Erbium-activated silica–titania planar waveguides on silica-on-silicon substrates prepared by rf sputtering

Share Embed


Descrição do Produto

Journal of Non-Crystalline Solids 284 (2001) 230±236 www.elsevier.com/locate/jnoncrysol

Erbium-activated silica±titania planar waveguides on silica-on-silicon substrates prepared by rf sputtering C. Tosello a, F. Rossi a, S. Ronchin a, R. Rolli a, G.C. Righini c, F. Pozzi d, S. Pelli c, M. Fossi c, E. Moser a, M. Montagna a, M. Ferrari b,*, C. Duverger a, A. Chiappini a,b,d, C. De Bernardi d b

a Dipartimento di Fisica and INFM, Universit a di Trento, via Sommarive 14, I-38050 Povo, Trento, Italy Consiglio Nazionale delle Ricerche CeFSA, Centro di Fisica delgi Stati Aggregati, via Sommarive 18, I-38050 Povo, Trento, Italy c Optoelectronics and Photonics Department, IROE±CNR, via Panciatichi 64, 50127 Firenze, Italy d OTC-CSELT, Via G. Reiss Romoli 274, 10148 Torino, Italy

Abstract Erbium-activated silica±titania planar waveguides were prepared by radio-frequency (rf) sputtering technique. Silicaon-silicon substrates obtained by plasma-enhanced chemical vapor deposition (PECVD) and rf sputtering (RFS) were employed. The refractive indices, the thickness and the propagation losses of the waveguides were measured. The refractive index and the roughness of the silica substrates produced by RFS appear to be dependent on the thickness. Thermal annealing, which is a necessary condition to obtain light propagation, induces a decrease of the refractive index in the silica substrates. The waveguide deposited on PECVD substrate exhibits several propagating modes with an attenuation coecient 1.7 dB/cm compared with 12.2 dB/cm measured for the waveguide deposited on silica substrate produced by RFS technique. Emission of the 4 I13=2 ! 4 I15=2 transition with a 53 nm bandwidth was observed. Ó 2001 Elsevier Science B.V. All rights reserved. PACS: 42.70; 48.82; 78.60; 78.20

1. Introduction The development of integrated optical devices such as frequency converters and ampli®ers based on glassy planar waveguides activated by Er3‡ ions is underway [1±4]. Silicate-based glasses have solubilities for rare-earth ions 66  1020 cm 3 , are transparent in the near infrared (NIR)±visible region and are compatible with integrated optics * Corresponding author. Tel.: +39-461 881 684; fax: +39-461 881 680. E-mail address: [email protected] (M. Ferrari).

technology [1,4±6]. In particular, the SiO2 ±TiO2 binary system is technologically important because with the possibility of producing planar waveguides with a controlled refractive index depending on the TiO2 =SiO2 molar ratio is a good compromise between performance and cost [3,8,9]. On the other hand, the optical properties of the waveguides, and particularly the losses, are dependent on the substrate±®lm interface. The silica-on-silicon format for integrated optics is favored for a variety of functions in optical communication systems [2,3,7]. The common trend is to use silicaon-silicon substrates produced by plasma-enhanced chemical vapor deposition (PECVD)

0022-3093/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 3 0 9 3 ( 0 1 ) 0 0 4 0 7 - 0

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

technique. The latter technique is well optimized in the microelectronic business [9,10], but to have suitable substrates for integrated optics with maximum con®nement of the light, it is necessary to produce a silica bu€er with a thickness of few micrometers. Although the PECVD technique is developed and in early stages of commercialization [9,10], this kind of substrate is not immediately commercially available. For research purposes it would be interesting to optimize alternative and lower cost techniques, such as radio-frequency (rf) sputtering, to prepare silica-on-silicon substrates whose properties permit to develop planar waveguides for integrated optics. Recently, we have shown that rf sputtering (RFS) is a suitable technique for preparing titania± silica planar waveguides activated by rare-earth ions [11,12]. In this work the RFS technique is employed to prepare both the silica substrate and the waveguide.

