We present a technological approach to the realization of channeled optical waveguides, starting from reactively sputtered tellurite glass thin films, grown on silica-coated 4” Si wafers. In particular, optical lithographic process and etching recipes have been developed to overcome the solubility of TeO2 films in aqueous solutions, and to process them into high-index contrast structures with minimized post-etch roughness. Optical tests on preliminary rib waveguide geometries feature 6.3 dB/cm propagation loss for fundamental TE mode at λ=1.5 µm.
©2008 Optical Society of America
Amorphous Tellurium oxides (TeOx x~2) have been extensively studied in recent years, as promising and versatile optical materials. Since pure bulk TeOx of good amorphous quality is hardly obtainable from the melt, due to rapid crystallization during the cooling process, multi-component glasses are usually realized, whose composition is tailored to application-required characteristics. For an overview of physical and optical properties of tellurium oxide glasses and related alloys, the reader can refer to the handbook by Mallawani . In general, by use of these materials it is possible to conjugate optical transparency in the visible and infrared optical spectra with elevated values of refractive index (n~2) and low chromatic dispersion. Moreover, measured values of nonlinear third-order susceptibility χ(3), in the range of 10–12 esu, appeal to the exploitation in all-optical processing of infrared signals [2, 3]. Drawing technology for tellurite fibers traces back to early 1990s [3, 4]. Thanks to the features of TeO2-based glasses as host matrices for rare-earth ions, broadband fiber optics amplifiers operating at λ=1.5 µm, as well as at λ=1.3 µm, have been demonstrated [5–7].
A fundamental step to full technological exploitation of these materials is the feasibility of integrated optics devices; actually, several solutions to this have already been proposed in the literature. Waveguiding effects have been demonstrated in multi-component bulk glass substrates, either by the direct laser writing of buried waveguides, reaching low-contrast regimes with Δn ranging from 1×10-4 to 9×10-3 [8–10], or by ion-exchange methods, which reach Δn/n up to 5 %  in slab waveguide geometry. A different approach, based on the fiber-on-glass method, has been proposed to obtain channeled waveguides . However, low Δn core-cladding contrast is an intrinsic limitation of technological approaches based on bulk glass substrates. On the contrary, tellurite glasses can be potentially exploited also to implement high-contrast integrated optics, what enables a consistent reduction in waveguide dimensions towards a large scale of integration. Besides, the bulk geometry strongly hinders the possibility to integrate guided optics with other functional elements onto a common technological platform. In perspective, these drawbacks can be both overcome by different approaches, based on the growth of thin tellurite films onto suitable substrates. An example in this direction has recently been suggested by Caricato et al.  and configures a slab waveguide, realized by reactive pulsed laser deposition of a thin film of Er-doped tellurite glass onto silica substrate.
On the other hand, reactive sputtering has been demonstrated as a viable approach to the growth of amorphous tellurite thin films onto silica and silicon large area substrates [14–16], also enabling the tuning of stoichiometry and optical properties. Multimode optical waveguiding at λ=633 nm for a TeO2 slab on Corning glass substrate, featuring propagation loss as low as 0.26 dB/cm was also demonstrated .
In this work we introduce a technological process for the realization of TeO2-based, high refractive index contrast channeled optical waveguides. The starting point is the deposition of an amorphous glass film on silica-coated silicon substrates, by radiofrequency reactive sputtering, as described in Section 2; structural, morphological and optical characterizations are also presented, that assess the quality of films in view of applications. Section 3 is devoted to the description of the design criteria and process steps to the realization of the optical structures. Examples of rib waveguides have been demonstrated and characterized and experimental results are discussed in view of effective applications.
2. The material: growth and characterization
Reactive sputtering from Tellurium targets have proved to be one of the effective techniques to grow pure TeOx glass thin films on substrates [15–18]. In particular, sputtering is effective in tuning the oxygen molar fraction, from sub-stoichiometric O/Te ratio to oxygen-enriched films. This in turn is a degree of freedom in the definition of optical properties like refractive index and absorption coefficient [15, 18].
