requirement is that the waveguide’s linear loss is negligible, as specified by the commonly used figures of merit . To date, linear and nonlinear propagation in chalcogenide-based waveguides has been studied using relatively weakly confining chalcogenide fibers  or integrated waveguides [7–10] with Aeff~8–40 µm2. However, the excellent thermo-mechanical compatibility and high index contrast between chalcogenide glasses (ex. As2Se3 with n~2.8) and certain high performance polymers (polysulfone, polyether sulfone, polyether imide with n~1.5–1.6) have enabled the realization of photonic bandgap fibers based on these materials [11–12] and illustrates the potential for high index contrast waveguiding.
In this paper, we describe experimental results for high index contrast planar rib waveguides fabricated using As2Se3 chalcogenide glass as a core material and a compatible polymer (polyamide-imide) as cladding material. We realized waveguides with Aeff~3.5–4 µm2 and linear losses as low as ~0.25 dB/cm at 1530 nm. Bragg gratings with stop band extinction as high as 20 dB were embedded in some of these waveguides, and were used to assess modal effective indices. Experimental and simulated results are in good agreement.
2. Device fabrication
Chalcogenide glasses exhibit numerous photostructural changes when exposed to near-bandgap light. For example, illumination of fully annealed chalcogenide films generally results in photodarkening (a red-shift in the absorption band edge, which is typically reversible by subsequent annealing) and a corresponding increase in refractive index . These effects have been used in the fabrication of laser-written channel waveguides  and holographically induced volume gratings . Photodarkening is often accompanied by a large change of etch rate (in suitable wet etchants), so that structures such as relief gratings  and rib or strip waveguides can be realized by patterned exposure of the chalcogenide film followed by etching.
Notwithstanding their unique options for processing, there are some challenges associated with the use of chalcogenide glasses in integrated optics. While toxicity and durability are limitations of some alloys , probably the key challenge is the large coefficient of thermal expansion (CTE) of these glasses relative to typical substrates and undercladding materials . While high quality thin films (As-Se, As-S, Ge-Se, etc.) can be deposited by various techniques such as evaporation, sputtering, etc., a post-deposition anneal is often required to stabilize and densify the film, and to reduce scattering losses . This annealing step can lead to film delamination or cracking if the chalcogenide film is deposited on a low thermal expansion substrate such as Si or SiO2. Rapid thermal annealing (RTA)  and use of as-deposited films (by pulsed laser deposition)  have been reported to mitigate this. We used a polyamide-imide (PAI) polymer, (Torlon AI-10 from Solvay Advanced Polymers) , with a CTE (~30 ppm/°C) well matched to that of As2Se3 (21 ppm/°C) , as both the undercladding and uppercladding material. PAI is a robust polymer with a high glass transition temperature (270 °C). Of particular importance here, it can be fully cured at temperatures below the glass transition temperature of As2Se3 (~190 °C) .
Several wafers were processed as follows. The PAI polymer undercladding was spun cast onto silicon wafers as described elsewhere . Subsequently, a 1.1 µm thin film of As2Se3 was deposited by thermal evaporation onto room temperature substrates. The evaporation rate was ~10 nm/s. Uniformity and stoichiometry of the chalcogenide films were verified by electron microprobe analysis. Waveguides were patterned directly in the as-deposited As2Se3 films by exposure through a photomask in a standard UV mask aligner . This exposure photodarkens the glass, and greatly increases its resistance to amine-based wet etchants . Exploiting this property, the unexposed chalcogenide film was etched ~100–200 nm using a monoethanolamine (MEA) based solution. It should be noted that annealed As2Se3 films exhibited reverse polarity in MEA, with exposed portions etching more quickly. A second PAI layer was spun cast and cured at 150 °C for one hour in nitrogen atmosphere; the As2Se3 core layer is effectively annealed at the same time. Wafers were cleaved using a conical diamond cutter, and an SEM image of a typical waveguide facet is shown in Fig. 1. Wafers were free from visible cracks, and there was no evidence of film delamination on cleaving. This attests to the good thermo-mechanical compatibility of the PAI polymer and As2Se3, and suggests that PAI is able to absorb the thermally induced mechanical stresses between the As2Se3 layer and the Si substrate.
