Philosophical Magazine, 1 May–1 June 2004 Vol. 84, Nos. 13–16, 1659–1666
Spectroscopic assessment of silica–titania and silica–hafnia planar waveguides R. M. Almeiday}}, A. C. Marquesy, S. Pelliz, G. C. Righiniz, A. Chiasera}, M. Mattarelliz, M. Montagna}, C. Tosello}, R. R. Gonc alvesk, H. Portalesô, S. Chaussedentyy, M. Ferrarizz and L. Zampedrizz y Departamento de Engenharia de Materiais, Instituto Superior Te´cnico, Avenida Rovisco Pais, 1049-001 Lisboa, Portugal z Istituto di Fisica Applicata ‘Nello Carrara’, Consiglio Nazionale delle Ricerche Via Panciatichi 64, I-50127 Firenze, Italy } Dipartimento di Fisica and Istituto Nazionale per la Fisica della Materia, Universita` di Trento, Via Sommarive 14, I-38050 Povo, Trento, Italy k Dipartimento di Ingegneria dei Materiali, Universita` di Trento, Via Mesiano 44, I-38050 Povo, Trento, Italy ô Dipartimento di Fisica and Istituto Nazionale per la Fisica della Materia, Universita` di Padova, Via Marzolo 8, I-35131 Padova, Italy yy Laboratoire Proprie´te´s Optiques des Mate´riaux et Applications, Unite´ Mixte de Recherche associe´e au CNRS 6136, Universite´ d’Angers, 2 boulevard Lavoisier, 49045 Angers Cedex, France zz Istituto di Fotonica e Nanotecnologie, Consiglio Nazionale delle Ricerche, Caracterizzazione e Suiluppo di Materiali per la Fotonica e l’Optoelectronica Group, Via Sommarive 14, I-38050 Povo, Trento, Italy
Abstract Silicate glasses remain the most investigated systems for optical planar waveguides, since they offer a reasonable solubility for rare-earth ions, they are transparent in the near-infrared–visible region and they are compatible with integrated optics (IO) technology. In the last decade, various technologies have been employed for the fabrication of silica (SiO2)-based IO components and a broad variety of silicate glass systems have been investigated. Besides the SiO2–titania (TiO2) system, which has been widely studied, it has recently been shown that SiO2–hafnia (HfO2) could be a further viable system for 1.5 mm applications. This paper compares spectroscopic results, in particular infrared and Raman spectra, in order to assess the structural and optical properties of erbium-activated SiO2–TiO2 and SiO2–HfO2 planar waveguides, prepared by two different techniques: rf sputtering and the sol–gel method. Particular attention is devoted to the homogeneity of the material structures obtained in each case.
}} Author for correspondence. Email:
[email protected]. Philosophical Magazine ISSN 1478–6435 print/ISSN 1478–6443 online # 2004 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/14786430310001644459
1660
R. M. Almeida et al.
} 1. Introduction Er3þ doping of waveguide materials, especially silicate glasses, remains one of the most important and thoroughly researched applications of rare-earth (RE) ions for optoelectronics. The main driving force behind the extensive work on these materials is the coincidence of the Er3þ emission band around 1530 nm with the wavelength region of choice for telecommunications, near the lowest-loss window in the absorption spectrum of silica (SiO2)-based optical fibres. SiO2-based glasses are suitable as hosts for those ions, basically because they offer a reasonable solubility for ions of about 1.1 1020 cm3; they are transparent in the near-infrared–visible region and they are compatible with integrated optics (IO) technology. Among the different binary systems, SiO2–titania (TiO2) has been extensively employed, mainly because it offers the possibility of producing planar waveguides with a controlled refractive index and a significant index contrast, depending on the TiO2-to-SiO2 ratio. Recently, it has been shown that SiO2–hafnia (HfO2) could be a further viable system for 1.5 mm applications (Gonc¸alves et al. 2002). In fact, HfO2 is transparent over a wide range of wavelengths, it exhibits a high refractive index and it allows the preparation of good-optical-quality waveguides. Several processing techniques, such as thermal oxidation, sputtering, flame hydrolysis deposition, chemical vapour deposition and sol–gel processing (Almeida 1999) have been used with success for the fabrication of SiO2-based IO components. In particular, rf sputtering has been shown to be a suitable technique for the fabrication of SiO2–TiO2 planar waveguides activated by RE3þ ions (Tosello et al. 2001, Chiasera et al. 2003b), whereas sol–gel processing is becoming one of the cheapest and most versatile methods for the fabrication of passive and active components for IO, including erbium-activated SiO2–TiO2 (Orignac et al. 1999, Almeida et al. 1999, 2003, Almeida 2002) and SiO2–HfO2 (Gonc¸alves et al. 2002, 2003) planar