- Email: [email protected]
Spectral properties and thermal stability of Er3+-doped oxyfluoride silicate glasses for broadband optical amplifier
Journal of Alloys and Compounds 361 (2003) 313–319
L
www.elsevier.com / locate / jallcom
Spectral properties and thermal stability of Er 31 -doped oxyfluoride silicate glasses for broadband optical amplifier Shiqing Xu*, Zhongmin Yang, Shixun Dai, Jianhu Yang, Lili Hu, Zhonghong Jiang Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China Received 10 March 2003; received in revised form 12 April 2003; accepted 12 April 2003
Abstract Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 glasses were prepared. The effect of PbF 2 content on refractive indices, densities, absorption spectra, the Judd–Ofelt parameters Vt (t52,4,6), fluorescence spectra and the lifetimes of 4 I13 / 2 level of Er 31 in 50SiO 2 –(502x)PbO– xPbF 2 glasses were investigated, and the stimulated emission cross-section was calculated from McCumber theory. With increasing PbF 2 content in the glass composition, the V6 parameter, fluorescence full width at half maximum (FWHM) and lifetimes of 4 I13 / 2 level of Er 31 increase, while refractive indices and densities decrease. Compared with other glass hosts, the gain bandwidth properties of Er 31 -doped 50SiO 2 –50PbF 2 glass is close to those of tellurite and bismuth glasses, and has an advantage over those of silicate, phosphate and germante glasses. The broad and flat 4 I13 / 2 → 4 I15 / 2 emission of Er 31 around 1.55 mm can be used as host material for potential broadband optical amplifier in the wavelength–division–multiplexing (WDM) network system. 2003 Elsevier B.V. All rights reserved. Keywords: Rare earth compounds; Optical materials; Thermal analysis; Optical spectroscopy
1. Introduction In recent years, due to the rapid increase of information capacity and the need for flexible networks, there exists urgent demand for optical amplifier with a wide and flat gain spectrum in the telecommunication window, to be used in the wavelength–division–multiplexing (WDM) network system [1–4]. Although the performance of the present silica-based erbium-doped fiber amplifier (EDFA) can be used in WDM but with fewer channels, there exists the requirement for larger number of channels, which could be possible by using an EDFA with a wider gain spectrum [2]. From a practical standpoint, the flatness of the gain is also critically important because the light intensity for different channels would be varied by the multistep amplification if the gain of optical amplifier were not independent of wavelengths. So far, much effort has been spent on phosphate, fluorophosphate, tellurite and bismuth glasses [1,4–9]. Specifically, tellurite and bismuth glasses are capable of providing the broad fluorescence full width at half maximum (FWHM) and large stimulated emission cross-section around 1.55 mm bands, and thus are *Corresponding author. E-mail address: [email protected] (S. Xu). 0925-8388 / 03 / $ – see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016 / S0925-8388(03)00447-X
attractive host materials for 1.55 mm broadband amplification [10,11]. However, chemical and mechanical stability and fiberizability of these glasses as practical materials still remains problematic [12]. Furthermore, in tellurite glasses, the long population duration in 4 I11 / 2 level with the 970 nm excitation owing to the low phonon energy of glass host results in strong upconversion, lowers the 970 nm pumping efficiency and enhances the noise figure of broadband optical amplifier. In bismuth glasses [2,11], the broad FWHM is mainly attributed to a large amount of B 2 O 3 , but strong fluorescence quenching exists owing to the introduction of B 2 O 3 at the same time. However, silicate glasses are the most chemically and mechanically stable and also are more easily fabricated into various shapes such as a rod and optical fiber [13]. Therefore, the design of a new silicate glass host for Er 31 for wide and flat spectra of the 4 I13 / 2 → 4 I15 / 2 transition around 1.55 mm bands is a target at present. In this work, the new oxyfluoride silicate glasses (50SiO 2 –(502x)PbO–xPbF 2 ) were prepared. The broad and flat fluorescence around 1.55 mm of the studied glass samples indicate that it is good host material for potential broadband optical amplifier in WDM. The physical and thermal properties, the absorption spectra, fluorescence
314
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
spectra and the lifetimes of 4 I13 / 2 level of Er 31 of the glasses were investigated. The absorption and stimulated emission cross-sections were calculated from McCumber theory, and the Judd–Ofelt intensity parameters Vt (t5 2,4,6) were determined using the Judd–Ofelt theory.
2. Experimental Glasses with the compositions of 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) were prepared. The starting materials are reagent grade SiO 2 , Pb 3 O 4 , PbF 2 and Er 2 O 3 . The Er 31 doping concentration in the glasses was 1.0 mol%, which was introduced as with 99.99% purity. About 100 g batches of starting materials were fully mixed and then melted between 1000 8C and 1100 8C in covered silicate crucibles in a SiC Globar furnace with an N 2 atmosphere. When the melting was completed, the glass liquids were cast into stainless steel plates. All glass samples were annealed for 2 h at temperatures ranging from 320 8C to 400 8C, depending upon the glass composition. The glass samples were then cooled to room temperature at a rate of 10 8C / h, and then were cut and polished carefully in order to meet the requirements for optical measurements. The glass transition temperature (T g ), and crystallization onset temperature (T x ) were determined by differential thermal analysis (DTA) at a heating rate of 10 8C / min, using an aluminum oxide ceramic pan. Densities were measured according to the Archimedes’ principle using distilled water as the medium. Refractive indices were measured by the prism minimum deviation method. UV/ VIS / NIR absorption spectra were recorded between 300 and 1700 nm using a spectrophotometer on 3-mm-thick glass samples optically polished. Fluorescence spectra were recorded using a laser diode excitation source at 970 nm. The fluorescence lifetimes of 4 I13 / 2 level of Er 31 were measured with light pulses of 970 nm laser diode (LD) and a HP546800B 100-MHz oscilloscope. All the measurements were taken at room temperature.
