Raman gain of selected tellurite glasses for IR fibre lasers calculated from spontaneous scattering spectra

Raman gain of selected tellurite glasses for IR fibre lasers calculated from spontaneous scattering spectra

Available online at www.sciencedirect.com Optical Materials 30 (2008) 946–951 www.elsevier.com/locate/optmat Raman gain of selected tellurite glasse...

494KB Sizes 1 Downloads 27 Views

Available online at www.sciencedirect.com

Optical Materials 30 (2008) 946–951 www.elsevier.com/locate/optmat

Raman gain of selected tellurite glasses for IR fibre lasers calculated from spontaneous scattering spectra M.D. O’Donnell a, K. Richardson a,*, R. Stolen a, C. Rivero b, T. Cardinal c, M. Couzi c, D. Furniss d, A.B. Seddon d a

d

Advanced Materials Research Laboratory (AMRL), School of Materials Science and Engineering, Clemson University, 91 Technology Drive, Anderson, SC 29625, USA b College of Optics and Photonics: CREOL and FPCE, Building 53, University of Central Florida, 4000 Central Florida Blvd., Orlando, FL 32816, USA c ICMCB-CNRS, University of Bordeaux, Pessac, France Novel Photonic Glasses Research Group, Wolfson Centre for Materials Research, School of Mechanical, Manufacturing and Materials Engineering, University Park, University of Nottingham, Nottingham, Nottinghamshire NG7 2RD, UK Received 31 August 2006; received in revised form 4 May 2007; accepted 5 May 2007 Available online 28 June 2007

Abstract In this paper, we present the spontaneous Raman scattering spectra and calculated Raman gain spectra of two TZN (TeO2–ZnO– Na2O) glasses and three tungsten tellurite glasses. Addition of lead(II) oxide to the TZN glass increased the amount of lower coordination [TeO3]/[TeO3+1] units (765 cm1) in the glass, and decreased the higher coordination [TeO4] units (665 cm1) and Te–O–Te chains (440 cm1). Addition of WO3 to the tungsten–tellurite glasses also resulted in the same trend as with PbO, and an additional band at around 925 cm1 was seen to increase in intensity due to [WO4+2] units. Finally, a band at around 370 cm1 was seen in the bismuth-doped tungsten tellurite glass, due to Te–O–Bi linkages. The calculated Raman gain of these tellurite glasses were found to be 20–30 times that of fused-silica (0.89 · 1013 m W1). The calculated Raman gain of the PbO-doped TZN glass also showed good agreement with direct gain measurements previously made at 1064 nm. The minimum laser powers required to stimulate Raman amplification were calculated for one TZN glass and one tungsten–tellurite glass for optical fibre with a 10 lm core. The power densities required were of the order of MW cm2 for fibre with 2–3 dB m1 loss at 1550 nm and much lower than the surface optical damage thresholds of the glasses which are of the order of GW cm2.  2007 Elsevier B.V. All rights reserved. Keywords: Raman gain; Glass; Tellurite; Infrared; Laser

1. Introduction There are two significant advantages of Raman fibre amplifiers over erbium-doped fibre amplifiers (EDFAs): a broad amplification bandwidth (>100 nm) and the maximum gain can be tuned throughout the transparency window of the material by varying the pump wavelength, to be compared with the fixed wavelength based on a laser tran*

Corresponding author. Tel.: +1 864 656 0549; fax: +1 864 656 5973. E-mail address: [email protected] (K. Richardson).

0925-3467/$ - see front matter  2007 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2007.05.010

sition in a lanthanide-doped amplifier. Therefore, amplification can be utilised in any of the telecommunication windows or mid-IR, depending on application and the use of the amplifier [5]. Tellurite glasses have received much attention recently in the literature due to their highly non-linear optical properties compared to fused-silica. The peak Raman response of tellurite glasses has been shown to be over 50 times that of fused-silica, with the height (peak gain) and breadth (bandwidth) easily tunable with glass composition by the addition of oxide and halide network modifiers and