After dicing the wafers into chips containing waveguides of various lengths, Bragg gratings were embedded in some of the waveguide samples. Gratings were holographically patterned by photodarkening the As2Se3 layer using a standard 633 nm wavelength HeNe source. To write first-order gratings (with required pitch smaller than half the free-space wavelength of the writing beam), samples were placed in intimate contact with a prism  as illustrated in Fig. 2. The PAI uppercladding has good transparency at 633 nm. Using this technique, high quality gratings with periods on the order of 290 nm (see Fig. 2, inset) were realized. Note that for a fully annealed chalcogenide film, we expect to be in the regime of reversible photodarkening . The induced index change has been shown to depend not only on thermal history of the film, but also on the energy and intensity supplied by the writing beam . The single beam intensity at the sample was estimated to be 0.3 W/cm2 and exposure time was typically on the order of 10 min.
3. Near field mode analysis
Waveguides were tested using laser sources at 980 nm, 1480 nm, and near 1550 nm wavelengths, by input coupling from either an objective lens or a high numerical aperture fiber (Nufern, NA~0.26, mode field diameter ~4.8 µm). Theoretical mode profiles were estimated by assuming a square profile for the rib, as illustrated in Fig. 1 (b). The ribs actually have a somewhat rounded profile due to the isotropic nature of the wet etch; nominal rib widths were in the range of 3.8 to 4.3 µm. Modal solutions were obtained using the effective index method and also with various commercial software packages (eg. OptiBPM, from Optiwave and FIMMWAVE from Photon Design). For an etch depth of ~100 nm, it was predicted that the narrowest rib widths (3.8 µm) would support the fundamental TE mode only, while wider ribs supported the second TE mode as well. All rib widths were predicted to support two TM modes (with weak confinement for the second TM mode in some cases). Theoretical TE mode profiles for 3.8 µm wide rib waveguide at 1480 nm wavelength, and for 4.2 µm wide rib waveguide at 980 nm wavelength are shown in Figs. 3 (a) and (b), respectively. These images were obtained using OptiBPM software with perfectly matched boundary conditions. The effective area for the fundamental mode (estimated by the effective index method and using FIMMWAVE) ranged from ~3.5 µm2 for the narrowest ribs to 4 µm2 for the widest ribs.
Experimental near-field profiles were obtained by imaging the waveguide output facets through a series of lenses onto either an infrared Vidicon camera or a silicon CCD camera. It was possible to selectively excite either the fundamental or first-order mode by careful alignment of the input fiber. Fig. 3 (c) shows the fundamental mode at 1480 nm, for a 3.8 µm wide waveguide and Fig. 3 (d) shows the first-order mode at 980 nm, for a 4.2 µm wide waveguide. As shown in Fig. 3, the experimental results and theoretical predictions were in good qualitative agreement. Because of camera nonlinearities, we also assessed the near field mode profiles by replacing the camera with a linear detector apertured by a 5 µm pinhole. This detector was mounted on a precision translation stage, and scanned in two dimensions to map the magnified mode intensity profiles. A typical scan for the horizontal direction (in the plane of the layers) is shown in Fig. 3 (e). These scans confirmed that the horizontal 1/e intensity widths (~3 µm) were in good agreement with those predicted theoretically. The 1/e width in the vertical direction (<1 µm) was below the resolving power of the objective lens employed (N.A.=0.85).
4. Waveguide propagation loss
Waveguides with rib widths ranging from 3.8 to 4.2 µm, and lengths ranging from 5 to 20 mm were used in Fabry-Perot measurements of propagation loss [21,22]. The schematic of the experimental setup for the loss measurement is shown in Fig. 4. The measurement employed a narrow line width, 1530 nm semiconductor DFB laser source with an external fiber isolator to minimize any fluctuation of laser power due to back reflections at the waveguide facets. The input power was tapped using a 90:10 fiber coupler and the output power was normalized in order to account for variation of input power with laser frequency. A fiber polarization controller was used to control the polarization of the input light. Input coupling was via the high NA fiber described previously. A 60X microscopic objective (N.A.=0.85) was used at the output end of the waveguide to focus the light onto a Ge detector. The polarization of the light was confirmed using a polarizer placed between the objective and the detector. By varying the temperature of the laser source, the emission wavelength was varied. The normalized output power was plotted as a function of time as the laser temperature was ramped. A typical Fabry-Perot fringe pattern is shown in Fig. 5 (a). Fringe quality was generally excellent and results were highly repeatable, providing evidence for the high quality of the waveguide facets.