Fig. 1. Compositional dependence of refractive indices and densities of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
and DT 5 T x 2 T g of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%). With increasing PbF 2 content, the value of T x , T g , and DT 5 T x 2 T g decrease. The difference between T x and T g , DT 5 T x 2 T g , has been frequently used as a rough estimate of the glass formation ability or glass stability. Since fiber drawing is a reheating process and any crystallization during the process will increase the scattering loss of the fiber and then degrade the optical properties [14]. To achieve a large working range of temperature during our sample fiber drawing, it is desirable for a glass host to have DT as large as possible [14,15]. From Fig. 2, it can be seen that the smallest value of DT is 231 8C, which is larger than those of tellurite (141.5 8C), bismuth (170 8C) and fluoride (105 8C) glasses [4,11,16], indicating these glass samples are more stable against devitrification [17]. The introduction of PbF 2 leads to a decrease of T x , T g and DT of these glass samples. Such features were explained as follows: with the substitution of PbF 2 for PbO, the gradual replacement of the Pb5O double bonds by two Pb–F
3. Results and discussion
3.1. Physical and thermal stability properties Fig. 1 shows the compositional dependence of refractive indices and densities of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. With increasing PbF 2 content, refractive indices and densities decrease, which was explained as follows: due to the gradual replacement of the Pb5O double bonds by two Pb–F single bonds, the structure of glasses becomes looser, which results in the decrease of density. Correspondingly, refractive indices also decrease. Fig. 2 shows the compositional dependence of T x , T g ,
Fig. 2. Compositional dependence of T g , T x and DT of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
315
Fig. 3. Absorption spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. The assignments of absorption bands are indicated by the excited level.
single bonds, the structure of glasses becomes looser, thus weakening the network structure; the T x and T g of the glass samples decrease. The DT value of these glass samples decrease because the decreased velocity of T x is more rapid than that of T g .
3.2. Absorption spectra and Judd–Ofelt analysis
Fig. 5. The partial energy level diagram of Er 31 in 50SiO 2 –50PbF 2 glass.
Fig. 3 shows the absorption spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Each assignment corresponds to the excited level of Er 31 . The cut-off band shifts to a shorter wavelength with the substitution of PbF 2 for PbO. The approximate 1.55 mm transition 4 I15 / 2 → 4 I13 / 2 is important for optical amplifier operation in the telecommunication window. The compositional dependence of absorption peak wavelength ( lp ) of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) is shown in
Fig. 4. With increasing PbF 2 content from 0 to 50 mol%, the lp shifts from 1535 to 1532 nm. The partial energy level diagram of Er 31 in 50SiO 2 –50PbF 2 glass is presented in Fig. 5. The Judd–Ofelt theory [15,19] was often used to calculate the spectroscopic parameters such as intensity parameters Vt (t52,4,6), spontaneous emission probability, fluorescence branching ratios, radiation lifetime and the line strength of electric-dipole transition (Sed ) and magnetic dipole transition (Smd ) of rare-earth ion-doped glasses. The three parameters Vt (t52,4,6) can be obtained experimentally from the measured absorption spectra and refractive indices of the glass hosts. The value of V2 , V4 and V6 for the glass samples are calculated by the
Fig. 4. Compositional dependence of absorption peak wavelength of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
Fig. 6. Compositional dependence of intensity parameters Vt (t52,4,6) of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
316
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
procedure provided in Refs. [20,21]. Fig. 6 shows the compositional dependence of Vt (t52,4,6) parameters of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Apparently, with increasing PbF 2 content, the values of V4 and V6 increase monotonically, and the value of V2 decreases monotonically. According to previous studies [2,22], V2 is related with the symmetry of the glass hosts while V6 is inversely proportional to the covalency of Er–O bond. The covalency of Er–O bond is assumed to be related with the local basicity around the rare-earth sites, which can be adjusted by the composition or structure of the glass hosts [8]. On the basis of the electronegativity theory [23], the smaller the difference of electronegativity between cation and anion ions, the stronger the covalency of the bond. The values of electronegativity, for Pb, O, and F elements, are 1.8, 3.5, and 4.0, respectively. As a result, the covalency of Pb–F bond is lower than that of Pb–O bond. Consequently, the value of V6 increase with the substitution of PbF 2 for PbO.
the bandwidth, Fig. 8 shows the compositional dependence of FWHM of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Obviously, FWHM of these glass samples increase monotonically with increasing PbF 2 content. The broadest FWHM value of Er 31 -doped 50SiO 2 –50PbF 2 glass is about 60 nm. In the case of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 , because the difference in the total angular momentum DJ equals 1, there exists the contribution of magnetic dipole transitions (Smd ) [18,19,26]. In order to obtain broadband and flat emission spectrum, it will be effective to increase the relative contribution of electric dipole transition (Sed ) [2]. Smd is independent of the ligand fields and is characteristic of the transition determined by the quantum numbers, while Sed is a function of the ligand fields, and it is possible to increase Sed by modifying the structure and compositions of the glass hosts. According to the Judd– Ofelt theory, the line strength of Sed of the 1.55 mm 4 I13 / 2 → 4 I15 / 2 transition of Er 31 is given in Ref. [27].