M.D. O’Donnell et al. / Optical Materials 30 (2008) 946–951

intermediates [6]. Tellurite glasses also exhibit superior laser damage resistance compared to other high gain materials such as chalcogenides, a feature important for applications which require high intensity pump lasers to induce non-linear effects in the material [7]. Raman amplifiers incorporating tellurite-based glass optical fibre have also received much interest lately due to the work of Mori and coworkers at NTT in Japan [1–4]. This group demonstrated gain of over 10 dB in the mid-IR over 250 m of fibre, amplifying with a wide bandwidth (160 nm) [4]. No compositional details of the glass fibres were given in these studies. The purpose of this study was to use spontaneous Raman scattering spectroscopy to understand the structure of the tellurite glasses, and accurately calculate the Raman gain response of candidate optical fibre drawing compositions. The laser powers required to stimulate the Raman amplification process were also evaluated, to ensure these were well below the surface optical damage thresholds of the materials. 2. Experimental Glasses were melted according to small-scale glass melting procedures described previously [7–9]. 2.1. Spontaneous raman scattering measurements Spontaneous Raman scattering spectra were taken for the tellurite glasses in VV and VH polarisation using a micro-Raman setup. The excitation wavelength used was 1064 nm (from a Nd:YAG laser), with a power of 30 mW. Spectra of the tellurite glasses were collected for 10 s and the standards (fused-silica and Schott SF6) for 120 s due to their lower Raman signals. The incoming polarised laser beam was focused onto the front polished surface of the sample via a 100· microscope objective, with a spatial resolution of about 2 lm. A polariser and quarterwave plate (k/4) combination were used to select the polarisation direction (vertical, V or horizontal, H) of the scattered light. A backscattering geometry was used to collect the Raman signal, which, in turn was spectrally analysed with a spectrometer and a CCD detector. The Rayleigh line was suppressed with a holographic notch filter. For these near-infrared experiments, an InGaAs array from Jobin– Yvon was employed. This CCD was cooled to liquid nitrogen temperature and was made of 512 pixels, each 50 lm wide. 2.2. Calculation of raman gain from VV spontaneous raman scattering spectra Spontaneous Raman scattering spectra were obtained as described in the previous section. The Raman gain of selected glasses was calculated from the spontaneous spectra as reported earlier [10,11]. The Raman gain coefficient of a glass, gR (m W1), can be calculated from spontaneous

947

Raman scattering spectra in VV (vertically polarised) scattering geometry after some corrections detailed below and by scaling the spectra to a well characterized standard (e.g. fused-silica or Schott SF6 glass). By obtaining the differential spontaneous Raman scattering cross-section (dr/dX) from the spontaneous Raman scattering spectra, the Raman gain spectra can be calculated. The ratio of the number of Stokes photons scattered spontaneously into a solid angle dX is given by [12] " # dr x3s xp 2 ¼ hjaR j i; 2 dX ð4pe0 c2 Þ

ð1Þ

where aR = Raman polarisability, e0 = permittivity of free space (8.8542 · 1012 F m1), xs = Stokes frequency, xp = pump frequency, c is the speed of light (2.998 · 108 m s1). Therefore, the Raman gain, gR, can be expressed as   8p2 c2 NT 1 dr 2 gR ¼ 2 2 2 hns np xs xp D þ T 2 dX   8p2 c2 dr  NT 2 ; dX hns np x2s xp

ð2Þ

where N = number density of molecules, h = h/2p, h is Plank’s constant (6.626 · 1034 J s), ns = refractive index at the Stokes frequency, np = refractive index at the pump frequency, D = optical frequency detuning from resonance and T2 = transverse relaxation time for homogeneous transition (at resonance DT2  1). To summarise, first the spectra were normalised for collection time and to the peak Raman intensity of fused-silica (440 cm1). To correct for the reflection loss and variation in the internal solid angle, the spectra relative to fused-silica must be multiplied by FR-SO, given by F R-SO ¼

ð1 þ nsam Þ

4

ð1 þ nSiO2 Þ

4

ð3Þ

;

where nsam is the refractive index of the sample at the pump wavelength and nSiO2 is the refractive index of fused-silica (1.46). This assumes dispersion is low at the pump, therefore ideally Raman excitation should be performed away from the electronic bandgap of the material. The spectral intensity should also be divided by the Bose–Einstein correction factor, FBE, to eliminate low wavenumber thermal effects shown by 



ht F BE ðt; T Þ ¼ 1 þ exp kT



1 1

;

