Quantifying Raman gain coefficients in tellurite glasses

Quantifying Raman gain coefficients in tellurite glasses

Journal of Non-Crystalline Solids 345&346 (2004) 396–401 www.elsevier.com/locate/jnoncrysol Quantifying Raman gain coefficients in tellurite glasses Cl...

437KB Sizes 0 Downloads 21 Views

Journal of Non-Crystalline Solids 345&346 (2004) 396–401 www.elsevier.com/locate/jnoncrysol

Quantifying Raman gain coefficients in tellurite glasses Clara Rivero a,*, Kathleen Richardson a,1, Robert Stegeman a, George Stegeman a, Thierry Cardinal b, Evelyne Fargin b, Michel Couzi c, Vincent Rodriguez c a

School of Optics/CREOL, Florida Photonics Center of Excellence, University of Central Florida, 4000 Central Florida Blvd., Orlando, FL 32816, USA b ICMCB (UPR 9048-CNRS) 87, Avenue du Docteur Schweitzer, 33608 Pessac cedex, France c LPCM (UMR 5803-CNRS), University of Bordeaux, I 351 Cours de la Liberation, 33405 Talence cedex, France Available online 23 September 2004

Abstract This paper presents results obtained on bulk glasses that have been fabricated and characterized for their Raman gain properties. We summarize relative values of the Raman gain coefficients of different tellurite-based glasses as compared to that of fused silica, and compare data to values estimated from their Raman cross-section data. Spontaneous Raman spectra measurements demonstrate a straight forward means to estimate Raman gain coefficients utilizing spontaneous Raman scattering cross-section data as referenced to known standards. This technique shows, for an initial set of tellurite-based bulk glass samples, excellent agreement with experimentally-obtained Raman gain coefficient data. Initial experimental measurement of Raman gain coefficients for several tellurite glasses reveal that some compositions examined exhibit an absolute Raman gain coefficient up to 30 times higher than silica, with an overall spectral bandwidth more than twice that of fused silica. Ó 2004 Elsevier B.V. All rights reserved.

1. Introduction The demands for high-speed optical communications are increasing at a tremendous rate. In order to satisfy this need for information flow to short and long distance networks, a more efficient use of available communication channels is required. This in turn, requires an expansion of the available spectral bandwidth, which in the existing communication sector is limited by the transmission losses and water absorption peak of current silica-based fibers. However, a recent breakthrough in communication was achieved with the reduction of the water absorption peak at 1400 nm, which has opened

* Corresponding author. Tel.: +1 407 823 6869; fax: +1 407 823 6880. E-mail address: [email protected] (C. Rivero). 1 On leave at Schott Glass Technologies, Duryea, PA, USA.

0022-3093/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2004.08.051

up the available communication range to span from 1270 to 1650 nm, corresponding to about 50 THz bandwidth [1]. This dramatic increase in bandwidth, and correspondingly in the number of channels available, rules out use of existing erbium-doped fiber amplifiers (EDFA), which until recently have been considered the primary means of amplification. This evolution leaves Raman gain as the main mechanism for future amplification needs. Silica fibers and germanium doped silica fibers are currently deployed in industry as Raman gain media. In fact, fused silica has been, for the past century, the key material used for long and short haul transmission of optical signals because of its good optical properties, and attractive figure of merit (i.e trade-off between Raman gain and losses). However, one of the main disadvantages of fused silica is its limited usable bandwidth for Raman amplification of about 5 THz (150 cm1). With current telecom demands, the transmission of data over larger bandwidths is becoming a crucial factor,

C. Rivero et al. / Journal of Non-Crystalline Solids 345&346 (2004) 396–401

consequently new materials which possess both large Raman gain coefficients with broad bandwidth will be required to satisfy these increasing demands. Heavy metal oxide glasses have been previously proposed for Raman gain applications due to their enhanced nonlinearity and transparency over the telecom window [2,3]. Tellurite fiber has been previously evaluated for its Raman gain performance and exhibited similar performance to that observed in our studies, however no compositional information on this glass was provided [4]. In this study we have examined a range of tellurite glasses due to their high nonlinearity as compared to other oxide glasses. Also, these glasses have been selected over other oxide families due to their high spontaneous Raman intensities and large spectral bandwidth as compared to fused silica.