3.3. Fluorescence spectra and stimulated emission crosssection
S ed [ I13 / 2 , I15 / 2 ] 5 0.0188 V2 1 0.1176 V4 1 1.4617 V6
The stimulated emission cross-section and FWHM are very important for optical amplifier to realize broadband and the gain amplification, which can be used to evaluate the gain properties of the optical amplifier. The gain bandwidth of optical amplifier is determined largely by the width of the fluorescent spectrum and stimulated emission cross-section [24,25]. The fluorescence spectra of Er 31 doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses are shown in Fig. 7. The highest emission intensities are normalized to 1 for all the spectra. FWHM is often used as a semiquantitative indication of
Fig. 7. Compositional dependence of the normalized emission spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
4
4
(1) Consequently, for Sed of 4 I13 / 2 → 4 I15 / 2 transition of Er , the V6 parameter is dominant and consequently is also dominant for 1.55 mm emission. Therefore, increasing the V6 parameter is helpful in increasing the bandwidth of the 1.55 mm emission. Fig. 9 shows the relationship between FWHM and the line strength of Sed of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 -doped 50SiO 2 –(502 x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. The line strength of Sed shows a strong dependence on FWHM of the 1.55 mm emission band, and the glass sample with a large line strength of Sed generally possesses a broad FWHM. In addition, the line strength of Sed 31
Fig. 8. Compositional dependence of FWHM of Er 31 -doped 50SiO 2 – (502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
Fig. 9. Relationship between FWHM and the line strength of electric dipole transition (Sed ) of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
increases with increase of the V6 parameter; from Fig. 6, it is seen that the V6 parameter increase with increasing PbF 2 content. Consequently, the glass sample with a large V6 also generally possesses a broad FWHM. This result demonstrates the validity of the above statement. The absorption cross-section was determined from the absorption spectra, and the stimulated emission crosssection is calculated from McCumber theory [28]. According to the McCumber theory, the absorption and stimulated emission cross-sections are related by:
se ( l) 5 sa ( l) exp [(´ 2 hy ) /kT ]
(2)
where sa and se are absorption and stimulated emission cross-sections, respectively, the h is the Planck constant, k is the Boltzmann constant, y is the photon frequency, ´ is the net free energy required to excite one Er 31 from the 4 I15 / 2 state to 4 I13 / 2 at temperature T. ´ was determined
Fig. 10. Absorption and stimulated emission cross-section of Er 31 in 50SiO 2 –50PbF 2 glass.
317
using the procedure provided in Ref. [29]. Fig. 10 illustrates the absorption and stimulated emission cross-sections of Er 31 in 50SiO 2 –50PbF 2 glass. The stimulated emission cross-section is proportional to the refractive index of the glass host [s | (n 2 1 2)2 /n] [12]. Since the glass host has a much larger refractive index, it is expected that Er 31 in 50SiO 2 –50PbF 2 glass is capable of providing large stimulated emission cross-section at the 1.55 mm bands. The emission parameters FWHM, the peak of the stimulated emission cross-section (s peak ) and FWHM3 e s epeak of Er 31 in different glass hosts are listed in Table 1 in order to compare the emission properties of Er 31 . The gain bandwidth of optical amplifier can be evaluated by FWHM3 s peak product. The bigger the product, the better e the property. From Table 1, It is seen that the value of FWHM, s peak and FWHM3 s epeak for Er 31 in 50SiO 2 – e 50PbF 2 glass is close to those of tellurite and bismuth glasses, and more than those of silicate, phosphate and germanate glasses. Consequently, the glass samples can be used as host material for potential broadband optical amplifier in WDM.
3.4. Lifetimes of 4 I13 / 2 level of Er 31 The lifetime of 4 I13 / 2 level of Er 31 is also an important parameter for optical amplifier. A critical factor in the success of Er 31 -doped fiber amplifier in optical communications is the long lifetime of the metastable state that permits the required high population inversions to be obtained under steady-state conditions using modest pump powers [32]. Fig. 11 shows the compositional dependence 4 31 of the lifetime of I13 / 2 level of Er in 50SiO 2 –(502 x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Obviously, the lifetimes of 4 I13 / 2 level of Er 31 increase with increasing PbF 2 content. The lifetime is related mainly to the phonon energy of glass hosts. If neglected, the energy transfer process, in accordance with the simplified rate relation among total lifetime tm , radiative lifetime tr and the non-radiative lifetime tnr , 1 /tm 5 1 /tr 1 1 /tnr , the shortening of the measured lifetime is due to the enhancement of non-radiative decay processes [33]. The larger the phonon energy of the glass host, the higher the probability for non-radiative decay, and consequently the lower the lifetime of 4 I13 / 2 level of Er 31 . On the contrary, the lower the phonon energy of the glass host, the higher the upconversion emission, which is also a disadvantageous property for optical amplifier. Thus the glass host should have medium phonon energy for Er 31 to realize good amplification properties. The increased lifetime in the studied glass samples could be explained as follows: firstly, because the phonon energy of Pb–F bonds is lower than that of Pb–O, with the substitution of PbF 2 for PbO, the non-radiative transition rate of Er 31 decreases, which leads to the increase of the measured lifetimes; secondly,
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
318
Table 1 Comparisons of FWHM, s peak and FWHM3 s peak of Er 31 in different glass hosts e e Glass
50SiO 2 –50PbF 2
Tellurite
Bismuth
Phosphate
Germanate
Silicate
FWHM (nm) s peak (10 221 cm 2 ) e FWHM3 s peak e Ref.
60 7.4 444 This work
65 7.5 487.5 [30]
79 7.0 554 [11]
37 6.4 236.8 [9]
42 5.7 239.4 [31]
40 5.5 220 [30]
the radiative lifetime calculated by the Judd–Ofelt theory is inversely proportional to refractive index of the glass host, due to the refractive index of the glass samples decreasing with increasing PbF 2 content, and so is the measured lifetime. In addition, free OH 2 in the glass is regarded as one of the important quenching centers in Er 31 -doped glasses which is attributed to the strong stretch vibration of OH 2 at 3600 cm 21 (2.79 mm) [34]. The energy of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 (|6500 cm 21 ) corresponds to the energy of the second harmonic of the OH 2 stretching vibration. Accordingly, non-radiative relaxation occurs by excitation of two OH 2 vibrational quanta when an Er 31 is coupled to OH 2 . However, the control of OH 2 bands isn’t taken into consideration in our experiment, if OH 2 bands of glass samples are eliminated, lifetimes of 4 I13 / 2 levels of Er 31 will be longer.