ð4Þ

where t is the frequency of the Raman shift relative to the pump, k the Boltzmann constant (1.381 · 1023 J K1) and T the temperature of the Raman experiment (298 K for the spectra presented here). Finally by performing the follow-

948

M.D. O’Donnell et al. / Optical Materials 30 (2008) 946–951

ing calculation, the Raman gain of the sample, gRs , can be approximated  2 nSiO2  gRSiO ; ð5Þ gRs ¼ I corr  2 nsam where Icorr is the corrected Raman intensity (normalised to the reference material and applying the correction factors above) and gRSiO the peak Raman gain of fused-silica 2 (0.89 · 1013 m W1) for 1064 nm pumping and a Stokes 1 shift of 440 cm . 3. Results Table 1 summarises the glass compositions analysed here, and the corresponding sampled IDs with peak Raman gain values of silica and for comparison. Fig. 2. Spontaneous Raman scattering spectra of tungsten–tellurite glasses in VV and VH polarisations.

3.1. Spontaneous raman scattering spectra Fig. 1 shows the spontaneous Raman scattering spectra of the TZN glasses T1 and T2 in VV and VH polarisations. The spectra exhibit a low wavenumber tail, which is a thermal effect and will be eliminated from the Raman gain spectra using the Bose–Einstein correction (Eq. (4)). Three bands can be seen in the spectra at around 440, 665 and 765 cm1. The intensities of the bands at 440 and Table 1 Glass compositions and calculated peak Raman gain values with fusedsilica and ZBLAN for comparison [17] Sample ID

Composition (mol.%)

gR ( · 1013m W1)

T1 T2 T5 T6 T7 ZBLAN Fused–silica

80TeO2–10Na2O–10ZnO 77TeO2–10Na2O–10ZnO–3PbO 90TeO2–5WO3–5Nb2O5 82.5TeO2–7.5WO3–10Nb2O5 70TeO2–25WO3–5Bi2O3 ZrF4–BaF2–LaF3–AlF3–NaF SiO2

19.9 ± 1.0 17.9 ± 0.9 29.7 ± 1.5 27.0 ± 1.4 23.9 ± 1.2 2.0 0.9

Fig. 1. Spontaneous Raman scattering spectra of TZN glasses in VV and VH polarisations.

665 cm1 decrease with PbO addition (T2) compared to the base glass (T1). The assignments of the band and how they relate to the underlying glass structure will be discussed in the next section. Fig. 2 displays the spontaneous Raman scattering spectra of tungsten–tellurite glasses T5–7 in VV and VH polarisations. These spectra exhibit more features than the spectra of the TZN glasses (Fig. 1). All three glasses exhibit the bands identified for the TZN glasses in Fig. 1. In addition, glasses T5–7 show a band at around 925 cm1. Glass T5 has an additional band at around 370 cm1. 3.2. Calculated raman gain spectra Fig. 3 depicts the calculated Raman gain spectra of TZN glasses (T1 and T2) with fused-SiO2 for comparison. It

Fig. 3. Calculated Raman gain spectra of TZN glasses with SiO2 for comparison.

M.D. O’Donnell et al. / Optical Materials 30 (2008) 946–951

Fig. 4. Calculated Raman gain spectra of tungsten–tellurite glasses.