The technological process introduced in this work, towards the definition of integrated optics elements, starts with the deposition of TeOx (x~2) by radio-frequency reactive sputtering from a TeO2 sintered target. The choice for quasi-stoichiometric composition of the films stems from the request to keep low absorption coefficient in the near infrared, conjugated to elevated refractive index and nonlinear response [14, 15]. The sputtering system (VS80, Kenosystec), is equipped with a loadlock station and pumped to a standard base pressure of 7×10-7 mbar. Sputtering of tellurium oxide films has been performed on 4” (100) Si wafers, both directly on Si surface and onto a SiO2 cladding layer grown by Plasma Enhanced Physical Vapor Deposition. The surface of the TeO2 target (circular, 4”, purity 99,999%) was cleaned by a presputtering step for about 15 min before deposition. Controlled Ar and O2 flows (purity 99,9999) were introduced in the growth chamber, the Ar flow being set at Ar=50sccm and the oxygen flow calibrated in order to compensate for O2 loss during the sputtering process and for stoichiometry tuning. Film deposition parameters are summarized in Table 1.
Compositional characterization of the films (EDAX system -Genesis 4000 XMS) excluded the presence of detectable impurities. Rutherford Back Scattering analysis (RBS) (estimating O/Te ratio within 1%) confirmed homogeneity of all deposited films and showed no variation of Te/O ratio in depth. RBS results have been used to calibrate the oxygen composition of the reactive atmosphere in the sputtering deposition process, leading to controlled stochiometric ratios as reported in Table 2.
Films between 300nm and 2000nm thick were sputtered and morphologically characterized. Good adhesion to the substrate and good quality of the film-substrate interface are evident for films grown both on Si and SiO2-coated wafers, from scanning electron microscopy analysis (SEM - LEO 1525). All samples show a densely packed fine grain structure, without the formation of columnar grains. Surface roughness of as-grown films was evaluated by AFM topography (AFM PSIA XE 150) with indipendent Z scanner, using high aspect ratio tips operating in non-contact mode. A mean roughness value of about 3 nm and standard deviation Rq=0.4 nm÷0.5 nm was measured on 1 µm2 areas, for 1100nm - thick film samples grown on silicon.
The amorphous quality of room-temperature sputtered films on both substrates was assessed by X-Ray Diffraction (Bruker AXS).
The dispersion of real and imaginary parts of the refractive index, respectively n and k, of sputtered film samples at different compositional ratio (1.9<x<2.15) has been characterized by variable angle spectroscopic ellipsometry  in the 260nm÷1700nm range, with a spectral resolution of 0.03 nm. Experimental results reported in Fig. 1 have been obtained using a Cauchy fitting model for the chromatic dispersion of n, as summarized in Table 3. The films have been assumed to be single layers of constant composition, as suggested by RBS, and the first guess for film thickness was estimated by SEM inspection.
Negligible absorption in the third telecommunication optical window is detected, while a 1.5 % spread in n is measured, as the O/Te ratio is varied. More specifically, n values at λ=1550 nm are plotted in the inset of Fig. 1(b). Present results substantially correspond, at same O/Te ratios, to those reported in Ref. 15 and referred to films sputtered from metallic Te target. This implies that the nature of the target, whether metallic or oxide, has negligible impact on the optical properties of the deposited amorphous thin film, once the same final stoichiometry and structure are guaranteed by respective process steps.
3. Realization of tellurite rib waveguides
Optical materials with elevated index of refraction can in principle be exploited to realize the core of high index contrast optical waveguides. The subwavelength cross-sections and micron-range bending radii of the wire-waveguides , obtainable at core-cladding index contrasts higher than 15%, appeal to optical large scale of integration, but are particularly challenging in terms of required high resolution lithography, minimization of surface and sidewall roughness and fiber-to-waveguide coupling. If the minimization of bending radius is not an aim in itself, an alternative geometry is represented by thick slab rib-waveguides. By suitable design of the rib, single mode conditions can still be guaranteed even in the presence of high refractive index in the core and large waveguide cross-sections . Moreover, the mode field diameter allows efficient coupling without recurring to mode converters and tapered/lensed fibers .
Tellurium oxide, featuring an index contrast over 48% with respect to SiO2 and about 100% with respect to air, can be selected as the core material in high-contrast geometries where either SiO2 or air are the cladding materials. Rib waveguides can be obtained by UV lithography and dry etching of the TeO2 core. The technological process flow is sketched in Fig. 2.