4. Conclusions The present oxyfluoride silicate glasses were prepared and analyzed. Er 31 -doped 50SiO 2 –50PbF 2 glass showed a broad emission spectra of 1.55 mm with a large stimulated emission cross-section, long lifetimes of 4 I13 / 2 level of Er 31 and good thermal stability. The gain bandwidth
properties of Er 31 -doped 50SiO 2 –50PbF 2 glass was evaluated by FWHM3 s peak product, and the FWHM3 s epeak e 31 value of Er -doped 50SiO 2 –50PbF 2 glass is 444, which is close to those of tellurite glass (487.5) and bismuth glass (554), is larger than that of silicate glass (220), phosphate glass (236.8) and germanate glass (239.4). The large stimulated emission cross-section of Er 31 -doped 50SiO 2 – 50PbF 2 glass (se 57.4310 221 cm 2 ) results from the large refractive index of the glass host. According to the Judd– Ofelt theory, the FWHM value is related to the V6 parameter and the line strength of Sed , and the larger the V6 parameter and the line strength of Sed , the broader the FWHM. DT has been used as a rough estimate of the glass formation ability or glass stability, and the DT value of Er 31 -doped 50SiO 2 –50PbF 2 glass is 231 8C, which is larger than those of tellurite (141.5 8C), bismuth (170 8C) 4 and fluoride (105 8C) glass. The lifetimes of the I13 / 2 level of Er 31 in the glass hosts were measured, and the lifetimes increase from 4.9 to 9.4 ms with increase of PbF 2 content from 0 to 50 mol%.
Acknowledgements We would like to express our thanks to Dr Nengli Dai for his useful discussion. This work is supported by the Project of the National Nature Science Foundation of China (Grant No 60207006) and Optical Science and Technology of Shanghai (Grant No 022261046).
References
Fig. 11. Compositional dependence of fluorescence lifetime of the 4 I13 / 2 level of Er 31 in 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
[1] S. Tanabe, N. Sugimoto, S. Ito, T. Hanada, J. Lumin. 87–89 (2000) 670. [2] S. Tanabe, J. Non-Cryst. Solids 259 (1999) 1. [3] M. Yamada, A. Mori, K. Kobayashi, H. Ono, IEEE Photonics Technol. Lett. 10 (1998) 1244. [4] Y. Ding, S.B. Jiang, B.C. Hwang, T. Luo, N. Peyghambarian, Y. Himei, T. Ito, Y. Miura, Opt. Mater. 15 (2000) 123. [5] N. Sugimoto, Rev. Laser. Eng. 29 (2001) 590. [6] J. Yang, S. Dai, L. Wen, Z. Liu, L. Hu, Z. Jiang, Chin. Phys. Lett. 19 (2002) 1516. [7] S. Tanabe, J. Non-Cryst. Solids 256–257 (1999) 282. [8] S. Tanabe, S. Yoshii, K. Hirao, N. Soga, Phys. Rev. B 45 (1992) 4620. [9] S. Jiang, T. Luo, B.C. Hwang, F. Smekatala, J. Non-Cryst. Solids 263–264 (2000) 364. [10] Y. Ohishi, A. Mori, M. Yamada, H. Ono, Opt. Lett. 23 (1998) 274.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319 [11] J. Yang, S. Dai, Y. Zhou, L. Wen, L. Hu, Z. Jiang, J. Appl. Phys. 93 (2003) 977. [12] J.S. Wang, E.M. Vogel, E. Snitzer, Opt. Mater. 3 (1994) 187. [13] F. Guinhos, P. Nobrega, P. Santa-Cruz, J. Alloys Comp. 323–324 (2001) 358. [14] L.L. Neindre, S.B. Jiang, B.C. Hwang, T. Luo, J. Watson, N. Peyghambarian, J. Non-Cryst. Solids 255 (1999) 97. [15] M.G. Drexhage, O.H. El Bayoumi, C.T. Moyniyan, J. Am. Ceram. Soc. C 65 (1982) 168. [16] D.R. Macfarlane, J.S. Javorniczky, P.J. Newman, D.J. Booth, V. Bogdanov, J. Non-Cryst. Solids 184 (1995) 249. [17] J.S. Wang, E.M. Vogel, E. Snitzer, J. Non-Cryst. Solids 178 (1994) 109. [18] B.R. Judd, Phys. Rev. 127 (1962) 750. [19] S. Ofelt, J. Chem. Phys. 37 (1962) 511. [20] S. Tanabe, T. Ohyagi, N. Soga, T. Hanada, Phys. Rev. B 46 (1992) 3305.
319
[21] H. Takebe, Y. Nageno, K. Morinaga, J. Am. Ceram. Soc. 78 (1995) 1161. [22] S. Tanabe, J. Appl. Phys. 73 (1993) 8451. [23] L. Pauling, J. Am. Chem. Soc. 51 (1929) 1010. [24] A. Mori, Y. Ohishi, S. Sudov, Electron. Lett. 33 (1997) 863. [25] S. Shen, M. Naftaly, A. Jha, Proc. SPIE 3489 (1999) 103. [26] S. Tanabe, T. Hanada, J. Non-Cryst. Solids 196 (1996) 101. [27] M.J. Weber, Phys. Rev. 157 (1967) 262. [28] D.E. McCuber, Phys. Rev. A 134 (1964) 299. [29] W.J. Miniscalo, R.S. Quimby, Opt. Lett. 16 (1992) 258. [30] X. Zou, T. Izumitani, J. Non-Cryst. Solids 162 (1993) 68. [31] H. Lin, E.Y.B. Pun, S.Q. Man, X.R. Liu, J. Opt. Soc. Am. B 18 (2001) 602. [32] R. Rolli, A. Chiasera, M. Montagna, Proc. SPIE 4282 (2001) 109. [33] L. Petit, T. Cardinal, J. Videau, G. Flem, Y. Guyot, G. Boulon, M. Couzi, T. Buffeteau, J. Non-Cryst. Solids 298 (2002) 76. [34] X. Feng, S. Tanabe, T. Hanada, J. Non-Cryst. Solids 281 (2001) 48.