2. Experimental The silica-on-silicon substrates, obtained by RFS, were obtained by deposition of SiO2 on silicon wafers with a deposition rate of about 1 lm/h. Substrates of di€erent thickness were obtained by changing the deposition time. The residual pressure, before deposition, was about 2  10 7 mbar. During evaporation the samples were not heated. The sputtering was carried out with an Ar gas at a pressure of 7  10 3 mbar and the applied rf power was 150 W with a re¯ected power of 18 W. The silica-on-silicon substrates prepared by PECVD technique were produced 1 with a deposition rate of 105 nm/min. On these PECVD substrates, a subsequent deposition of the erbium-activated silica±titania ®lm was performed by sputtering a target disc of silica (100 mm diameter) on which eight discs of TiO2 (10 mm diameter) and one disc of erbium oxide (8 mm diameter) were put. The same parameters were employed for the deposition of the waveguide on rf sputtered SiO2 substrates. The deposition of the 1

At the OTC-CSELT Laboratories.

231

two ®lms, silica substrate and erbium-activated silica±titania waveguide, was performed in a single run by changing the targets without opening the chamber. An erbium content of about 0.7 at.% can be estimated from previous preparation runs [12]. It should be noted that the as-prepared waveguides do not propagate the light and a thermal annealing in air for 6 h at 600°C was necessary to achieve light propagation. Scanning electron microscopy (SEM) was used to analyze the morphological properties and quality of the silica substrates. The surface of the ®lms was analyzed by an JEOL-JSM 6300 apparatus at 15 kV by covering the ®lms with a 20 nm gold layer. The refractive index and the thickness of the substrates and the waveguides were measured by an m-line apparatus. 2 Two He±Ne lasers, operating at 632.8 and 543.5 nm, were employed. The resolution in the determination of the angles synchronous to the propagation modes was 0.0075°. To measure propagation losses the light intensity scattered out of the waveguide plane, which is proportional to the guided intensity, was recorded by a video camera. The losses were evaluated by ®tting the intensity to an exponential decay function, assuming a homogeneous distribution of the scattering centers in the waveguide [13]. The measurements were performed by exciting the transverse electric TE0 mode of the waveguide with the 632.8 nm laser beam of a He±Ne laser. Photoluminescence spectroscopy, in the region of the 4 I13=2 ! 4 I15=2 transition of Er3‡ ions, was performed using the 514.5 nm line of an Ar laser as the excitation source and dispersing the luminescence light with a 270 mm single grating monochromator with a resolution of 2 nm. The light was detected using a InGaAs photodiode and a lock-in technique.

3. Results The optical parameters of some silica-on-silicon substrates, obtained by modal measurements, 2

www.metricon.com

232

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

are reported in Table 1. The as-prepared silica substrates have refractive indices >1.455. After thermal annealing in air the refractive index decreases approaching the vitreous (v)-silica index 1.4575 at 632.8 nm. We attribute this e€ect to the non-stoichiometric structure of SiOx with x < 2 of the as-prepared substrates. After annealing the number of Si±O±Si bonds increases and a structure close to that of v-silica is obtained. The nonstoichiometry is more important for the ®lms prepared by rf sputtering than for those prepared by PECVD technique, optimized to produce stoichiometric SiO2 . Furthermore, we observe that for both kind of samples a more non-stoichiometric e€ect is present in the ®lms with smaller thickness. Fig. 1 shows the refractive index dependence on the thickness for the substrates produced by RFS. Fig. 2 shows the SEM micrograph of two substrates of di€erent thickness, obtained by RFS. The optical parameters of the erbium-activated silica±titania ®lms, deposited by RFS both on the prepared silica-on-silicon and v-silica substrates, are reported in Table 2. Actually, the W1 waveguide was prepared in a single run by successively sputtering about 3 lm of silica and 2 lm of SiO2 ±TiO2 . In this case the refractive indices and the thicknesses of the two layers were determined from the m-line measurements in the dual ®lms approximation. 2 This process is the reason for the errors on the optical parameters of the W1

Fig. 1. Refractive index dependence on the thickness for the substrates produced by RFS. (a) As-prepared, (b) annealed in air at 600°C for 6 h. The lines are drawn between data symbols.