2. Experimental procedures 2.1. Glass fabrication Glasses in the system TeO2–PbO–P2O5–Sb2O3 were prepared from high purity raw materials: TeO2 (Cerac 99.99%), (NH4)2 HPO4 (Merck minimum 98%), PbO (Cerac 99.99%), Sb2O3 (Cerac 99.999%). Before melting, a pre-heat treatment was conducted at 200 °C and 400 °C to eliminate the water and ammonia respectively, present in the primary starting materials. The batch mixture was melted in alumina crucibles at a temperature range of 900 °C to 1000 °C depending on the composition, for 30 min. Following melting, the glasses were quenched on to a pre-heated carbon plate, and annealed at a temperature of 40 °C below their glass transition temperature (Tg). Approximately 1 mol% of alumina attributable to the crucible material was detected in the glass using elemental dispersive spectroscopy (EDS). Finally, the glasses were cut and optically polished. The compositions examined in the study are listed in Table 1. 2.2. Measurements The volumetric weight of the different glass samples was measured by the Archimedes method in diethyl-


phtalate at room temperature (24 °C). The accuracy of the measurement is within 0.02 g/cm3. The linear absorption spectra of the different compositions were obtained using a Cary 5E (VARIAN) spectrophotometer. However, no absolute absorption measurement has been conducted in the IR region of the spectra, where the application of these glasses is expected. To ascertain these data for our glasses at key wavelengths, we are currently setting up a photothermal deflection apparatus (PDS) to accurately measure the absorption coefficients of these glasses at 1.3 lm. Results of these measurements will not be discussed here. The linear refractive index of the glasses was measured using the BrewsterÕs angle method at a wavelength of 532 nm and 1064 nm respectively. The recorded signal was fitted using Fresnel equations. The experimental error was found to be ±0.05. The IR transmittance spectra of 2 mm thick bulk glasses were recorded using a NicoletTM 740 (BRUKER) FTIR spectrometer, in the range of 2000–5000 cm1, with a spectral resolution of 4 cm1. In order to estimate the spontaneous Raman crosssection of the different tellurite-based glasses, the polarized (VV and VH) spontaneous Raman spectra of the vitreous materials were measured using a micro-Raman setup with an excitation wavelength of 514 nm. A 100X microscope objective, with a spatial resolution of about 2 lm, was used to focus the light on the front polished surface of the sample. A polarizer and quarter-wave plate (k/4) combination were used to select the polarization direction (vertical, V or horizontal, H) of the backscattered light. The scattered light was collected and spectrally analyzed with a CCD detector mounted on the exit port of a single grating spectrograph, with a typical resolution of about 6 cm1. The Rayleigh line was suppressed with a holographic notch filter.

3. Results Table 1 summarizes the values of volumetric weight (q) and linear refractive index (n) measurements for the three families of compositions studied in this paper. Note that all samples evaluated represented glasses in

Table 1 Physical properties of different tellurite-based glasses and reference materials Glass composition (mol%)

Sample code

q (g/cm3) ±0.02 g/cm3

n ± 0.05 532 nm

1064 nm

76.5TeO2–9PbO–9P2O5–5.5Sb2O3 56TeO2–20PbO–20P2O5–4Sb2O3 48TeO2–17PbO–17P2O5–18Sb2O3 85TeO2–15WO3 85TeO2–10Nb2O5–5MgO SiO2 SF6

sam sam sam sam sam sam sam

5.38 5.13 5.34 5.89 5.26 2.205 5.18

2.05 1.99 1.98 2.20 2.15 1.462 (514.5 nm) 1.82 (514.5 nm)

1.99 1.94 1.97 2.08 2.13

1 2 3 4 5 6 7


C. Rivero et al. / Journal of Non-Crystalline Solids 345&346 (2004) 396–401

the tellurite family, but varied significantly in their network constituents between phospho-tellurites (sam 1– 3), tungsten tellurites (sam 4) and niobium tellurites (sam 5). It can be seen that the density and refractive index decreases with decreasing tellurite content for samples 1 and 2; however, sample 3 shows an increase in the density and refractive index value at 1064 nm. The results obtained for sample sam 4 and sam 5 have been previously discussed [5], but their q, and n values are also summarized for completeness in the discussion. A fused silica (Suprasil) sample from Heraeus and SCHOTT SF6 glass, (provided by Schott Glass Technologies Inc.), have also been included in Table 1 for comparison purposes. The visible and IR absorption spectra of bulk samples 1–3 are shown in Fig. 1(a) and (b) respectively. Notice the blue shift in the absorption edge with decreasing tellurite concentration for compositions 1 and 2, and the apparent red shift of the optical band gap for sample 3.