L
www.elsevier.com / locate / jallcom
Spectral properties and thermal stability of Er 31 -doped oxyfluoride silicate glasses for broadband optical amplifier Shiqing Xu*, Zhongmin Yang, Shixun Dai, Jianhu Yang, Lili Hu, Zhonghong Jiang Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China Received 10 March 2003; received in revised form 12 April 2003; accepted 12 April 2003
Abstract Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 glasses were prepared. The effect of PbF 2 content on refractive indices, densities, absorption spectra, the Judd–Ofelt parameters Vt (t52,4,6), fluorescence spectra and the lifetimes of 4 I13 / 2 level of Er 31 in 50SiO 2 –(502x)PbO– xPbF 2 glasses were investigated, and the stimulated emission cross-section was calculated from McCumber theory. With increasing PbF 2 content in the glass composition, the V6 parameter, fluorescence full width at half maximum (FWHM) and lifetimes of 4 I13 / 2 level of Er 31 increase, while refractive indices and densities decrease. Compared with other glass hosts, the gain bandwidth properties of Er 31 -doped 50SiO 2 –50PbF 2 glass is close to those of tellurite and bismuth glasses, and has an advantage over those of silicate, phosphate and germante glasses. The broad and flat 4 I13 / 2 → 4 I15 / 2 emission of Er 31 around 1.55 mm can be used as host material for potential broadband optical amplifier in the wavelength–division–multiplexing (WDM) network system. 2003 Elsevier B.V. All rights reserved. Keywords: Rare earth compounds; Optical materials; Thermal analysis; Optical spectroscopy
1. Introduction In recent years, due to the rapid increase of information capacity and the need for flexible networks, there exists urgent demand for optical amplifier with a wide and flat gain spectrum in the telecommunication window, to be used in the wavelength–division–multiplexing (WDM) network system [1–4]. Although the performance of the present silica-based erbium-doped fiber amplifier (EDFA) can be used in WDM but with fewer channels, there exists the requirement for larger number of channels, which could be possible by using an EDFA with a wider gain spectrum [2]. From a practical standpoint, the flatness of the gain is also critically important because the light intensity for different channels would be varied by the multistep amplification if the gain of optical amplifier were not independent of wavelengths. So far, much effort has been spent on phosphate, fluorophosphate, tellurite and bismuth glasses [1,4–9]. Specifically, tellurite and bismuth glasses are capable of providing the broad fluorescence full width at half maximum (FWHM) and large stimulated emission cross-section around 1.55 mm bands, and thus are *Corresponding author. E-mail address: [email protected] (S. Xu). 0925-8388 / 03 / $ – see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016 / S0925-8388(03)00447-X
attractive host materials for 1.55 mm broadband amplification [10,11]. However, chemical and mechanical stability and fiberizability of these glasses as practical materials still remains problematic [12]. Furthermore, in tellurite glasses, the long population duration in 4 I11 / 2 level with the 970 nm excitation owing to the low phonon energy of glass host results in strong upconversion, lowers the 970 nm pumping efficiency and enhances the noise figure of broadband optical amplifier. In bismuth glasses [2,11], the broad FWHM is mainly attributed to a large amount of B 2 O 3 , but strong fluorescence quenching exists owing to the introduction of B 2 O 3 at the same time. However, silicate glasses are the most chemically and mechanically stable and also are more easily fabricated into various shapes such as a rod and optical fiber [13]. Therefore, the design of a new silicate glass host for Er 31 for wide and flat spectra of the 4 I13 / 2 → 4 I15 / 2 transition around 1.55 mm bands is a target at present. In this work, the new oxyfluoride silicate glasses (50SiO 2 –(502x)PbO–xPbF 2 ) were prepared. The broad and flat fluorescence around 1.55 mm of the studied glass samples indicate that it is good host material for potential broadband optical amplifier in WDM. The physical and thermal properties, the absorption spectra, fluorescence
314
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
spectra and the lifetimes of 4 I13 / 2 level of Er 31 of the glasses were investigated. The absorption and stimulated emission cross-sections were calculated from McCumber theory, and the Judd–Ofelt intensity parameters Vt (t5 2,4,6) were determined using the Judd–Ofelt theory.
2. Experimental Glasses with the compositions of 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) were prepared. The starting materials are reagent grade SiO 2 , Pb 3 O 4 , PbF 2 and Er 2 O 3 . The Er 31 doping concentration in the glasses was 1.0 mol%, which was introduced as with 99.99% purity. About 100 g batches of starting materials were fully mixed and then melted between 1000 8C and 1100 8C in covered silicate crucibles in a SiC Globar furnace with an N 2 atmosphere. When the melting was completed, the glass liquids were cast into stainless steel plates. All glass samples were annealed for 2 h at temperatures ranging from 320 8C to 400 8C, depending upon the glass composition. The glass samples were then cooled to room temperature at a rate of 10 8C / h, and then were cut and polished carefully in order to meet the requirements for optical measurements. The glass transition temperature (T g ), and crystallization onset temperature (T x ) were determined by differential thermal analysis (DTA) at a heating rate of 10 8C / min, using an aluminum oxide ceramic pan. Densities were measured according to the Archimedes’ principle using distilled water as the medium. Refractive indices were measured by the prism minimum deviation method. UV/ VIS / NIR absorption spectra were recorded between 300 and 1700 nm using a spectrophotometer on 3-mm-thick glass samples optically polished. Fluorescence spectra were recorded using a laser diode excitation source at 970 nm. The fluorescence lifetimes of 4 I13 / 2 level of Er 31 were measured with light pulses of 970 nm laser diode (LD) and a HP546800B 100-MHz oscilloscope. All the measurements were taken at room temperature.