waveguide. For the other waveguides the parameters of the substrates were measured prior to deposition of the waveguiding ®lm. The TiO2 content can be determined by comparing the refractive indices of Table 2 with those measured for samples with di€erent silica±titania ratios. A TiO2 content ranging from 10 to 13 mol% is deduced [14]. Fig. 3 shows the refractive index pro®le of the W2 waveguide reconstructed from the e€ective indices at 633 nm by an inverse Wentzel±Kramers± Brillouin method [15]. From the equivalence of the refractive index pro®les obtained for TE and TM modes, it appears that the birefringence in this

Table 1 Refractive index n and thickness t obtained by modal measurements at 543.5 and 632.8 nm for TE polarization, of the silica substrates deposited by PECVD and RFS techniques as-prepared and after annealing at 600°C for 6 h Deposition technique

Deposition time

k (nm)

PECVD

11 min

PECVD

34 min

RFS

1.25 h

RFS

2h

RFS

5h

543.5 632.8 543.5 632.8 543.5 632.8 543.5 632.8 543.5 632.8

As-prepared

After annealing

n 0:0005

t (lm) 0:1

n 0:0005

t (lm) 0:1

1.4658 1.4631 1.4582 1.4551 1.4722 1.4691 1.4696 1.4661 1.4659 1.4625

1.2

1.4550 1.4530 1.4550 1.4522 1.4604 1.4590 1.4610 1.4590 1.4582 1.4550

1.2

3.6 1.4 2.2 5.3

3.4 1.4 2.0 5.2

233

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

Fig. 2. SEM micrographies of the as-prepared 1.4 lm (a) and 5.3 lm (b) thick silica ®lms deposited by RFS on silicon wafers.

Table 2 Optical parameters of the erbium-activated silica±titania waveguides deposited by RFS both on silica-on-silicon (SiO2 /Si) and v-silica (SiO2 ) substrates. ns , ts , nf and tf indicate the refractive index n and the thickness t of substrate and ®lm, respectively Substrate preparation

Number of modes at 632.8 nm Number of modes at 543.5 nm ns at 632.8 nm ns at 543.5 nm ts nf at 632.8 nm nf at 543.5 nm tf Attenuation coecient (dB/cm)

Waveguide labeling W1 RFS

W2 PECVD

W3 v-SiO2

3 3 1:47  0:01 1:47  0:01 3:0  0:1 1:53  0:01 1:54  0:01 2:2  0:1 12:2  0:1

3 4 1:4522  0:0005 1:4550  0:0005 3:4  0:1 1:4981  0:0005 1:5030  0:0005 2:6  0:1 1:7  0:1

4 4 1:4575  0:0005 1:4603  0:0005

waveguide is negligible. The other waveguides have similar properties. Fig. 4 shows the electric ®eld pro®les of the TE modes of the W2 waveguide, calculated at 633 and 1530 nm by using the parameters obtained by the m-line measurements. The refractive index at 1530 nm has been estimated by extrapolation of the green and red measurements. Fig. 5 shows the propagation losses measure for the W2 waveguide. An attenuation coecient of 1.7 dB/cm was measured at 633 nm. Fig. 6 shows the room temperature photoluminescence spectrum relative to the 4 I13=2 ! 4 I15=2 transition of Er3‡ for the W1 planar waveguide upon 514.5 nm excitation. The width of the emission band is about 53 nm.

1:508  0:001 1:513  0:001 2:6  0:1 1:8  0:1

4. Discussion The as-prepared waveguides do not propagate light and a thermal annealing in air for 6 h at 600°C was necessary to achieve light propagation. The thermal annealing performed on the silica bu€ers induces a decrease of the refractive index and a decrease of the thickness that is just greater than errors of measurement. From an application viewpoint the former e€ect is bene®cial for the production of waveguides with SiO2 as substrate, since larger the di€erence between the refractive index of the substrate and the material of the waveguiding ®lm, better the light con®nement. On the other hand, the variation of the thickness could produce cracking or microfractures of the wave-

234

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

Fig. 3. Refractive index pro®les of the W2 planar waveguide reconstructed from modal measurements at 633 nm for (a) the TE and (b) TM polarization. The e€ective indices of the TE ( ) and TM (}) modes are reported. Lines are drawn as guides for the eye.