Fig. 2(a) and (b) illustrate the VV and VH polarized Raman spectra of the various tellurite glasses, as well as that for fused silica and SF6. These data were used to calculate absolute Raman cross-section as described in [3]. The VH spectra were multiplied by a factor of 5 for all the compositions due to lower intensity, to allow for comparison. The absolute intensities were compared to that of SiO2, and the Raman scattering cross-section for each glass was calibrated to the absolute values of SiO2 and SF6, as reported in [3]. Fig. 3 shows the relative differential Raman cross-section values obtained for the various bulk glasses after correction for Fresnel reflection and internal solid angle as given in [6,7]. Data has been normalized to that of SiO2, for Raman data obtained for the TeO4 Raman vibration at 665 cm1. The cross-section results have been plotted against the number density (N) of tellurium atoms in the glass system. As predicted by Hellwarth in 1963, there is a relation between spontaneous and stimulated Raman scattering processes. The results of this derivation are presented in [8], where the Raman gain coefficient (g [cm/W]), also known as the gain factor, is given by:

20 -1

Absorption Coefficient α (cm )


sam 1 sam 2 sam 3


15 10 5


0 350





0 500










g ½cm=W ¼

 2  4p3 Nc2 or ; 2 2 hxS xP nS oxoX


where N is the number density of molecules, xS and xP are the Stokes (signal) and pump (laser) frequencies respectively, nS is the refractive index at the Stokes o2 r wavelength, and oxoX is the differential Raman crosssection. The differential spontaneous Raman cross-section can be obtained from the absolute measurement of the Raman scattered intensity I(xS) of a given Raman active mode, yielding the relationship described in [6]: 

 or IðxS Þ dX ¼ N V EðxL Þ dX; oX


sam 1 sam 2 sam 3


Absorption Coefficient α (cm )





0.0 2000






W avenumber (cm -1 )

Fig. 1. (a) Absorption spectra and (b) IR spectra.

where NV represents the number density of molecules in the scattered volume V, and E(xL) is the irradiance of the excitation laser beam. The relative values of the Raman cross-section as compared to SiO2 illustrated in Fig. 3, were used to calculate the relative Raman gain coefficient, at the 665 cm1 Stokes Raman peak, for the different tellurite compositions (sam 1–5). The results of this calculation are also shown in Fig. 3. The experimentally determined Raman gain spectra of compositions 4 and 5 have been reported previously in Ref. [9]. The estimated results presented in Fig. 3 for these two glasses are in very good agreement with the experimental values obtained in [9], which are also illustrated in Fig. 3 for comparison purposes. The experimental error of the Raman cross-section measurements are within ±15%.

C. Rivero et al. / Journal of Non-Crystalline Solids 345&346 (2004) 396–401 VV VH ( x5)

VV VH ( x5)

sam 5

Raman Intensity (a.u)

Raman Intensity (a.u)

sam 3

sam 2


sam 4

SiO2 (VV) SiO2 (VH) SF6 (VV) SF6 (VH)

sam 1





1000 1200 1400 -1




Wavenumber (cm )




1000 1200 1400

Wavenumber (cm-1)

Raman Cross-Section Estimated Raman gain coeff Experimental Raman gain coeff


sam 4 35

90 80

sam 5

at υ =665 cm-1




60 20

50 sam 1

40 30


sam 3 sam 2


20 5

10 0







Relative Raman Gain Coefficient

Relative Differential Raman Cross-Section

Fig. 2. Parallel (VV) and perpendicular (VH) Raman spectra.


Number Density of Te Atoms Fig. 3. Differential Raman cross-section and Raman gain coefficient at 665 cm1, relative to SiO2.

