Optical intensities of Pr3+ ions in transparent oxyfluoride glass and glass-ceramic. Applications of the standard and modified Judd-Ofelt theories

Journal of Alloys and Compounds 380 (2004) 167–172 Optical intensities of Pr3+ions in transparent oxyfluoride glass and glass–ceramic. Applications of the standard and modified Judd–Ofelt theories R.T. Génovaa, I.R. Mart´ ına, U.R. Rodr´ ıguez-Mendozaa, F. Lahoza, A.D. Lozano-Gorr´ ınb, P. Núñezb, J. González-Platasc,d,V.Lav ´ ına,∗ aDepartamento de F´ısica Fundamental y Experimental, Electrónica y Sistemas, Universidad de La Laguna, E-38200 La Laguna, Tenerife, Spain bDepartamento de Qu´ımica Inorgánica, Universidad de La Laguna, E-38200 La Laguna, Tenerife, Spain cDepartamento de F´ısica Fundamental II, Universidad de La Laguna, E-38200 La Laguna, Tenerife, Spain dInstituto de Bioorgánica “Antonio González”, Servicio General de Difracción Rx (SIDIX), Av. Astrof´ısico Francisco Sánchez s/n, E-3800 La Laguna, Tenerife, Spain Abstract The optical characterisation of Pr3+ions in transparent SiO2–Al2O3–CdF2–PbF2–YF3based glass and glass–ceramic have been performed. Fromabsorption andemission spectratheoscillator strengthsof the4f2–4f2electronictransitions have beenobtained. Theintensityparameters havebeencalculated using boththeJudd–Ofelt theory andthe modified theorydevelopedby Kornienko,Kaminskiiand Dunina. Acomparison of the experimental oscillator strengths, the spontaneous emission probabilities and the lifetimes of the 3P0level and those calculated using the above theoretical procedures has been performed for both samples. The root mean square deviation found using the standard Judd–Ofelt theory is larger than the value obtained with the modified treatment. © 2004 Published by Elsevier B.V. Keywords: Glass–ceramic; Nanostructures; Pr3+; Optical properties 1. Introduction Transparent oxyfluoride glass–ceramics doped with lanthanide ions have received much attention in the recent literature due to their promising applications in optical devices for telecommunication systems, such as fibers, optical amplifiers and solid-state lasers [1,2]. These matrices can be easily melted and handled in air-atmosphere, after a thermal treatment of the precursor glass the subsequently obtained glass–ceramic still shows a large transparency and the lanthanide ions are mainly incorporated in fluoride nanocrystals (low-energy phonon environments) embedded in the oxide glassy bulk [1–5]. Moreover, they have superior macroscopic (optical, mechanical, chemical, ...) properties compared to the precursor glass [1,2]. The Judd–Ofelt theory [6–9] is the most suitable theory to characterise the intensities of forced electric–dipole tran- ∗Corresponding author. Tel.: +34-922-31-8321; fax: +34-922-31-8228. E-mail address: [email protected] (V. Lav´ ın). sitions between 4f states of lanthanide ions, since the absorption and emission probabilities for an ion-matrix combination are given as a function of a set of three parameters, i.e. the intensity or Judd–Ofelt parameters. Although this theory has been successfully applied to account for the optical properties of lanthanide ions in solid matrices, it is well known that it works less well in the case of Pr3+[8,9]. The problem has always been attempting to fit all the absorption transition intensities with the same set of parameters [8]. The discrepancies come from the theoretical treatments of the configuration mixing, i.e. the opposite-parity excited configurations are considered far away from the ground configuration and are regarded as completely degenerated. Moreover, the energy difference between the states of the ground and the opposite-parity excited configurations is considered constant. In the case of Pr3+ions, the first opposite-parity excited configuration is 4f15d1and it is located at around 60,000cm−1, with an extension of around 20,000cm−1, so it is obvious that the above assumptions do not seem to hold for this trivalent ion. Different modifications of the standard Judd–Ofelt theory have been developed 0925-8388/$ – see front matter © 2004 Published by Elsevier B.V. doi:10.1016/j.jallcom.2004.03.025 168 R.T. G´enova et al./Journal of Alloys and Compounds 380 (2004) 167–172 in order to take into account the energy dependence of the multiplets involved in the transition with the energy of the 4f15d1configuration [10,11]. Tick et al. [12] and