should be noted that the low wavenumber tail (see Fig. 1) has been eliminated by the Bose–Einstein correction. The decrease in intensity of the bands at 440 and 665 cm1 with PbO addition to the glass can be seen more clearly here. Fig. 4 shows the calculated Raman gain spectra of the tungsten–tellurite glasses (T5–7). It can be seen there is a decrease of intensity in the bands at around 440 and 665 cm1 and an increase in the bands at around 765 and 925 cm1 with WO3 addition. The peak Raman gains for these five tellurite glasses are summarised in Table 1 with glass compositions. 4. Discussion 4.1. Raman spectra Figs. 1 and 3 show the spontaneous Raman scattering spectra and calculated Raman gain spectra, respectively, for the TZN glasses T1 and T2. It can be seen that PbO addition decreases the intensity of the bands at around 440 and 665 cm1. These correspond to Te–O–Te chains and [TeO4] units in the glass, respectively. Addition of lead increases the intensity of the band at 765 cm1 corresponding to [TeO3]/[TeO3+1] units [13]. The [TeO3+1] unit can be thought of as distorted trigonal bipyramid [TeO4] units, with one oxygen further away from the central tellurium than the remaining three oxygens. This increase in lower coordination at the expense of higher coordination units is indicative of depolymerisation of the tellurite glass network. PbO is an intermediate, which can link the glass network in places (coordination number 4), or loosen the network (coordination number 6–8). PbO enters the tellurite network as [PbO4] tetrahedra and [PbO6] octahedra [14]. Therefore, at this doping level (3 mol.% PbO), the lead must be predominantly in the [PbO6] sites as the Raman

949

spectra and DTA data [7] indicating the glass network is loosened with lead addition, due to an increase in lower coordination [TeO3]/[TeO3+1] units, manifested by a drop in Tg. The high non-linearity of the tellurite glasses can be also clearly seen compared to the fused-silica also shown in Fig. 3. The peak Raman gain of fused-silica is around 0.89 · 1013 m W1 at 440 cm1. By comparison, the peak Raman gain of the TZN glass T1 studied here was Raman shifted further (663 cm1), with a peak intensity over 20 times more than fused-silica (19.9 · 1013 m W1). Figs. 2 and 4 present the spontaneous Raman scattering spectra and calculated Raman gain spectra, respectively, for the tungsten–tellurite glasses T5–7. In addition to the structural units from the tellurite glass network, additional features can be seen. Bands at 370 cm1 (Te–O–Bi [15]) for glass T7 and 925 cm1 (corresponding to distorted [WO4+2] octahedral units [16]) for glasses T5–7 can be seen in these spectra. Bismuth tellurite compounds have not been studied extensively with Raman spectroscopy. However, Udovic et al. [15] performed Raman spectroscopy on crystalline Bi2O3, Bi2Te4O11 and a series of TeO2Bi2O3-TiO2 glasses. The spectra of the Bi2Te4O11 showed two bands around 400 cm1 (376 and 413 cm1) which were attributed, after lattice dynamics calculations, to Te–O–Bi linkages [15]. Other units (e.g. niobate units) may be hidden by the strong Raman response of the TeO2 matrix and/or are of low intensity. Addition of WO3 decreases the intensity of the bands at 440 and 665 cm1attributed to Te–O–Te chains and [TeO4] units, respectively. At the expense of these units, WO3 addition increases the intensity of the band at 765 cm1 due to [TeO3]/[TeO3+1] units and the band at 925 cm1 from [WO4+2] units. With the addition of tungsten it would be expected that the tellurite glass network is depolymerised. W6+ is a large heavy ion, therefore its addition to the glass will loosen and strain the TeO2 network. The band positions and identification are summarised in Table 2. The Raman shifts were converted to absolute and shifted wavelengths (ka and ks, respectively) by the following relations ka ¼

kp ; 1  ð100~ts kp Þ

ð6Þ

ks ¼ k a  k p ;

ð7Þ

where ~ts is the Raman shift measured in cm1 and kp is the pump wavelength. The calculated peak Raman gain values of these glasses are summarised in Table 1. The peak Table 2 Bands identified from Raman spectra of TZN and tungsten tellurite glasses showing the wavelength shift of the bands from 1064 nm pumping Structural unit

~ts (cm1)

ks (nm)

ka (nm)