waveguides. In the present work, we compare spectroscopic results, in particular infrared (IR) and Raman spectra, in order to assess the structural and optical properties of erbium-activated SiO2–TiO2 and SiO2–HfO2 planar waveguides, prepared by two different techniques: rf sputtering and the sol–gel method. } 2. Experimental procedures SiO2–TiO2–ErO1.5–YbO1.5 films and waveguides were prepared by the rf sputtering technique, according to the basic procedure described in previous work (Chiasera et al. 2003a, b). The films were deposited on silicon substrates polished on one side. Sputtering deposition was performed using a 4 in (10.16 cm) SiO2 target, on which discs of TiO2, plus metallic erbium and metallic ytterbium pieces were placed. The annealing of the as-deposited films was carried out in air for 6 h at 600 C, in order to achieve good light propagation. Erbium-doped SiO2–TiO2 films and waveguides were also prepared by sol–gel processing, according to the basic procedure described in previous work (Almeida et al. 1999, Almeida 2002). Finally, erbium-doped SiO2–HfO2 films and waveguides were prepared by the sol–gel method, according to a previously described procedure (Gonc¸alves et al. 2002, 2003). Three compositions were produced, with different silicon-to-hafnium molar ratios of 70 to 30, 80 to 20 and 90 to 10. The erbium-activated silica–hafnia waveguides were deposited on vitreous silica (v-SiO2) substrates by dip coating. SiO2–HfO2 films were deposited on silicon substrates polished on one side. Table 1 lists the compositions and processing conditions.
Spectroscopy of SiO2 –TiO2 and SiO2 –HfO2 planar waveguides Table 1. Sample A B C D E F G H I J K
1661
Compositions and processing conditions of the different samples. Composition (mol%)
Processing method
Thickness (nm)
Final heat treatment
1800 2200 2800 420 450 500 600 2200 450 300
6 h at 600 C 6 h at 600 C 6 h at 600 C 5 min at 900 C 3 h 30 min at 900 C 30 h at 900 C 5 min at 900 C 15 min at 900 C 1 h at 900 C 5 min at 900 C
93 SiO2–6 TiO2–1 ErO1.5 Rf sputtering 91 SiO2–6 TiO2–1 ErO1.5–2 YbO1.5 Rf sputtering 92 SiO2–7 TiO2–1 ErO1.5 Rf sputtering 69.8 SiO2–29.9 HfO2–0.3 ErO1.5 Sol–gel 79.8 SiO2–19.9 HfO2–0.3 ErO1.5 Sol–gel 89.7 SiO2–10.0 HfO2–0.3 ErO1.5 Sol–gel 69.6 SiO2–29.9 HfO2–0.5 ErO1.5 Sol–gel 92.5 SiO2–7 TiO2–0.5 ErO1.5 Sol–gel 79.6 SiO2–19.9 TiO2–0.5 ErO1.5 Sol–gel 100 SiO2 Sol–gel 100 SiO2 Bulk glass
Fourier transform infrared (FTIR) absorption spectra were recorded with a Nicolet 5DXC spectrometer, for films deposited on silicon substrates polished on one side, in the range 400–5000 cm1, at 4 cm1 resolution, with an average of 100 scans per spectrum. TE0 mode waveguiding excitation at 457 nm, by prism coupling, was used for Raman spectroscopy measurements, with the scattered light being detected at right angles from the front of the waveguides. The Raman spectra were collected in VV (parallel) and HV (perpendicular) polarizations. The signal was analysed with a double monochromator and detected by a photon-counting system. } 3. Results and discussion 3.1. Infrared spectra The FTIR spectra of Er3þ-doped SiO2–TiO2 films prepared by rf sputtering, in figure 1, exhibit some overlapping with interference fringes, owing to their large thickness (1.8–2.8 mm), which, despite the increased sensitivity, makes their analysis more difficult. The main peaks present in the spectra, which are typical of SiO2–TiO2 glassy materials, are the dominant band at about 1070 cm1 due to asymmetric stretching of the Si–O–Si bonding sequences (transverse optical (TO) component (Almeida and Pantano 1990)), whereas the weaker peak on its low-frequency side at about 950 cm1 is due to a superposition of asymmetric stretching of oxygen atoms in Si–O–Ti bonding environments and stretching vibrations of the silanol (Si–OH) groups; the shoulder at about 1200 cm1 is related to the longitudinal optical (LO) component of the Si–O–Si asymmetric stretching (Almeida and Pantano 1990, Almeida 1992). The heat treatment of 6 h at 600 C performed on each sample appears sufficient to eliminate most OH groups, since no significant OH stretching bands (at about 3400 and about 1650 cm1) can be detected in the spectra of figure 1. Thus, neglecting the possible contribution of residual Si–OH groups, the peak at 950 cm1 is characteristic of homogeneous gel–glass network regions, whereas the approximately 1070 cm1 band arises from SiO2-rich regions and it may indicate the occurrence of phase separation (Almeida 1998, Almeida et al. 2002). The shoulder at about 1200 cm1 usually increases in the presence of porosity in incompletely densified films (Almeida and Pantano 1990, Gallardo et al. 2002).