The propagation loss in dB/cm is given by,
where K=I max/I min is the ratio between the maximum and minimum output intensity, R is the geometric mean of the power reflection coefficients of the waveguide end facets, and L is the length of the waveguide in cm . For small core high index contrast waveguides, estimating R using Fresnel reflection expressions with waveguide effective index is not accurate. Hence, we used the approximations provided by Buus . At 1530 nm for example, the Buus formula predicts R~0.266 for the TE0 mode of the waveguides studied (more than 25% larger than the Fresnel estimation). From the data shown in Fig. 5 (a) (representative of the best waveguides realized), for a waveguide length of ~1.75 cm, K is about 2.652 and transmission loss is estimated as 0.26 dB/cm.
The Fabry-Perot technique is well-suited to analysis of low loss, singlemode waveguides. In general, experimental non-idealities (such as the finite line width of the laser source, poor quality facets, or the existence of a second waveguide mode) will result in an overestimate of the loss . As mentioned above, the waveguides studied here were generally bi-moded (except for TE modal propagation in the narrowest rib guides). Well-defined Fabry-Perot fringes, such as those shown in Fig. 5 (a), can provide evidence of single mode propagation. Fig. 5 (b) shows representative data for a set of 8 waveguides lying on the same chip. Predicted losses were generally in the 0.25–2 dB/cm range, reflecting the experimental nature of the fabrication process. While some waveguides contained visual defects, we believe that loss was overestimated in many cases due to presence of two waveguide modes or possibly due to inconsistent facet quality. The scattered light streak images (Fig. 5 inset) and overall insertion losses exhibited less variation.
5. Effective modal indices from Bragg grating characterization
We have previously reported photodarkening-induced index changes as high as ~0.06 in as-deposited As2Se3 thin films [24,25], in agreement with earlier results . Exposure (by near-band gap light) of as-deposited (non-annealed) chalcogenide films produces an irreversible shift of the band edge (photodarkening or photobleaching, depending on the alloy). Annealing near the glass-transition temperature also shifts the band-edge of as-deposited films, and typically brings the film closer to the properties of the bulk glass. Exposure of fully annealed films produces a reversible shift of the band edge, typically photodarkening irrespective of the alloy . This band edge shift (and corresponding increase in refractive index) can be erased by a subsequent anneal, and the cycle is repeatable. The maximum irreversible and reversible index change depends on the alloy. Irreversible and reversible index changes of 0.06 and 0.004, respectively, have been reported for As2Se3 .
We embedded Bragg gratings in several of the low loss waveguides discussed above. The spectral response of the gratings was probed by injecting light from an erbium doped fiber ring laser producing broadband noise-like pulses. The output light was coupled into an optical spectrum analyzer. A typical transmission scan is shown in Fig. 6 (a). Good quality gratings, with stop band extinctions >20 dB were realized. From the width of the main stop band feature, an index contrast Δn~0.004 was estimated , in good agreement with the reversible index change expected for As2Se3 . Note that the ringing features on the short wavelength side of the stop band are expected for a Gaussian-apodized Bragg grating, as produced by our writing technique. These gratings did not show good stability under illumination by room and microscope lights, typically degrading within days. This can be attributed to further photodarkening of the glass by ambient light, which washes out the grating pattern. Stability is improved when photodarkening is induced by high intensity light, an approach we hope to explore further.
As suggested originally by Poulsen et al. , the Bragg grating stop band positions can be used to characterize the modal effective indices and birefringence of waveguides. As shown in Fig. 6 (b), we were able to identify several stop bands in the output spectra of the grating-embedded waveguides. Further, by adjusting the input polarization we could identify whether each stop band was attributable to a TE or TM mode. The grating pitch could be estimated by the angle between the writing beams, and was approximately verified by SEM inspection in some cases (see Fig. 2). For the 4.0 and 4.2 µm wide waveguides, the grating angle was adjusted for a nominal grating period of 286 and 288 nm, respectively. Using these estimates and by slightly varying the grating period as a parameter, we found very good agreement between the effective indices predicted numerically (OptiBPM) and those extracted from the grating stop band positions. This agreement also provides evidence for the stability of the grating writing setup. Representative data is shown in Table 1.