Fig. 5. Power scattered out of the W2 waveguide as a function of the propagation length of the light streak reported in the photograph. The line is the ®t of the data to an exponential decay function.

Fig. 4. Calculated electric ®eld pro®les of the TE0 mode at 633 nm (a) and 1530 nm (b) across the layered structure, cladding of air (c), waveguide (w) and SiO2 substrate (s), of the W2 planar waveguide.

guiding ®lm after thermal annealing of the waveguide. In fact, this could be a cause of the larger losses measured for the W1 waveguide which was produced in a single run without previous annealing of the silica bu€er. In conclusion, when annealing is necessary to favor the waveguiding properties, it is necessary to perform a previous annealing on the silica-on-silicon substrate alone. The total loss of a planar waveguide consists of absorption and scattering contributions, with the latter being usually larger at the wavelengths of

Fig. 6. Room temperature photoluminescence spectrum relative to the 4 I13=2 ! 4 I15=2 transition of Er3‡ for the W1 planar waveguide, upon 514.5 nm excitation.

interest in integrated optics [16]. The scattering optical loss measured for an amorphous waveguide is the sum of two contributions: volume scattering, due to local ¯uctuation in the refractive index resulting from density and compositional variation, and surface scattering due to surface roughness [16]. A detailed discussion about the optical loss mechanism in silica±titania planar

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

waveguides deposited on silica-on-silicon substrates and produced by sol±gel route is reported in [16]. For these waveguides with 20 mol% TiO2 the main cause of losses was assigned to surface scattering with an attenuation coecient estimated at 1.2 dB/cm not so far from our best result of 1.7 dB/cm. In the present case we measured very larger losses for the W1 waveguide where both silica substrate and waveguiding ®lm are produced by RFS. On the contrary, an attenuation coecient of 1.7 dB/cm is measured for the W2 waveguide deposited on the PECVD substrate, comparable or smaller than that of the W3 waveguide deposited on v-silica. It appears evident that the substrate a€ects these waveguides although losses due to volume scattering could also be considered. The SEM images of Fig. 2 can give an explanation for the larger losses measured in the waveguides deposited on the silica-on-silicon substrate produced by RFS. It is easily seen that the ®lm surface has di€erent roughness features depending on the thickness of the silica substrate. The thin substrate (Fig. 2(a)) shows a homogeneous surface and appears practically particulatefree, but the surface of the thicker substrate (Fig. 2(b)) has important rough features with di€erent sizes and many islands on the scale of 100 to 300 nm. These results indicate that the RFS technique has to be optimized to be considered an alternative to the PECVD technique for the preparation of silica substrate suitable for integrated optics. However, we note that the RFS technique gives satisfying results in the preparation of the optically active ®lm. In fact, as shown in Fig. 3 and Table 2, planar waveguides which have many propagation modes and a single step pro®le with an uniform refractive index throughout the thickness can be produced. The modeling shown in Fig. 4 for the fundamental mode …m ˆ 0† indicates that the optical parameters of the W3 waveguide, i.e., refractive index and thickness, appear appropriate also for application in the third telecommunication window at 1.53 lm. In fact, the ratio of integrated intensity, i.e., the square of the electric ®eld, in the waveguide to the total intensity, which includes also the squared evanescent ®elds, is 0.99 and 0.92 at 632.8 and 1530 nm, respectively. This di€erence

235

means that an ecient injection at 1530 nm is possible for the produced waveguide. The emission spectrum of Fig. 6 is characteristic of the 4 I13=2 ! 4 I15=2 transition of Er3‡ . It has a main peak at 1.530 lm with a shoulder at about 1.548 lm. The spectral width of the emission band is about 53 nm and is due to inhomogeneous and homogeneous broadening plus additional Stark splitting of the excited and ground states [1,6]. This spectral width is an advantage, because optical ampli®cation is possible over a larger bandwidth. For application in wavelength division multiplexed signal ampli®cation, a better gain plateau could be obtained by codoping the silica±titania waveguide with alumina [9].