1662
R. M. Almeida et al.
Transmittance
Transmittance (a.u.)
0.8
A B C
0.6
0.4
0.2
0.0 1600
1400
1200
1000
800
600
400
-1
Wavenumber (cm )
A: x = 1.8∝m B: x = 2.2∝m C: x = 2.8∝m
5000
4000
3000
2000
1000
-1
Wavenumber (cm ) Figure 1. IR absorption spectra of films prepared by rf sputtering and heat treated at 600 C for 6 h (a.u., arbitrary units): sample A, 93 mol% SiO2–6 mol% TiO2–1 mol% ErO1.5, thickness of about 1.8 mm; sample B, 91 mol% SiO2–6 mol% TiO2–1 mol% ErO1.5– 2 mol% YbO1.5, thickness of about 2.2 mm; sample C, 92 mol% SiO2–7 mol% TiO2– 1 mol% ErO1.5, thickness of about 2.8 mm. The inset shows a detail of the IR spectra in the 400–1600 cm1 region.
Figure 1 shows Si–O–Ti peaks at about 950 cm1 of significant intensity, compared with the approximately 1070 cm1 peak, for all the rf sputtered films, which is unusual for such a small TiO2 concentration (only about 6 mol%), especially in the absence of Si–OH groups. Moreover, a rather intense shoulder is apparent at about 1200 cm1, similar to the case of gel samples while they are still quite porous (Almeida and Pantano 1990, Gallardo et al. 2002). Also, the ratio of the shoulder (about 1200 cm1) intensity to the TO (about 1070 cm1) intensity is observed to increase with increasing thickness of the films. When the IR spectrum of a typical sputtered film is compared with that of a typical sol–gel-derived film in the same composition system, the case for samples A and I in figure 2, it is clear that the Si–O–Ti and the shoulder peaks of the former are stronger, compared with the approximately 1070 cm1 peak, indicating a higher degree of chemical homogeneity in the sputtered sample, but also suggesting the simultaneous presence of porosity (or other type of scattering centre). Figure 3 shows the FTIR spectra of Er3þ-doped SiO2–HfO2 sol–gel films, for different HfO2 contents (10, 20 and 30 mol%). These samples were essentially fully densified and no significant residual OH levels (bands at about 3400 and
1663
Transmittance (a.u.)
Spectroscopy of SiO2 –TiO2 and SiO2 –HfO2 planar waveguides
E I J A 1600
1400
1200
1000
800
600
400
-1
Wavenumber (cm ) Figure 2. IR absorption spectra of sample E (79.8 mol% SiO2–19.9 mol% HfO2– 0.3 mol% ErO1.5), sample I (79.6 mol% SiO2–19.9 mol% TiO2–0.5 mol% ErO1.5) and sample J (100 mol% SiO2), prepared by the sol–gel method and heat treated at 900 C, plus sample A (93 mol% SiO2–6 mol% TiO2–1 mol% ErO1.5), prepared by rf sputtering and heat treated at 600 C (a.u., arbitrary units).
D
Transmittance (a.u.)
E 0.
F
0.
0.