Quimby et al. [13] have analysed the optical properties of these oxyfluoride glass and glass–ceramic applying the Judd–Ofelt technique, but using only absorption oscillator strengths. Moreover, they have excluded the 3H4→3P2transition and two different sets of Judd–Ofelt parameters have been found depending on the method of calculation. The purpose of this work is to complete the optical analysis of the Pr3+ions in these matrices using the experimental optical oscillator strengths in absorption and emission. Moreover, a comparison of the optical parameters calculated using the Judd–Ofelt (J–O) parameterisation and the modified theory developed by Kornienko, Kaminskii and Dunina (K–K–D) is given. 2. Experimental The praseodymium doped oxyfluoride samples used in this study have the following chemical composition (in mol%): 30SiO2, 15Al2O3, 29CdF2, 22PbF2,(4−x)YF3and xPrF3(x=0.1 and 1). The samples were obtained by melting the thoroughly mixed composition in an electric furnace at 1050◦C for 2h and quickly casting the melt into a slab by pressing between two stainless steel plates at RT. The transparent glass–ceramic was obtained after a thermal treatment of the precursor glass at 470◦C for 36h. The X-ray analysis was performed in an X’Pert PANalytical diffractometer on a solid sample by using Cu radiation. Optical absorption spectra were recorded in a Perkin-Elmer lambda9 spectrophotometer. Luminescence spectra were obtained by exciting the samples with light coming from a 250W halogen lamp through a 1/4.5m Spex single-grating monochromator. Detection was performed through a 1/4.5m Spex double-grating monochromator using a Hamamatsu R-928 photomultiplier. Spectra were corrected from instrumental response. For lifetime measurements a MOPO pumped laser (pulse width 10ns) was used. For measurements at 13K an APD Cryogenics helium continuous-flow cryostat was used. 3. Theoretical background The experimental measure of the intensity of a lanthanide ion intra-configurational transition is the area under an absorption peak, that can be related to the transition probability through a dimensionless quantity called the oscillator strength. For transitions between multiplets of the 4fNconfiguration in randomly-oriented systems the experimental oscillator strengths corresponding to absorption bands can be obtained using the following formula: f=mc2 πe2N2.303OD(λ) λ2ddλ(1) where mand eare the electron mass and charge, respectively, cthe speed of light, Nthe number of absorbing ions in the unit volume, dthe thickness of the sample and OD(λ)isthe optical density as a function of wavelength. In the standard Judd–Ofelt theory, the electric dipole oscillator strength of a transition from a multiplet |aJto a multiplet |bJ, with an average frequency ν,isgivenby[6–9] the following: fED(aJ,bJ)=8π2mν 3(2J+1)n2hχED × λ=2,4,6 Ωλ|aJ||Uλ||bJ|2(2) where his the Planck constant, χED =n(n2+2)2/9 the field correction factor and nthe refraction index (n=1.75 for these samples), 2J+1 the degeneracy of multiplet |aJ and Ωλare the Judd–Ofelt parameters. The doubly reduced matrix elements ||Uλ|| are almost independent of the host, and those calculated by Weber for Pr3+in LaF3have been used [14]. In order to take into account the energy dependence of the multiplets involved in the optical transition with the energy of the 4f15d1configuration, Kornienko, Kaminskii and Dunina have considered the non-orthogonality of the wave functions to get the following formula [11]: fED(aJ,bJ)=8π2mν 3h(2J+1)n2χED × λ=2,4,6 Ωλ|aJ||Uλ||bJ|2[1 +2α(E(aJ) +E(bJ)−2E(4fN))] (3) where α=(1/2)[E(4f15d1)−E(4fN)], being E(4fN) and E(4f15d1) the mean energies of the ground and first opposite-parity configurations, respectively, and E(aJ) and E(bJ) are the energies of the aJ and bJmultiplets of the 4fNconfiguration (N=2 for the Pr3+ion). The spontaneous emission probability AED(aJ,bJ)ofan electric–dipole transition is given as the following equation: AED(aJ,bJ)=8π2ν2e2n2 mc3fED(aJ,bJ)(4) and the radiative lifetime τris given by τr=1  bJ A(aJ,bJ)(5) where the magnetic–dipole contributions to the spontaneous emission probabilities have been considered. 