Glass

Te–O–Bi Te–O–Te [TeO4] [TeO3]/[TeO3+1] W–O

367 452 663 726 922

43 54 81 89 116

1107 1118 1145 1153 1180

T7 T1–2, T5–7 T1–2, T5–7 T1–2, T5–7 T5–7

950

M.D. O’Donnell et al. / Optical Materials 30 (2008) 946–951

Raman gains of the tellurite glasses studied are 20–25 times greater than fused-silica. The peak Raman gain of the TZN glasses decreases with lead addition for the glass pair T1 and T2. This is perhaps contrary to the expected trend, as lead is a heavier more polarisable ion than tellurium. But on closer inspection of the Raman gain curves in Fig. 1 we can see that the peak at 765 cm1 due to [TeO3]/[TeO3+1] units is increasing at the expense of the 665 cm1 [TeO4] peak with PbO addition. Although the peak Raman gain for the 3 mol.% PbO glass (T2) is lower than the TZN base glass (T1), at higher lead contents the peak gain may be higher than the 0 mol.% PbO glass (T1) and positioned on the 765 cm1 band rather than the 665 cm1 band. It can be seen that addition of tungsten oxide decreases the peak Raman gain of the glass series T5–7. Again this may be contrary to what we expect due to the heavy polarisable nature of W6+. However, as with lead addition in the TZN glasses, the band at 765 cm1 due to [TeO3]/[TeO3+1] units is increasing at the expense of the 665 cm1 [TeO4] peak with WO3 addition. For compositions with >25 mol.% WO3 the intensity of the 765 cm1 band may exceed the 665 cm1 band, resulting in higher gain around 765 cm1 compared to the low WO3 glasses. The peak gain of glass T2 shows good agreement with calculations and direct measurements previously made on an identical glass composition [7]. In the previous study, the peak gain was directly measured at 1064 nm and found to be 17 ± 2 · 1013 m W1, compared to 13 17.9 · 10 m W1 calculated here.

The minimum amount of power, Pmin, required to be launched into a fibre for Raman amplification with a core of effective area, Aeff, and a peak Raman gain, gR, to overcome absorption, a (in cm1 and proportional to loss), is given by [8]. aAeff : gR

ð8Þ

The intensity of the light transmitted after distance L, IL, is shown by the Beer–Lambert law I L ¼ I 0 eaL ;

Loss/dB m1

434 100 10 5 2 1 0.1 0.01 0.001

T2

T7 2

Pmin (kW)

Pq (GW cm )

Pmin (kW)

Pq (GW cm2)

4.61999 1.06379 0.10638 0.05319 0.02128 0.01064 0.00106 0.00011 0.00001

5.88235 1.35446 0.13545 0.06772 0.02709 0.01354 0.00135 0.00014 0.00001

3.27249 0.75352 0.07535 0.03768 0.01507 0.00754 0.00075 0.00008 0.00001

4.16667 0.95941 0.09594 0.04797 0.01919 0.00959 0.00096 0.00010 0.00001

Fig. 5. Minimum power densities required to stimulate Raman gain with variation in loss for glasses (T2 and T7) with core diameters: 10 and 20 lm.

4.2. Calculations of raman fibre amplifier performance

P min ¼

Table 3 Variation in minimum laser power, Pmin, required to stimulate Raman gain in glasses T2 and T3 with loss, and the corresponding power densities, Pq, for a 10 lm diameter core (Aeff = 7.85 · 1011 m2)

ð9Þ

where I0 is the initial intensity. The loss in dB is shown by the following relation, which can be modified to incorporate Eq. (9) if we assume I0 = 1   10log10 II L0 10log10 ðeaL Þ : ð10Þ ¼ Loss ðdB:m1 Þ ¼ L L Therefore, an absorption coefficient (a) of 1 cm1 is equivalent to a loss of 434 dB m1. Table 3 summarises the variation in minimum laser power, Pmin, required to stimulate Raman gain in glasses T2 and T3 with loss, and the corresponding power densities, Pq, for a 10 lm

diameter core (Aeff = 7.85 · 1011 m2). Typical loss values for unclad fibres made from these materials without any drying of precursors and the melt atmosphere, at 1550 nm were 2–3 dB m1 [7]. It can be seen that the power densities are well below the surface optical damage thresholds for these materials (15 GW cm2 [7]), therefore damage of the Raman amplifier would not occur during operation. Fig. 5 illustrates the minimum power densities required to stimulate Raman scattering with variation in loss for two glasses (T2 and T7) and two optical fibre core diameters: 10 and 20 lm. It can be seen that the tungsten tellurite glass requires less laser power to stimulate Raman gain for both diameters compared to the TZN glass, due to the higher non-linearity and Raman response of this glass. Again, low losses, low core diameters and high gain materials are preferred to avoid laser damage. 5. Conclusions The Raman gain intensity and bandwidth of tellurite glasses can be fine tuned with cationic additions to the