5. Conclusion Erbium-activated silica±titania planar waveguides were deposited by rf sputtering techniques on di€erent substrates such as v-silica and silicaon-silicon substrates produced by PECVD and RFS. After thermal annealing in air for 6 h at 600°C light propagation occurs in all waveguides.The waveguides deposited on PECVD substrates have an attenuation coecient of 1.7 dB/cm at 632.8 nm and represent a single step refractive index pro®le with negligible birefringence. Luminescence in the third telecom region with a large spectral width (53 nm) has been observed. The results indicate that the RFS technique can be a practicable route to prepare optically active planar waveguides in the silica-on-silicon format.

Acknowledgements This research was partially supported by the `Progetto Finalizzato MADESS II' CNR Project, a MURST-Co®n 99, and a French±Italian Program Galileo 98±99. S.P. acknowledges the grant by MADESS II for his PhD fellowship at the University of Firenze, Electronic Engineering Department. A.C. acknowledges the grant by Camera di Commercio e Turismo of Trento.

236

C. Tosello et al. / Journal of Non-Crystalline Solids 284 (2001) 230±236

References [1] Y.C. Yan, A.J. Faber, H. de Waal, P.G. Kik, A. Polman, Appl. Phys. Lett. 71 (1997) 2922. [2] M. Benatsou, B. Capoen, M. Bouazaoui, W. Tchana, J.P. Vilcot, Appl. Phys. Lett. 71 (1997) 428. [3] X. Orignac, D. Barbier, X.M. Du, R.M. Almeida, O. McCarty, E. Yeatman, Opt. Mater. 12 (1999) 1. [4] G.C. Righini, M. Bettinelli, M. Brenci, C. Duverger, M. Ferrari, M. Fossi, M. Montagna, S. Pelli, A. Speghini, SPIE 3749 (1999) 755. [5] W.J. Miniscalco, J. Lightwave Technol. 9 (1991) 234. [6] M.P. Hehlen, N.J. Cockroft, T.R. Gosnell, A.J. Bruce, Phys. Rev. B 56 (1997) 9302. [7] Y. Jaouen, L. du Mouza, D. Barbier, J.M. Delavaux, P. Bruno, IEEE Photon. Technol. Lett. 11 (1999) 1105. [8] G. Brusatin, M. Guglielmi, P. Innocenzi, A. Martucci, G. Battaglin, S. Pelli, G.C. Righini, J. Non-Cryst. Solids 220 (1997) 202.

[9] E.M. Yeatman, M.M. Ahmad, O. McCarthy, A. Vannucci, P. Gastaldo, D. Barbier, D. Mongardien, C. Moronvalle, Opt. Commun. 164 (1999) 19. [10] Y. Kuo, Vacuum 51 (1998) 741. [11] M. Ferrari, L.M. Gratton, C. Mazzoleni, M. Montagna, E. Moser, F. Tonelli, C. Tosello, S. Turrell, in: P. Vincenzini (Ed.), Innovative Light Emitting Materials, vol. 27, Techna, Faenza, 1999, p. 231. [12] C. Tosello, S. Ronchin, E. Moser, M. Montagna, P. Mazzoldi, F. Gonella, M. Ferrari, C. Duverger, R. Belli, G. Battaglin, Philos. Mag. B 79 (1999) 2103. [13] S. Pelli, G.C. Righini, in: M. Martellucci, A.N. Chester, M. Bertolotti (Eds.), Advances in Integrated Optics, Plenum, New York, 1994, p. 1. [14] M. Montagna, M. Ferrari, F. Rossi, F. Tonelli, C. Tosello, Phys. Rev. B 58 (1998) R547. [15] K.S. Chiang, J. Ligthwave Technol. LT 3 (1985) 385. [16] R.M. Almeida, P.J. Morais, H.C. Vasconcelos, SPIE 3136 (1997) 296.

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.