The starting stacked structure is substantially described in Fig. 2(a). A standard SiO2 film (about 3 µm thick), deposited by Plasma Enhanced Chemical Vapor Deposition (PECVD) onto the Si wafer, has been used as the lower cladding. The TeO2 film has been sputtered onto it and covered by a silica cap-layer, sputtered from a rectangular 12”x2” SiO2 target, (RF Power 800W, Ar 50sccm, Pressure 1.1×10-2 mbar); typical thickness of the cap-layer measures 50nm÷120nm, the exact value being dependent on the etch depth of the rib structure and on the dry etching selectivity of silica over tellurite being equal to 1:5. The need for a silica cap-layer can be explained when considering that a standard optical UV lithography technique is used, which employs a broadband mask-aligner (MA6 Karl-Suss, working at 280nm÷450nm), standard resists and alkaline developers. However, sputtered amorphous TeOx films have proven to be soluble both in acid and alkaline aqueous solutions, what therefore represents a critical issue, because the developer would also dissolve the waveguide core. The SiO2 cap-layer protects tellurium oxide films during lithography. Photoresist (UV6 - Rohm and Haas) is spun on top of the cap-layer and the masked stack is patterned in the mask-aligner, as in Fig. 2(b)–(c). After mask removal, the resist is developed while the cap-layer shields the tellurite film, as shown in Fig. 2(d). The SiO2 cap layer is etched by a Reactive Ion Etching (RIE) step in fluorine chemistry, optimized by tuning the reactive plasma characteristics through acting on radio-frequency source power, feed gas composition and pressure. An example of recipe and related parameters are reported in Table 4. The endpoint of RIE step has been checked by SEM. The resulting structure is shown in Fig. 2(e). A wet etching step in boiling acetone is used to remove the resist without etching the tellurite film. After the latter process step, a patterned silica hardmask is left, as sketched in Fig. 2(f), which is eventually used for the dry physical etching in Ar plasma of the TeO2 film, down to the selected depth, to obtain the rib geometry depicted in Fig. 2(g). An example of related recipe is reported in the third column of Table 4.
The choice for a non-reactive dry etching solution for the definition of the rib in tellurite glass is motivated by the need to minimize post-etch surface roughness. Figs. 3(a) and 3(b) report the results of performing, respectively, a dry physical etching of sputtered tellurite films, compared to what issued by reactive ion etching in fluorine chemistry. In case of fluorine-based RIE, for a depth of about 100 nm, a peak-to-peak roughness of entity comparable to the etched depth is originated, what would introduce excessive scattering loss to optical waveguiding. Actually, even if some volatile reaction products like Tellurium esafluoride (TeF6) and Tellurium tetrafluoride (TeF4) are expected to be formed, additional non volatile reaction products may be generated, eventually causing micromasking effects and roughness. By performing non-reactive dry etching at low RF power and cathode temperature, the smoothness of surface is strongly improved with respect to the reactive approach. The sample shown in Fig. 3(b) has been etched in inert gas plasma, without generating any plumes and resulting roughness is negligible, if compared to the situation depicted in Fig. 3(a). By the same recipe, etch depths down to 300 nm have been demonstrated, with similar roughness quality.
Waveguide chips were prepared by cleavage, without any additional polishing step. Examples of as-cleaved chips are shown in Fig. 4. The somewhat rough quality of the tellurite rib edge in Fig. 4(b) can be ascribed to a stressed condition of the silica under-cladding. The parameters of the structure shown in Fig. 4(a) are listed in Table 5. The slab fractional thickness is defined as the ratio between the etched depth and the overall thickness of the core layer. In the present case it corresponds to an etch depth of the TeO2 core by 27nm, in addition to the full etching of the cap layer.
Optical propagation at λ=1.5µm in tellurite rib waveguides has been tested on 7 mm-long specimen. The optical transmission, spectral response and insertion losses have been measured over a wavelength spectrum λ=1520nm÷1620nm and for both TE and TM polarization states. The optical set up is depicted in Fig. 5, and based on a tunable laser source (Pout=5dBm) and an optical spectrum analyzer. Waveguides were placed on a vacuum holder and thermally stabilized; launch polarization was selected through a polarization controller. For a most efficient fiber-to-waveguide coupling, small-core fibers with a Mode Field Diameter (MFD) equal to 4±0.3 µm have been used and aligned by use of feedbacked nanopositioning stages (5nm of spatial resolution in closed-loop). The modal distributions have been recorded by substituting the fiber output stage with an infrared camera equipped with 100X objective.
Mode profiles and coupling efficiency have been simulated with the aid of a Beam Propagation Method (BPM). Waveguide geometries as in Fig. 4 are multi-modal. Examples of calculated distributions for the fundamental TE and TM modes are reported in Figs. 6(a) and 6(b) respectively, and compared to corresponding measured far-field patterns.