5000
4000
3000
2000
1000
Wavenumber (cm-1) Figure 3. IR absorption spectra of (100x)SiO2–xHfO2–0.3 mol% ErO1.5 films prepared by the sol–gel method and heat treated at 900 C (a.u., arbitrary units): sample D, x ¼ 29.9 mol%; sample E, x ¼ 19.9 mol%; sample F, x ¼ 10 mol%.
1664
R. M. Almeida et al.
about 1650 cm1) were detected. These samples were nevertheless much thinner than the sputtered samples (only about 300–500 nm) and therefore less sensitive to the observation of OH impurity peaks. If one compares the sample with 10 mol% HfO2 with those of figure 1, which have a TiO2 content of the same order (about 6 mol% TiO2), the major difference in the strong absorption region between 870 and 1300 cm1 consists of a much weaker peak at about 970 cm1 in the former, attributed to Si–O–Hf asymmetric stretching (Neumayer and Cartier 2001) and characteristic of homogeneous gel–glass network regions. The reason why the Si–O–Hf peak of sol–gel films is much weaker than the Si–O–Ti peak in sputtered or sol–gel-derived films may be due to a lower degree of homogeneity in the former, but it could also arise in part from a lower IR activity of the Si–O–Hf mode, due to structural differences such as different distributions of the Si–O–Hf and Si–O–Ti angles. The shoulder-to-TO-intensity ratio is also lower in the case of SiO2–HfO2 films than in the case of sputtered SiO2–TiO2 films, which might suggest that the former are better densified than the sputtered SiO2–TiO2 films. Finally, comparing the spectrum of sample E (20 mol% HfO2) with that of sample I (20 mol% TiO2) in figure 2, it is possible to observe that the dominant peak at about 1070 cm1 in the HfO2-containing sample (full width at half-maximum (FWHM), about 84 cm1) is substantially broader than the corresponding peak of the SiO2–TiO2 sol–gel film (FWHM, about 70 cm1) and that of the SiO2 film, indicating that the structure of the SiO2–HfO2 glasses is considerably more disordered, especially concerning the Si–O–Si bond angle distribution, which must be broader (Almeida and Pantano 1990, Innocenzi 2003). It should be noted that the FWHM is even larger in the case of the sputtered SiO2–TiO2 sample A (roughly about 115 cm1), probably owing to an extra degree of disorder induced by the sputtering process.
3.2. Raman spectra Figure 4 shows the VV-polarized Raman spectra of v-SiO2, SiO2–TiO2 waveguides prepared by rf sputtering and by sol–gel processing, and SiO2–HfO2 waveguides prepared by sol–gel processing. It is apparent that the level of Rayleigh scattering (stray light wing) is quite low for all samples, attesting to their good optical quality. The Raman band at about 920 cm1 in samples A and H, which was found to be depolarized, is assigned to the asymmetric stretching vibration of mixed Si–O–Ti linkages (Chiasera et al. 2003a, b) and it is rather intense for such low TiO2 contents (6–7 mol%), indicating a high degree of homogeneity of these glassy structures. The peak at about 1100 cm1, which was found to be strongly polarized, can be assigned to the LO component of the stretching mode of silicon-nonbridging oxygen (Si–O) bonds, whereas the corresponding TO component (also IR active at about 950 cm1) is probably hidden under the approximately 920 cm1 Raman peak. For the Er3þ-doped SiO2–HfO2 sol–gel sample, in addition to the low intensity of the band centred at 970 cm1, attributed to the IR-active Si–O–Hf stretching (Gonc¸alves et al. 2002, 2003), the band at about 800 cm1 almost disappeared, as well as the band at about 1100 cm1. As already inferred from their IR spectra, these films appear to have lower chemical homogeneity, unless some of these weak vibrations actually have unusually low IR and Raman activities.
Spectroscopy of SiO2 –TiO2 and SiO2 –HfO2 planar waveguides
1665
Intensity (a.u)
G A
H
K -200
0
200
400
600
800
1000
1200
Raman shift (cm-1) Figure 4. Polarized (VV) Raman spectra of sample G (69.6 mol% SiO2–29.9 mol% HfO2–0.5 mol% ErO1.5) prepared by sol–gel processing, sample A (93 mol% SiO2–6 mol% TiO2–1 mol% ErO1.5) prepared by rf sputtering and sample H (92.5 mol% SiO2–7 mol% TiO2–0.5 mol% ErO1.5) prepared by sol–gel processing (a.u., arbitrary units). Excitation was by TE0 waveguided mode at 457.9 nm. The spectrum of bulk v-SiO2 (sample K) is also shown for comparison.