Fig. 1. Compositional dependence of refractive indices and densities of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
and DT 5 T x 2 T g of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%). With increasing PbF 2 content, the value of T x , T g , and DT 5 T x 2 T g decrease. The difference between T x and T g , DT 5 T x 2 T g , has been frequently used as a rough estimate of the glass formation ability or glass stability. Since fiber drawing is a reheating process and any crystallization during the process will increase the scattering loss of the fiber and then degrade the optical properties [14]. To achieve a large working range of temperature during our sample fiber drawing, it is desirable for a glass host to have DT as large as possible [14,15]. From Fig. 2, it can be seen that the smallest value of DT is 231 8C, which is larger than those of tellurite (141.5 8C), bismuth (170 8C) and fluoride (105 8C) glasses [4,11,16], indicating these glass samples are more stable against devitrification [17]. The introduction of PbF 2 leads to a decrease of T x , T g and DT of these glass samples. Such features were explained as follows: with the substitution of PbF 2 for PbO, the gradual replacement of the Pb5O double bonds by two Pb–F
3. Results and discussion
3.1. Physical and thermal stability properties Fig. 1 shows the compositional dependence of refractive indices and densities of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. With increasing PbF 2 content, refractive indices and densities decrease, which was explained as follows: due to the gradual replacement of the Pb5O double bonds by two Pb–F single bonds, the structure of glasses becomes looser, which results in the decrease of density. Correspondingly, refractive indices also decrease. Fig. 2 shows the compositional dependence of T x , T g ,
Fig. 2. Compositional dependence of T g , T x and DT of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
315
Fig. 3. Absorption spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. The assignments of absorption bands are indicated by the excited level.
single bonds, the structure of glasses becomes looser, thus weakening the network structure; the T x and T g of the glass samples decrease. The DT value of these glass samples decrease because the decreased velocity of T x is more rapid than that of T g .
3.2. Absorption spectra and Judd–Ofelt analysis
Fig. 5. The partial energy level diagram of Er 31 in 50SiO 2 –50PbF 2 glass.
Fig. 3 shows the absorption spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Each assignment corresponds to the excited level of Er 31 . The cut-off band shifts to a shorter wavelength with the substitution of PbF 2 for PbO. The approximate 1.55 mm transition 4 I15 / 2 → 4 I13 / 2 is important for optical amplifier operation in the telecommunication window. The compositional dependence of absorption peak wavelength ( lp ) of Er 31 -doped 50SiO 2 –(502x)PbO– xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) is shown in
Fig. 4. With increasing PbF 2 content from 0 to 50 mol%, the lp shifts from 1535 to 1532 nm. The partial energy level diagram of Er 31 in 50SiO 2 –50PbF 2 glass is presented in Fig. 5. The Judd–Ofelt theory [15,19] was often used to calculate the spectroscopic parameters such as intensity parameters Vt (t52,4,6), spontaneous emission probability, fluorescence branching ratios, radiation lifetime and the line strength of electric-dipole transition (Sed ) and magnetic dipole transition (Smd ) of rare-earth ion-doped glasses. The three parameters Vt (t52,4,6) can be obtained experimentally from the measured absorption spectra and refractive indices of the glass hosts. The value of V2 , V4 and V6 for the glass samples are calculated by the
Fig. 4. Compositional dependence of absorption peak wavelength of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
Fig. 6. Compositional dependence of intensity parameters Vt (t52,4,6) of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
316
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
procedure provided in Refs. [20,21]. Fig. 6 shows the compositional dependence of Vt (t52,4,6) parameters of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Apparently, with increasing PbF 2 content, the values of V4 and V6 increase monotonically, and the value of V2 decreases monotonically. According to previous studies [2,22], V2 is related with the symmetry of the glass hosts while V6 is inversely proportional to the covalency of Er–O bond. The covalency of Er–O bond is assumed to be related with the local basicity around the rare-earth sites, which can be adjusted by the composition or structure of the glass hosts [8]. On the basis of the electronegativity theory [23], the smaller the difference of electronegativity between cation and anion ions, the stronger the covalency of the bond. The values of electronegativity, for Pb, O, and F elements, are 1.8, 3.5, and 4.0, respectively. As a result, the covalency of Pb–F bond is lower than that of Pb–O bond. Consequently, the value of V6 increase with the substitution of PbF 2 for PbO.
the bandwidth, Fig. 8 shows the compositional dependence of FWHM of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Obviously, FWHM of these glass samples increase monotonically with increasing PbF 2 content. The broadest FWHM value of Er 31 -doped 50SiO 2 –50PbF 2 glass is about 60 nm. In the case of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 , because the difference in the total angular momentum DJ equals 1, there exists the contribution of magnetic dipole transitions (Smd ) [18,19,26]. In order to obtain broadband and flat emission spectrum, it will be effective to increase the relative contribution of electric dipole transition (Sed ) [2]. Smd is independent of the ligand fields and is characteristic of the transition determined by the quantum numbers, while Sed is a function of the ligand fields, and it is possible to increase Sed by modifying the structure and compositions of the glass hosts. According to the Judd– Ofelt theory, the line strength of Sed of the 1.55 mm 4 I13 / 2 → 4 I15 / 2 transition of Er 31 is given in Ref. [27].