Figure 7(a) shows the BPM simulated fiber-to-waveguide butt-coupling efficiency at λ=1570nm, as a function of the fiber MFD, in case of the rib described in Fig. 5(a) and Table 5. The best coupling condition is achieved in case of small-core fibers and is characterized by estimated coupling losses of about 3.4 dB and 1.8 dB, in case respectively of TE and TM polarization; this result is substantially wavelength insensitive, as less than 0.1 dB of variation is numerically estimated on the whole spectral range 1520 nm÷1620 nm. In order to evaluate propagation penalties, waveguides have been input and output-coupled to small-core fibers and total insertion loss has been measured. By assuming equal coupling losses at both waveguide facets, by neglecting additional loss contribution due to chip edge roughness and by subtracting the simulated values from measured ones, 6.3 dB/cm and 11.4 dB/cm are respectively estimated as transmission loss for the TE and TM modes, as shown in Fig. 7(b). The discrepancy can be justified when noticing that the TM mode is more sensitive to the presence of film interfaces than the TE mode, as in Fig. 6, and is therefore expected to undergo higher loss by scattering due to waveguide roughness. On the other hand, the high amount of loss is probably due to the presence of higher order leaky modes that might be excited by waveguides non-idealities such as sidewall roughness or by a slight launch tilt. The excitation of such higher modes would contribute to enhance both propagation loss and output waveguide-to-fiber coupling loss; this last issue would make our propagation loss values slightly over-estimated.
Besides, by reducing the rib width and by increasing the etching depth, a true single-mode condition is eventually reached, when only fundamental TE and TM modes propagate. The technological feasibility of such a deep etching step is demonstrated in Fig. 4(a). By BPM design, a rib width a=1.7 µm and an etch depth of 220 nm would result in single mode propagation. For such a waveguide, less than 3 dB loss is expected for direct coupling with standard small-core fibers and a bending radius smaller than 0.5 mm for TM polarization (larger than 1 mm for TE polarization) is allowed, if a bending loss equal to 10-2 dB/Rad is taken as a reference.
A technological path to the realization of high index contrast planar optical structures made of tellurite glass thin films has been demonstrated. Pure TeOx (x≈2) sputtered amorphous thin films of sub-nanometer surface roughness, homogeneous structure and controlled optical properties have been deposited onto 4” silica-coated Si wafers. A suitable dry etching process has been developed, to pattern the tellurite glass films while avoiding the introduction of unacceptable amount of roughness and preventing the dissolution of sputtered TeO2. The process involves the use of low RF powers and etching temperatures and exploits a silica protective cap-layer, which also serves as a hardmask for the lithographic steps. Preliminary examples of rib waveguide structures have been reported, experimentally tested for optical propagation at λ=1.5 µm. Less that 4 dB coupling loss are estimated with small-core fibers and 5.7 dB/cm propagation loss have been experimentally derived in case of fundamental TE mode. The realization of sub-µm single-mode buried strip waveguides in tellurite-silica (air) systems could also evolve from the process approach presently described. Results here presented can be considered as a step in the direction of integrating tellurite-based optics onto silicon platforms for more complex optical subsystems.
Authors acknowledge Dr. Angel Muñoz Martìn, from Centro de Micro-Anàlisis de Materiales, Madrid, Spain, for performing RBS characterizations of samples and Dr. Marco Arimondi, from Pirelli Labs, Milano, Italy, for the XRD characterization.
References and links
1. R. A. H. El-MallawanyTellurite Glasses (CRC Press, Boca Raton, Florida, 2002).
2. M. E. Lines, “Oxide glasses for fast photonic switching: A comparative study,” J. Appl. Phys. 69, 6876–6884 (1991). [CrossRef]
3. H. Bookey, K. Bindra, A. Kar, and B. Wherrett, “Telluride glass fibres for all optical switching: - nonlinear optical properties and fibre characterization,” in Proceedings of the 2002 IEEE/LEOS Workshop on Fibre and Passive Optical Components, (IEEE, 2002) pp.29–34. [CrossRef]
4. J. S. Wang, E.M. Vogel, and E. Snitzer, “Tellurite glass: a new candidate for fiber devices,” Opt. Mat. 3, 187–203 (1994). [CrossRef]
5. S. Q. Man, E. Y. B. Pun, and P. S. Chung, “Tellurite glasses for 1.3 µm optical amplifiers,” Opt. Commun. 168, 369–373 (1999). [CrossRef]
6. A. Shaoxing-Xhen, Jha Xiaobo-Liu, M. Naftaly, K. Bindra, H. J. Bookey, and A. K. Kar, “Tellurite glasses for broadband amplifiers and integrated optics,” J. Am. Ceram. Soc. 85, 1391–1395 (2002). [CrossRef]
7. K. Tikhomirov, A. B. Seddon, D. Furniss, and M. Ferrari, “Intrinsic Defects and Glass Stability in Er3+-Doped TeO2 Glasses and the Implications for Er3+-Doped Tellurite Fiber Amplifiers,” J. Non-Cryst. Sol. 326–327, 296–300 (2003). [CrossRef]
8. E. Chierici, M. C. Didavide, A. Moro, O. Rossotto, L. Tallone, and E. Monchiero “Direct writing of waveguide on a tellurite glass using a focused ultraviolet laser beam,” in Proceedings of IEEE/LEOS Workshop on Fibre and Optical Passive Components, (IEEE, 2002), pp. 29–33.