4. Discussion The three bulk glass samples within the family TeO2– PbO–P2O5–Sb2O3 were engineered such that, in the case of samples 1 and 2, the Te–Sb ion ratio is almost identical, while the Te–Pb and Te–P ratios decrease by almost three times. On the other hand, compositions 2 and 3 possess almost the same Te–Pb and Te–P ratio, and different Te–Sb ion ratio. These compositional modifications were chosen to evaluate the impact of glass former/modifier types on resulting structure and corresponding spontaneous Raman bandwidth. Recall, that the goal of our study is to both maximize Raman intensity and spectral bandwidth within a given system. From the density results shown in Table 1 we can see that, for compositions 1 and 2, as the tellurite concentra-

tion decreases, the density also decreases. However, sample 3 shows an increase in density, which is correlated to the concurrent molar increase of antimony, resulting in a corresponding increase in the total amount of heavy atoms inside the glass matrix, as compared to sample 2. Notice that the index of refraction values at 1064 nm exhibit the same behavior as the density measurements since there is a well-established correlation between the index of refraction and the density, (or number density of oscillators) given by the simple harmonic oscillator model [10]. Using this argument, there is a slight discrepancy in the index measurements at 532 nm, however the results are within the experimental error of the measurement. Fig. 1(a) illustrates the absorption spectra of compositions 1–3 respectively. As shown in the figure, the transparency window of these glasses spans the range from approximately 400 nm to 2500 nm, making them suitable for telecom applications. In fact, the key trade-off in Raman amplification applications is given by the ratio between the gain and materialÕs loss spectra. Absolute absorption measurements at 1.3 lm will be conducted in the near future using a photothermal deflection setup (PDS). It can be seen, in the inserted graph in Fig. 1(a), that there is an evident blue shift of the absorption edge for sample 2, resulting from the decrease in tellurite concentration. However, as we increase the antimony content (sam 3) with a slight decrease in Te content, there is a relative red shift of the absorption edge. This absorption edge shift we believe is due to a charge transfer from the oxygen to the metal, although it is not clear for the moment which metal, Te, Pb, or Sb, is responsible for this shift in absorption. Once again, there is a correlation between the behavior of the absorption and index of refraction measurements given by Kramers–Kronig relation [10].


C. Rivero et al. / Journal of Non-Crystalline Solids 345&346 (2004) 396–401

Fig. 1(b) illustrates the IR absorption spectra of the same three compositions previously described. As shown in the figure, sample 2 exhibits a higher absorption than sample 1 due to the increase in phosphate concentration, as illustrated by the peak around 3000 cm1. This vibration has been assigned to the (R)–OH groups participating in oxygen bonding inside the glass network, such as P–O bonds. Moreover, the vibration at about 3500 cm1 has been attributed to the stretching of (R)–OH free groups [11]. Furthermore, it can be seen that the addition of antimony decreases the IR absorption of the glass. This effect could be related to the formation of different phosphate groups, with formation of P–O–Sb bonds, for large antimony content. This assumption remains to be confirmed using reflectance IR spectroscopy. Fig. 2(a) illustrates the polarized VV and VH Raman spectra of samples 1–3 respectively. The Raman vibration at 460 cm1 is attributed to the Te–O–Te chain unit symmetric stretching mode, while the spectral features from 610 cm1 to 670 cm1 and 750 cm1 correspond to the TeO4 bi-pyramidal arrangement and the TeO3+1 (or distorted TeO4), and TeO3 trigonal pyramids structures respectively [12]. It can be clearly observed that the evolution of the TeO3+1 and TeO3 units occurs, as the TeO2 concentration decreases. Also, the VH Raman spectra show that these last mentioned entities are polarization sensitive as compared to TeO4 bi-pyramidal structures. The spontaneous Raman spectra of compositions 4, 5, SiO2, and SF6 are also shown in Fig. 2(b) for comparison purposes. Note that, as discussed previously, the TeO4 Raman vibrations increase with increasing TeO2 content. In the case of sample 4, the Raman band located at 920 cm1 is attributed to the isolated W–O short bond [13]. Finally, the VV and VH Raman spectra of SiO2 and SF6 glass are shown for comparison purposes and completeness of our analysis. The experimentally recorded values of the relative Raman intensities (counts/s) of these two last compositions are illustrated in the figure. The main SiO2 Raman vibration has been assigned to the O–Si–O symmetric stretching, while the main Raman bands of SF6 glass, at the low frequency range and at 1000 cm1, are attributed to the lead introduction in the glass network [3]. The differential Raman cross-section, relative to SiO2, was obtained from the relative intensity values obtained in the experiment, following the use of various corrections as described in Ref. [6]. Substituting these values into Eq. (1), the relative Raman gain coefficients of the five different tellurite compositions, at the 665 cm1 Raman peak, were estimated. The results are illustrated in Fig. 3. As predicted in the literature [14] and also illustrated in Fig. 2(a) and (b), the higher the TeO2 concentration, the more TeO4 units, and thus the higher the differential Raman cross-section and Raman gain coefficient respectively. These results are consistent for the