We have produced low loss, highly confining rib waveguides in As2Se3 glass by employing polyamide-imide polymer as a cladding material. We also embedded strong Bragg gratings in these guides, although the gratings were not stable. Key attributes of the fabrication process are the absence of film cracking or delamination, the annealing of the chalcogenide film without thermal expansion related defects, and the good waveguide facets realized by cleaving.
The waveguide loss was as low as ~0.25 dB/cm, which is amongst the lowest values reported for integrated chalcogenide waveguides [8–10]. Further, to our knowledge the waveguides exhibit the highest modal confinement reported in chalcogenide glass. Given the promising 3rd-order nonlinearities of As2Se3 glass at 1550 nm, these waveguides have potential for integrated all-optical processing in the fiber communication band.
This work was supported by the Natural Science and Engineering Research Council of Canada, the Canadian Institute for Photonic Innovation, Canada Foundation for Innovation, and TRLabs. We would like to thank George Braybrook for capturing excellent SEM images. The devices were fabricated at the Nanofab of the University of Alberta.
References and links
1. R. Rangel-Rojo, T. Kosa, E. Hajto, P. J. S. Ewen, A. E. Owen, A. K. Kar, and B. S. Wherrett, “Near-infrared optical nonlinearities in amorphous chalcogenides,” Opt. Commun. 109, 145–150 (1994). [CrossRef]
2. G. Lenz and S. Spalter, “Chalcogenide glasses,” in Nonlinear Photonic Crystals, R. E. Slusher and B.J. Eggleton eds. (Springer-Verlag, New York, 2003).
3. J. M. Harbold, F. O. Ilday, F. W. Wise, J. S. Sanghera, V. Q. Nguyen, L. B. Shaw, and I. D. Aggarwal, “Highly nonlinear As-S-Se glasses for all-optical switching,” Opt. Lett. 27, 119–121 (2002). [CrossRef]
4. T. Cardinal, K. A. Richardson, H. Shim, A. Schulte, R. Beatty, K. Le Foulgoc, C. Meneghini, J. F. Viens, and A. Villeneuve, “Non-linear optical properties of chalcogenide glasses in the system As-S-Se,” J. Non-Crystalline Sol. 256&257, 353–360 (1999). [CrossRef]
5. R. E. Slusher, G. Lenz, J. Hodelin, J. Sanghera, L. B. Shaw, and I. D. Aggarwal, “Large Raman gain and nonlinear phase shifts in high-purity As2Se3 chalcogenide fibers,” J. Opt. Soc. Am. B 21, 1146–1155 (2004). [CrossRef]
6. K. Ogusu, H. Li, and M. Kitao, “Brillouin-gain coefficients of chalcogenide glasses,” J. Opt. Soc. Am. B 21, 1302–1304 (2004). [CrossRef]
7. S. Spalter, H. Y. Hwang, J. Zimmermann, G. Lenz, T. Katsufuji, S. W. Cheong, and R. E. Slusher, “Strong self-phase modulation in planar chalcogenide glass waveguides,” Opt. Lett. 27, 363–365 (2002). [CrossRef]
8. Y. Ruan, W. Li, R. Jarvis, N. Madsen, A. Rode, and B. Luther-Davies, “Fabrication and characterization of low loss rib chalcogenide waveguides made by dry etching”, Opt. Express 12, 5140–5145 (2004), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-12-21-5140. [CrossRef]
9. J. F. Viens, C. Meneghini, A. Villeneuve, T. V. Galstian, E. J. Knystautas, M. A. Duguay, K. A. Richardson, and T. Cardinal, “Fabrication and characterization of integrated optical waveguides in sulfide chalcogenide glasses,” J. Lightwave Technol. 17, 1184–1191 (1999). [CrossRef]
10. A. Zakery, Y. Ruan, A. V. Rode, M. Samoc, and B. Luther-Davies, “Low-loss waveguides in ultrafast laser-deposited As2S3 chalcogenide films,” J. Opt. Soc. Am. B 20, 1844–1852 (2003). [CrossRef]