M.D. O’Donnell et al. / Optical Materials 30 (2008) 946–951

glass. Addition of PbO and WO3 increases the population of lower coordination units ([TeO3]/[TeO3+1]) in the glass network at the expense of structures which reinforce the network (Te–O–Te and [TeO4]). This change in structure can shift the intensity and position of the most intense Raman peak. For example, a tellurite glass with a stronger reinforced network would typically show peak Raman gain around 665 cm1 due to [TeO4] units. A glass with a looser depolymerised network may exhibit peak Raman gain around 765 cm1 due to [TeO3]/[TeO3+1] units. Therefore, the peak Raman gain of a glass system does not just depend on the polarisability of the ions in the glass, but also the structural sites the cations reside in, and how they affect the structure of the other constituents in the glass. Material damage should not be an issue for these tellurite materials when incorporated into a Raman fibre-optic amplifier. The high laser powers typically used (MW cm2) to stimulate the non-linear Raman effect will be well below the power densities which could potentially damage the material (GW cm2). If losses can be achieved in the final core-clad preforms of the order of <1 dB m1 efficient Raman amplifiers could be realised. References [1] H. Masuda, A. Mori, K. Shikano, K. Oikawa, K. Kato, M. Shimizu, Electronics Letters 38 (2002) 867. [2] H. Masuda, A. Mori, K. Shikano, M. Shimizu, Journal of Lightwave Technology 24 (2006) 504.

951

[3] A. Mori, H. Masuda, K. Shikano, K. Oikawa, K. Kato, M. Shimizu, Electronics Letters 37 (2001) 1442. [4] A. Mori, H. Masuda, K. Shikano, M. Shimizu, Journal of Lightwave Technology 21 (2003) 1300. [5] J. Toulouse, Journal of Lightwave Technology 23 (2005) 3625. [6] R. Stegeman, L. Jankovic, H. Kim, C. Rivero, G. Stegeman, K. Richardson, P. Delfyett, Y. Guo, A. Schulte, T. Cardinal, Optics Letters 28 (2003) 1126. [7] M.D. O’Donnell, A.B. Seddon, D. Furniss, V.K. Tikhomirov, C. Rivero, M. Ramme, R. Stegeman, G. Stegeman, K. Richardson, R. Stolen, M. Couzi, T. Cardinal, Journal of the American Ceramic Society 90 (2007) 1448. [8] M.D. O’Donnell, C.A. Miller, D. Furniss, V.K. Tikhomirov, A.B. Seddon, Journal of Non-Crystalline Solids 331 (2003) 48. [9] M.D. O’Donnell, D. Furniss, V.K. Tikhomirov, A.B. Seddon, Physics and Chemistry of Glasses: European Journal of Glass Science and Technology B 47 (2006) 121. [10] C. Rivero, R. Stegeman, M. Couzi, D. Talaga, T. Cardinal, K. Richardson, G. Stegeman, Optics Express 13 (2005) 4759. [11] G. Dai, F. Tassone, A. Li Bassi, V. Russo, C.E. Bottani, F. D’Amore, IEEE Photonics Technology Letters 16 (2004) 1011. [12] P.N. Butcher, D. Cotter, The Elements of Nonlinear Optics, Cambridge University Press, Cambridge, 1990. [13] M. Vithal, P. Nachimuthu, T. Banu, R. Jagannathan, Journal of Applied Physics 81 (1997) 7922. [14] M.A.P. Silva, Y. Messaddeq, S.J.L. Ribeiro, M. Poulain, F. Villain, V. Briois, Journal of Physics and Chemistry of Solids 62 (2001) 1055. [15] M. Udovic, P. Thomas, A. Mirgorodsky, O. Durand, M. Soulis, O. Masson, T. Merle-Mejean, J.C. Champarnaud-Mesiard, Journal of Solid State Chemistry 179 (2006) 3252. [16] I. Shaltout, Y. Tang, R. Braunstein, A.M. Abuelazm, Journal of Physics and Chemistry of Solids 56 (1995) 141. [17] T. Mizunami, H. Iwashita, K. Takagi, Optics Communications 97 (1993) 74.