3.3. Fluorescence spectra and stimulated emission crosssection
S ed [ I13 / 2 , I15 / 2 ] 5 0.0188 V2 1 0.1176 V4 1 1.4617 V6
The stimulated emission cross-section and FWHM are very important for optical amplifier to realize broadband and the gain amplification, which can be used to evaluate the gain properties of the optical amplifier. The gain bandwidth of optical amplifier is determined largely by the width of the fluorescent spectrum and stimulated emission cross-section [24,25]. The fluorescence spectra of Er 31 doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses are shown in Fig. 7. The highest emission intensities are normalized to 1 for all the spectra. FWHM is often used as a semiquantitative indication of
Fig. 7. Compositional dependence of the normalized emission spectra of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
4
4
(1) Consequently, for Sed of 4 I13 / 2 → 4 I15 / 2 transition of Er , the V6 parameter is dominant and consequently is also dominant for 1.55 mm emission. Therefore, increasing the V6 parameter is helpful in increasing the bandwidth of the 1.55 mm emission. Fig. 9 shows the relationship between FWHM and the line strength of Sed of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 -doped 50SiO 2 –(502 x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. The line strength of Sed shows a strong dependence on FWHM of the 1.55 mm emission band, and the glass sample with a large line strength of Sed generally possesses a broad FWHM. In addition, the line strength of Sed 31
Fig. 8. Compositional dependence of FWHM of Er 31 -doped 50SiO 2 – (502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
Fig. 9. Relationship between FWHM and the line strength of electric dipole transition (Sed ) of Er 31 -doped 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
increases with increase of the V6 parameter; from Fig. 6, it is seen that the V6 parameter increase with increasing PbF 2 content. Consequently, the glass sample with a large V6 also generally possesses a broad FWHM. This result demonstrates the validity of the above statement. The absorption cross-section was determined from the absorption spectra, and the stimulated emission crosssection is calculated from McCumber theory [28]. According to the McCumber theory, the absorption and stimulated emission cross-sections are related by:
se ( l) 5 sa ( l) exp [(´ 2 hy ) /kT ]
(2)
where sa and se are absorption and stimulated emission cross-sections, respectively, the h is the Planck constant, k is the Boltzmann constant, y is the photon frequency, ´ is the net free energy required to excite one Er 31 from the 4 I15 / 2 state to 4 I13 / 2 at temperature T. ´ was determined
Fig. 10. Absorption and stimulated emission cross-section of Er 31 in 50SiO 2 –50PbF 2 glass.
317
using the procedure provided in Ref. [29]. Fig. 10 illustrates the absorption and stimulated emission cross-sections of Er 31 in 50SiO 2 –50PbF 2 glass. The stimulated emission cross-section is proportional to the refractive index of the glass host [s | (n 2 1 2)2 /n] [12]. Since the glass host has a much larger refractive index, it is expected that Er 31 in 50SiO 2 –50PbF 2 glass is capable of providing large stimulated emission cross-section at the 1.55 mm bands. The emission parameters FWHM, the peak of the stimulated emission cross-section (s peak ) and FWHM3 e s epeak of Er 31 in different glass hosts are listed in Table 1 in order to compare the emission properties of Er 31 . The gain bandwidth of optical amplifier can be evaluated by FWHM3 s peak product. The bigger the product, the better e the property. From Table 1, It is seen that the value of FWHM, s peak and FWHM3 s epeak for Er 31 in 50SiO 2 – e 50PbF 2 glass is close to those of tellurite and bismuth glasses, and more than those of silicate, phosphate and germanate glasses. Consequently, the glass samples can be used as host material for potential broadband optical amplifier in WDM.
3.4. Lifetimes of 4 I13 / 2 level of Er 31 The lifetime of 4 I13 / 2 level of Er 31 is also an important parameter for optical amplifier. A critical factor in the success of Er 31 -doped fiber amplifier in optical communications is the long lifetime of the metastable state that permits the required high population inversions to be obtained under steady-state conditions using modest pump powers [32]. Fig. 11 shows the compositional dependence 4 31 of the lifetime of I13 / 2 level of Er in 50SiO 2 –(502 x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses. Obviously, the lifetimes of 4 I13 / 2 level of Er 31 increase with increasing PbF 2 content. The lifetime is related mainly to the phonon energy of glass hosts. If neglected, the energy transfer process, in accordance with the simplified rate relation among total lifetime tm , radiative lifetime tr and the non-radiative lifetime tnr , 1 /tm 5 1 /tr 1 1 /tnr , the shortening of the measured lifetime is due to the enhancement of non-radiative decay processes [33]. The larger the phonon energy of the glass host, the higher the probability for non-radiative decay, and consequently the lower the lifetime of 4 I13 / 2 level of Er 31 . On the contrary, the lower the phonon energy of the glass host, the higher the upconversion emission, which is also a disadvantageous property for optical amplifier. Thus the glass host should have medium phonon energy for Er 31 to realize good amplification properties. The increased lifetime in the studied glass samples could be explained as follows: firstly, because the phonon energy of Pb–F bonds is lower than that of Pb–O, with the substitution of PbF 2 for PbO, the non-radiative transition rate of Er 31 decreases, which leads to the increase of the measured lifetimes; secondly,
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319
318
Table 1 Comparisons of FWHM, s peak and FWHM3 s peak of Er 31 in different glass hosts e e Glass
50SiO 2 –50PbF 2
Tellurite
Bismuth
Phosphate
Germanate
Silicate
FWHM (nm) s peak (10 221 cm 2 ) e FWHM3 s peak e Ref.
60 7.4 444 This work
65 7.5 487.5 [30]
79 7.0 554 [11]
37 6.4 236.8 [9]
42 5.7 239.4 [31]
40 5.5 220 [30]
the radiative lifetime calculated by the Judd–Ofelt theory is inversely proportional to refractive index of the glass host, due to the refractive index of the glass samples decreasing with increasing PbF 2 content, and so is the measured lifetime. In addition, free OH 2 in the glass is regarded as one of the important quenching centers in Er 31 -doped glasses which is attributed to the strong stretch vibration of OH 2 at 3600 cm 21 (2.79 mm) [34]. The energy of the 4 I13 / 2 → 4 I15 / 2 transition of Er 31 (|6500 cm 21 ) corresponds to the energy of the second harmonic of the OH 2 stretching vibration. Accordingly, non-radiative relaxation occurs by excitation of two OH 2 vibrational quanta when an Er 31 is coupled to OH 2 . However, the control of OH 2 bands isn’t taken into consideration in our experiment, if OH 2 bands of glass samples are eliminated, lifetimes of 4 I13 / 2 levels of Er 31 will be longer.