11. D. J. Gibson and J. A. Harrington, “Extrusion of hollow waveguide performs with a one-dimensional photonic bandgap structure,” J. Appl. Phys. 95, 3895–3900 (2004). [CrossRef]
12. K. Kuriki, O. Shapira, S. D. Hart, G. Benoit, Y. Kuriki, J. F. Viens, M. Bayindir, J. D. Joannopoulos, and Y. Fink, “Hollow multilayer photonic bandap fibers for NIR applications,” Opt. Express 12, 1510–1517 (2004), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-12-8-1510. [CrossRef]
13. J. P. DeNeufville, S. C. Moss, and S. R. Ovshinsky, “Photostructural transformations in amorphous As2Se3 and As2S3 films,” J. Non-Crystalline Sol. 13, 191–223 (1973/74). [CrossRef]
14. M. Vlcek, S. Schroeter, J. Cech, T. Wagner, and T. Glaser, “Selective etching of chalcogenides and its application for fabrication of diffractive optical elements,” J. Non-Crystalline Sol. 326&327, 515–518 (2003). [CrossRef]
15. R. Vallee, S. Frederick, K. Asatryan, M. Fischer, and T. Galstian, “Real-time observation of Bragg grating formation in As2S3 chalcogenide ridge waveguides,” Opt. Comm. 230, 301–307 (2004). [CrossRef]
16. S. Ramachandran and S. G. Bishop, “Low loss photoinduced waveguides in rapid thermally annealed films of chalcogenide glasses,” Appl. Phys. Lett. 74, 13–15 (1999). [CrossRef]
17. R. M. Bryce, H. T. Nguyen, P. Nakeeran, T. Clement, C. J. Haugen, R. R. Tykwinski, R. G. DeCorby, and J. N. McMullin, “Polyamide-imide polymer thin films for integrated optics,” Thin Solid Films 458, 233–236 (2004). [CrossRef]
18. P. N. Kumta and S. H. Risbud, “Review: Rare-earth chalcogenides - an emerging class of optical materials,” J. Mat. Sci. 29, 1135–1158 (1994). [CrossRef]
19. R. M. Bryce, H. T. Nguyen, P. Nakeeran, R. G. DeCorby, P. K. Dwivedi, C. J. Haugen, J. N. McMullin, and S. O. Kasap, “Direct UV patterning of waveguide devices in As2Se3 thin films,” J. Vac. Sci. Tech. A 22, 1044–1047, (2004). [CrossRef]
20. C. V. Shank and R. V. Schmidt, “Optical technique for producing 0.1-µ periodic surface structures,” Appl. Phys. Lett. 23, 154–155 (1973). [CrossRef]
21. R. G. Walker, “Simple and accurate loss measurement technique for semiconductor optical waveguides,” Electron. Lett. 21, 581–583 (1985). [CrossRef]
22. L. S. Yu, Q. Z. Liu, S. A. Pappert, P. K. L. Yu, and S. S. Lau, “Laser spectral linewidth dependence on waveguide loss measurements using the Fabry-Perot method,” Appl. Phys. Lett. 64, 536–538 (1994). [CrossRef]
23. J. Buus, “Analytical approximation for the reflectivity of DH lasers,” IEEE J. Quantum Electron. 17, 2256–2267 (1981). [CrossRef]
24. A. C. van Popta, R. G. DeCorby, C. J. Haugen, T. Robinson, J. N. McMullin, and S. O. Kasap, “Photoinduced refractive index change in As2Se3 by 633 nm illumination,” Opt. Express 10, 639–644 (2002), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-10-15-639. [CrossRef]
25. T. G. Robinson, R. G. DeCorby, J. N. McMullin, C. J. Haugen, S. O. Kasap, and D. Tonchev, “Strong Bragg gratings photoinduced by 633-nm illumination in evaporated As2Se3 thin films,” Opt. Lett. 28, 459–461 (2003). [CrossRef]
26. K. Shimakawa, A. Kolobov, and S. R. Elliott, “Photoinduced effects and metastability in amorphous semiconductors and insulators,” Advances in Physics 44, 475–588, (1995). [CrossRef]
27. C. V. Poulsen, J. Hubner, T. Rasmussen, L. U. A. Anderson, and M. Kristensen, “Characterization of dispersion properties in planar waveguides using UV-induced Bragg gratings,” Electron. Lett. 31, 1437–1438 (1995). [CrossRef]