} 4. Conclusions The overall optical qualities of the three types of waveguide studied were similar. However, the SiO2–TiO2 material appears to be more homogeneous, at the atomic level, than SiO2–HfO2. The sol–gel process is cheaper and usually somewhat faster than rf sputtering. Among the systems investigated in this work, the best candidates for active IO applications at the moment appear to be SiO2–TiO2 obtained by rf sputtering and SiO2–HfO2 obtained by sol–gel processing. Acknowledgements We would like to acknowledge the financial support of Fundac¸a˜o para a Cieˆncia e a Tecnologia, through the Programa Operacional ‘Cieˆncia, Tecnologia, Inovac¸a˜o’/ Cieˆncia e Tecnologia de Materiais/36109/99 project, of Fundo Social Europeu– Fundo Europeu de Desenudvimento Regional and Instituto de Cooperaca˜o Cientı´ fica e Tecnologica Internacional–Consiglio Nazionale delle Ricerche, through a collaborative grant on ‘Optical amplification’, of Consiglio Nazionale delle Ricerche–Centre Nationale de la Recherche Scientifique through a collaborative grant on ‘Improvement of the multi-target rf sputtering and sol–gel techniques for silica-based photonic component’ project, as well as that of Foreign Investment Review Board ‘Nanotecnologie, microtecnologie, sviluppo integrato di materiali’ and Ministero dell’Istruzione, dell’Universita` e della Ricerca (Programmi cofinanziati 2002) ‘Materiali nanostrutturati per l’ottica integrata’ Italian projects.
1666
Spectroscopy of SiO2 –TiO2 and SiO2 –HfO2 planar waveguides References
Almeida, R. M., 1992, Phys. Rev. B, 45, 161; 1998, J. Sol–Gel Sci. Technol., 13, 51; 1999, J. non-crystalline Solids, 259, 176; 2002, Quaderni Ottica Fotonica, 8, 81. Almeida, R. M., and Pantano, C. G., 1990, J. appl. Phys., 68, 4225. Almeida, R. M., Du, X. M., Barbier, D., and Orignac, X., 1999, J. Sol–Gel Sci. Technol., 14, 209. Almeida, R. M., Marques, A. C., and Ferrari, M., 2003, J. Sol–Gel Sci. Technol., 26, 891. Almeida, R. M., Morais, P. J., and Marques, A. C., 2002, Phil. Mag. B, 82, 707. Chiasera, A., Montagna, M., Tosello, C., Gonc alves, R. R., Chiappini, A., Ferrari, M., Zampedri, L., Pelli, S., Righini, G. C., Monteil, A., Foglietti, V., Minotti, A., Almeida, R. M., Marques, A. C., and Soares, V., 2003a, Proc. SPIE, 4990, 38. Chiasera, A., Montagna, M., Tosello, C., Pelli, S., Righini, G. C., Ferrari, M., Zampedri, L., Monteil, A., and Lazzeri, P., 2003b, Opt. Mater. (in press). Gallardo, J., Dura¤n, A., Di Martino, D., and Almeida, R. M., 2002, J. non-crystalline Solids, 298, 219. Gonc alves, R. R., Carturan, G., Montagna, M., Ferrari, M., Zampedri, L., Pelli, S., Righini, G. C., Ribeiro, S. J. L., and Messaddeq, Y., 2003, Opt. Mater. (in press). Gonc alves, R. R., Carturan, G., Zampedri, L., Ferrari, M., Montagna, M., Chiasera, A., Righini, G. C., Pelli, S., Ribeiro, S. J. L., and Messaddeq, Y., 2002, Appl. Phys. Lett., 81, 28. Innocenzi, P., 2003, J. non-crystalline Solids, 316, 309. Neumayer, D. A., and Cartier, E., 2001, J. appl. Phys., 90, 1801. Orignac, X., Barbier, D., McCarthy, O., Yeatman, E., and Almeida, R. M., 1999, Opt. Mater., 12, 1. Tosello, C., Rossi, F., Ronchin, S., Rolli, R., Righini, G. C., Pozzi, F., Pelli, S., Fossi, M., Moser, E., Montagna, M., Ferrari, M., Duverger, C., Chiappini, A., and De Bernardi, C., 2001, J. non-crystalline Solids, 284, 230.