4. Results and discussion The X-ray diffraction patterns given in Fig. 1 clearly show the structural transformation expected after the thermal treatment of the original glass. Broad curves, typical of R.T. G´enova et al./Journal of Alloys and Compounds 380 (2004) 167–172 169 25 50 75 Intensity (arb. units) 2Theta GC G Fig. 1. X-ray diffraction patterns of oxyfluoride glass (G) and glass– ceramic (GC). structures with no long-range order, are found for the glass (G), whereas for the glass–ceramic (GC) there are also a number of narrow and relatively intense peaks that reveals the coexistence of a crystalline phase and a glassy phase. From the analysis of these peaks, it has been concluded that this partially crystallised sample contains cubic fluoride crystallites of the ␤-PbF2phase (a=5.94Å) and using the Scherrer formula an average size of the crystals around 18nm has been found. The room temperature absorption spectra of the 1mol% Pr3+-doped oxyfluoride glass and glass–ceramic have been measured in the UV-Vis–IR range (Fig. 2). These spectra confirm that the transparency of the glass–ceramic is quite similar to that found in the precursor glass since, apart from those bands associated to the Pr3+ions, both samples show the typical increase in the absorbance in the UV-A region. The bands correspond to intra-configurational 4f2–4f2elec400 500 600 1000 1500 2000 2500 0,3 0,4 0,5 0,6 0,7 25000 20000 10000 5000 3H6 3F2 3F3 3F4 1G4 1D2 3P0 3P1,1I6 3P2 G GC Optical density Wavelength (nm) Energy (cm-1) RT 1 mol% Pr3+ Fig. 2. Absorption spectra of transparent oxyfluoride glass (G) and glass–ceramic (GC) doped with 1mol% of Pr3+at RT. All transitions start from the 3H4ground level to the indicated levels. tronic transitions, starting from the 3H4ground state to the different excited levels of the Pr3+ion, and all the transitions are assumed to be electric dipole in nature [13]. As can be seen, there are practically no differences in the energies of the transitions in the glass and the glass–ceramic, although for the latter the bands are sharper and some of them, such as 3H4→1D2and 3H4→3F3, clearly show structure, giving an experimental evidence that the Pr3+ions have been incorporated in the fluoride nanocrystals. The structural changes in which the Pr3+ions are involved during the thermal treatment of the precursor glass are clearly reflected in the values of the oscillator strengths (fexp) of the transitions corresponding to the absorption bands in the glass–ceramic, obtained using Eq. (1).As shown in Table 1, except for the 3H4→3P0transition, all the oscillator strengths increase after the thermal treatment. A feature already observed in transparent glass–ceramics doped with Er3+[15]. When two peaks could not be separated only one oscillator strength has been assigned to both transitions, although a deconvolution process has been performed in order to calculate the areas of the 3H4→ 3P0,1,2,1I6bands. As pointed out by different authors [9,11,16], it is worth increasing the number of experimental oscillator strengths in the fitting process taking into account those obtained from the transitions observed in the Pr3+luminescence. For this purpose, 0.1mol% Pr3+-doped samples have been used and the emission spectra of the 3P0level in the glass and glass–ceramic have been measured at 13K after exciting the 3H4→3P2transition at 440nm (Fig. 3). The low temperature condition avoids the thermalization effects that could allow the presence of emission from the 3P1level, although the emission from the 1D2level cannot be ignored. This seems to be true for the glass in which a contribution can be 500 600 700 22000 20000 18000 16000 14000 Intensity (arb. units) Wavelength (nm) 13 K 0.1 mol% Pr3+ G 3H5 1D2 3P0 3H4 3F3,3F4 3F2 3H6 3H5 3H4 GC G Energy (cm-1) Fig. 3. Luminescence spectra from the 3P0level in transparent oxyfluoride glass (G) and glass–ceramic (GC) doped with 0.1mol% of Pr3+after exciting at 440nm the 3H4→3P2transition at 13K. Emission from the 1D2level in the glass after exciting at 580nm the 3H4→1D2transition at 13K is also included for comparison. 