data from this study depicted in Fig. 3. The exponential behavior of the spontaneous Raman cross-section data, and consequently the Raman gain coefficient, observed in the figure can be correlated to the fact that as the TeO2 concentration increases, the relative amount of TeO3+1 and TeO3 units inside the network structure decreases, and consequently, the number of TeO4 units has to increase to maintain a constant TeO2 concentration. This result also agrees with the fact that the TeO4 bi-pyramidal arrangement possesses the highest polarizability value as compared to TeO3+1 and TeO3 structures [14] and leads to the observable index variation in the glasses. Lastly, we compared the estimated Raman gain values for samples 4 and 5 with the experimental values reported by our group previously in Ref. [9]. The estimated results presented in Fig. 3 for these two glasses are in very good agreement (within ±10%) with the experimental values obtained in [9], within the errors of the theoretical and experimental predictions. The next step will be to experimentally acquire the Raman gain spectra of samples 1–3, using the experimental setup described in [9], to observe if they too match with the Raman gain values predicted here.

5. Conclusions Based on an evaluation of a range of oxide glasses examined to date, tellurite based glasses appear to be promising candidates for Raman amplification applications. The proposed method to predict the Raman gain coefficient from spontaneous Raman cross-section estimations provides a means to estimate to the right order of magnitude, for expected Raman gain coefficients. The theoretical values obtained in the present study are in very good agreement with the experimental data obtained in Ref. [9]. The results obtained reveal that some of these compositions exhibit an absolute Raman gain coefficient of up to 30 times higher than silica, with an overall spectral bandwidth of twice of that of fused silica.

Acknowledgments This work was carried out with the support of a number of research, equipment, and educational grants, including ECS-0123484, ECS-0225930, INT-0129235, NSF IGERT grant # DGE-0114418, and NSF-CNRS # 13050. The authors acknowledge the assistance and support of all the staff at ICMCB/LPCM, and particularly to Laeticia Petit and Frederic Adamietz. Finally, special thanks to Schott Glass Technologies Inc. for providing us with the SF6 glass sample.

C. Rivero et al. / Journal of Non-Crystalline Solids 345&346 (2004) 396–401

References [1] J.J. Refi, Bell Labs Tech. J. (January-March) (1999) 246. [2] A.E. Miller, K. Nassau, K.B. Lyons, M.E. Lines, J. Non-Cryst. Solids 99 (1988) 289. [3] Z. Pan, S.H. Morgan, B.H. Long, J. Non-Cryst. Solids 185 (1995) 127. [4] A. Mori et al., Electron. Lett. 37 (24) (2001) 1442. [5] A. Berthereau, PhD Dissertation, University of Bordeaux I, Bordeaux, France, 1995. [6] Y. Kato, H. Takuma, J. Opt. Soc. Am. 61 (1971) 341. [7] F.L. Galeener, J.C. Mikkelsen Jr., R.H. Geils, W.J. Mosby, Appl. Phys. Lett. 32 (1) (1978) 34.


[8] R.W. Boyd, Nonlinear Optics, Academic, Elsevier Science, USA, 1992, p. 365. [9] R. Stegeman et al., Opt. Lett. 28 (13) (2003) 1126. [10] A. Yariv, P. Yeh, Optical Waves in Crystals, John Wiley & Sons, Inc., 1984. [11] A.M. Efimov, V.G. Pogareva, J. Non-Cryst. Solids 275 (2000) 189. [12] T. Sekiya et al., J. Non-Cryst. Solids 144 (1992) 128. [13] S. Lee et al., Electrochim. Acta 46 (2001) 1995. [14] E. Fargin et al., J. Non-Cryst. Solids 203 (1996) 96.