9. Y. Tokuda, M. Saito, M. Takahashi, K. Yamada, W. Watanabe, K. Itoh, and T. Yoko “Waveguide formation in niobium tellurite glasses by pico- and femtosecond laser pulses,” J. Non-Cryst. Sol. 326–327, 472–475 (2003). [CrossRef]
10. P. Nandi, G. Jose, C. Jayakrishnan, S. Debbarma, K. Chalapathi, K. Alti, A. K. Dharmadhikari, J. A. Dharmadhikari, and D. Mathur, “Femtosecond laser written channel waveguides in tellurite glass,” Opt. Express 14, 12145–12150 (2006). [CrossRef] [PubMed]
11. S. Berneschi, G. Nunzi Conti, I. Banyasz, A. Watterich, N. Q. Khanh, M. Fried, F. Paszti, M. Brenci, S. Pelli, and G. C. Righini “Ion beam irradiated channel waveguides in Er3+-doped tellurite glass,” Appl. Phys. Lett. 90, 121136–121138 (2007). [CrossRef]
12. V. A. G. Rivera, E. Rodriguez, E. F. Chillcce, C. L. Cesar, and L. C. Barbosa, “Waveguide produced by fiber on glass method using Er3+-doped tellurite glass,” J. Non-Cryst. Sol. 353, 339–343 (2007). [CrossRef]
13. A. P. Caricato, M. Fernández, M. Ferrari, G. Leggieri, M. Martino, M. Mattarelli, M. Montagna, V. Resta, L. Zampedri, R. M. Almeida, M. C. Conçalves, L. Fortes, and L. F. Santos, “Er3+-doped tellurite waveguides deposited by excimer laser ablation,” Mat. Sci. Eng. B 105, 65–69 (2003). [CrossRef]
14. F. D’Amore, M. Di Giulio, S. M. Pietralunga, A. Zappettini, L. Nasi, V. Rigato, and M. Martinelli, “Sputtered stoichiometric TeO2 films: dispersion of linear and nonlinear optical properties,” J. Appl. Phys. 94, 1654–1661 (2003). [CrossRef]
16. R. Nayak, V. Gupta, A. L. Dawar, and K. Sreenivas, “Optical waveguiding in amorphous tellurium oxide thin films,” Thin Solid Films 445, 118–126 (2003). [CrossRef]
17. M. F. Al-Kuhaili, S. M. A. Durrani, E. E. Khawaja, and J. Shirokoff, “Effects of preparation conditions on the optical properties of thin films of tellurium oxide,” J. Phys. D: Appl. Phys. 35, 910–915 (2002). [CrossRef]
18. N. Dewan, V. Gupta, K. Sreenicas, and R. S. Katiyar, “Growth of amorphous TeOx (2<x<3) thin film by radio frequency sputtering,” J. Appl. Phys. 101, 084910–084916 (2007). [CrossRef]
19. R. M. A. Azzam and N. M. Bashara, Ellipsometry and Polarized Light (Elsevier, Amsterdam, 1992).
20. A. Melloni, R. Costa, G. Cusmai, F. Morichetti, and M. Martinelli, “Waveguide index contrast: implications for passive integrated optical components,” in Proceedings of IEEE/LEOS Workshop Fibres and Optical Passive Components, (IEEE, 2005) pp. 246–253.
21. R. A. Soref, J. Schmidtchen, and K. Petermann, “Large single-mode rib waveguides in GeSi-Si and Si-on-SiO2,” IEEE J. Quantum Electron. 27, 1971–1974 (1991). [CrossRef]
22. R. Costa, G. Cusmai, A. Melloni, and M. Martinelli, “TE-TM coupling of a standard fiber to a Si-wire waveguide,” in Proc. of 13th European Conference on Integrated Optics, 25–27 Apr. 2007, Copenhagen, Denmark, paper ThC3.