4. Conclusions The present oxyfluoride silicate glasses were prepared and analyzed. Er 31 -doped 50SiO 2 –50PbF 2 glass showed a broad emission spectra of 1.55 mm with a large stimulated emission cross-section, long lifetimes of 4 I13 / 2 level of Er 31 and good thermal stability. The gain bandwidth
properties of Er 31 -doped 50SiO 2 –50PbF 2 glass was evaluated by FWHM3 s peak product, and the FWHM3 s epeak e 31 value of Er -doped 50SiO 2 –50PbF 2 glass is 444, which is close to those of tellurite glass (487.5) and bismuth glass (554), is larger than that of silicate glass (220), phosphate glass (236.8) and germanate glass (239.4). The large stimulated emission cross-section of Er 31 -doped 50SiO 2 – 50PbF 2 glass (se 57.4310 221 cm 2 ) results from the large refractive index of the glass host. According to the Judd– Ofelt theory, the FWHM value is related to the V6 parameter and the line strength of Sed , and the larger the V6 parameter and the line strength of Sed , the broader the FWHM. DT has been used as a rough estimate of the glass formation ability or glass stability, and the DT value of Er 31 -doped 50SiO 2 –50PbF 2 glass is 231 8C, which is larger than those of tellurite (141.5 8C), bismuth (170 8C) 4 and fluoride (105 8C) glass. The lifetimes of the I13 / 2 level of Er 31 in the glass hosts were measured, and the lifetimes increase from 4.9 to 9.4 ms with increase of PbF 2 content from 0 to 50 mol%.
Acknowledgements We would like to express our thanks to Dr Nengli Dai for his useful discussion. This work is supported by the Project of the National Nature Science Foundation of China (Grant No 60207006) and Optical Science and Technology of Shanghai (Grant No 022261046).
References
Fig. 11. Compositional dependence of fluorescence lifetime of the 4 I13 / 2 level of Er 31 in 50SiO 2 –(502x)PbO–xPbF 2 (x50, 10, 20, 30, 40, and 50 mol%) glasses.
[1] S. Tanabe, N. Sugimoto, S. Ito, T. Hanada, J. Lumin. 87–89 (2000) 670. [2] S. Tanabe, J. Non-Cryst. Solids 259 (1999) 1. [3] M. Yamada, A. Mori, K. Kobayashi, H. Ono, IEEE Photonics Technol. Lett. 10 (1998) 1244. [4] Y. Ding, S.B. Jiang, B.C. Hwang, T. Luo, N. Peyghambarian, Y. Himei, T. Ito, Y. Miura, Opt. Mater. 15 (2000) 123. [5] N. Sugimoto, Rev. Laser. Eng. 29 (2001) 590. [6] J. Yang, S. Dai, L. Wen, Z. Liu, L. Hu, Z. Jiang, Chin. Phys. Lett. 19 (2002) 1516. [7] S. Tanabe, J. Non-Cryst. Solids 256–257 (1999) 282. [8] S. Tanabe, S. Yoshii, K. Hirao, N. Soga, Phys. Rev. B 45 (1992) 4620. [9] S. Jiang, T. Luo, B.C. Hwang, F. Smekatala, J. Non-Cryst. Solids 263–264 (2000) 364. [10] Y. Ohishi, A. Mori, M. Yamada, H. Ono, Opt. Lett. 23 (1998) 274.
S. Xu et al. / Journal of Alloys and Compounds 361 (2003) 313–319 [11] J. Yang, S. Dai, Y. Zhou, L. Wen, L. Hu, Z. Jiang, J. Appl. Phys. 93 (2003) 977. [12] J.S. Wang, E.M. Vogel, E. Snitzer, Opt. Mater. 3 (1994) 187. [13] F. Guinhos, P. Nobrega, P. Santa-Cruz, J. Alloys Comp. 323–324 (2001) 358. [14] L.L. Neindre, S.B. Jiang, B.C. Hwang, T. Luo, J. Watson, N. Peyghambarian, J. Non-Cryst. Solids 255 (1999) 97. [15] M.G. Drexhage, O.H. El Bayoumi, C.T. Moyniyan, J. Am. Ceram. Soc. C 65 (1982) 168. [16] D.R. Macfarlane, J.S. Javorniczky, P.J. Newman, D.J. Booth, V. Bogdanov, J. Non-Cryst. Solids 184 (1995) 249. [17] J.S. Wang, E.M. Vogel, E. Snitzer, J. Non-Cryst. Solids 178 (1994) 109. [18] B.R. Judd, Phys. Rev. 127 (1962) 750. [19] S. Ofelt, J. Chem. Phys. 37 (1962) 511. [20] S. Tanabe, T. Ohyagi, N. Soga, T. Hanada, Phys. Rev. B 46 (1992) 3305.
319
[21] H. Takebe, Y. Nageno, K. Morinaga, J. Am. Ceram. Soc. 78 (1995) 1161. [22] S. Tanabe, J. Appl. Phys. 73 (1993) 8451. [23] L. Pauling, J. Am. Chem. Soc. 51 (1929) 1010. [24] A. Mori, Y. Ohishi, S. Sudov, Electron. Lett. 33 (1997) 863. [25] S. Shen, M. Naftaly, A. Jha, Proc. SPIE 3489 (1999) 103. [26] S. Tanabe, T. Hanada, J. Non-Cryst. Solids 196 (1996) 101. [27] M.J. Weber, Phys. Rev. 157 (1967) 262. [28] D.E. McCuber, Phys. Rev. A 134 (1964) 299. [29] W.J. Miniscalo, R.S. Quimby, Opt. Lett. 16 (1992) 258. [30] X. Zou, T. Izumitani, J. Non-Cryst. Solids 162 (1993) 68. [31] H. Lin, E.Y.B. Pun, S.Q. Man, X.R. Liu, J. Opt. Soc. Am. B 18 (2001) 602. [32] R. Rolli, A. Chiasera, M. Montagna, Proc. SPIE 4282 (2001) 109. [33] L. Petit, T. Cardinal, J. Videau, G. Flem, Y. Guyot, G. Boulon, M. Couzi, T. Buffeteau, J. Non-Cryst. Solids 298 (2002) 76. [34] X. Feng, S. Tanabe, T. Hanada, J. Non-Cryst. Solids 281 (2001) 48.