170 R.T. G´enova et al./Journal of Alloys and Compounds 380 (2004) 167–172 Table 1 Experimental and calculated oscillator strengths (×10−8) of the absorption and emission transitions of Pr3+in transparent oxyfluoride glass and glass–ceramic. The calculated values are obtained using the standard (J–O) and modified (K–K–D) Judd–Ofelt theories [6,7,11]. The intensity parameters Ωλ(×10−20 cm2) and the rms (×10−6) values are also given Transitions Glass Glass–ceramic fexp fJO fKKD,αfixed (α=1×10−5)fKKD,αvar. (α=2.2 ×10−5)fexp fJO fKKD,αfixed (α=1×10−5)fKKD,αvar. (α=2.4 ×10−5) Absorptions 3H4→3H6,3F2336 339 283 173 345 369 311 164 3H4→3F3,3F4932 1198 1193 989 1023 1353 1356 1049 3H4→1G419 18 20 20 31 20 22 22 3H4→1D2214 127 161 218 277 144 184 263 3H4→3P0278 286 280 265 246 321 315 296 3H4→3P1,1I6505 426 449 481 633 479 505 551 3H4→3P21096 451 650 1014 1307 510 740 1247 Emissions 3P0→3H518 0 0 0 19 0 0 0 3P0→3H6112 103 164 279 112 117 187 348 3P0→3F238 22 39 52 16 −912 30 3P0→3F3,3F427 135 150 162 23 152 169 184 Ω20.13 0.21 0.25 −0.06 0.06 0.14 Ω44.09 3.94 3.65 4.60 4.42 4.05 Ω66.33 9.10 13.95 7.18 10.41 17.10 rms 2.53 1.92 1.10 3.18 2.48 1.36 The uncertainty of the measured oscillator strengths is about 5%. rms =(fexp−fcalc)2 number of oscillator stengths−number parameters 1/2. observed as a shoulder in the low-energy side of the 3P0→ 3H6band and also as an isolated band around 700nm corresponding to the 1D2→3H5transition. Moreover, the presence of high-energy phonons, around 900cm−1, associated to bonds with silicon and oxygen in the glass favours the non-radiative de-excitation from the 3P0level to the 1D2 level [4,5]. On the other hand, the incorporation of the Pr3+ions in the cubic fluoride nanocrystals after the thermal treatment gives rise to a complete different 3P0luminescence pattern compared to that of the precursor glass. As can be observed in Fig. 3, the 1D2emission seems to be less important in the glass–ceramic due to a structural-induced decrease of the multiphonon de-excitation. This hypothesis has been confirmed using optical spectroscopic techniques in this material but doped with Eu3+ions. Analysing the vibronic bands before and after the thermal treatment, it has been found a relative decrease of the electron–phonon coupling of the Eu3+ions with the high-energy phonons of the glass phase and a relative increase of this coupling with the low-energy phonons, around 300cm−1, of the fluoride nanocrystals in the glass–ceramic [4,5]. Moreover, this effect may explain the different experimental values observed in the lifetime of the 3P0level, i.e. 19.6␮s in the glass and 26.4␮sin the glass–ceramic, which gives rise to different multiphonon de-excitation probabilities. Considering the oscillator strength of the 3H4→3P0absorption transition, the Eq. (4) and the ratio of areas of the visible emission bands relative to that of the 3P0→3H4 emission, the values of the oscillator strengths of the luminescence transitions from the 3P0level have been obtained (see Table 1). The contribution of the 1D2luminescence to the band centred at 605nm band has been estimated taken into account the 1D2→3H5band area. As a first step in the theoretical calculation, the intensity parameters Ωλhave been obtained by fitting the experimental oscillator strengths of all the transitions measured from the absorption spectra. Applying the Judd–Ofelt theory to the glass and the glass–ceramic, low and negative values for Ω2parameter have been obtained, results that do not have any physical sense within this theory. The largest discordance is found for the 3H4→3P2transition, being this one of the problems when the standard Judd–Ofelt approximation is applied to Pr3+[9]. Only the 3H4→3F2transition has a significant dependency on ||U2||, and since the 3H4→3P2transition has a relatively large oscillator strength and only depends on the ||U4|| and ||U6||,in the fitting process the latter transition forces large values for the Ω4and Ω6parameters. It is usually found in the literature that one of the above transitions has not been taken into account in the fitting process. On the other hand, using the Kornienko–Kaminskii– Dunina treatment, this problem is completely solved when αis allowed to vary in the fitting process, being the rms values much lower than applying the Judd–Ofelt theory. Using the modified theory, large positive values for Ω2are obtained (12.29 for the glass and 19.44 for the glass–ceramic) and the values for Ω4are similar in the glass and the glass–ceramic for both theories. Concerning the Ω6parameter, it is well known that large values have been usually found for Pr3+ions [8] but they are two times larger using the modified treatment compared to the Judd–Ofelt theory R.T. G´enova et al./Journal of Alloys and Compounds 380 (2004) 167–172 171 Table 2 Experimental and calculated spontaneous emission probabilities (s−1) and lifetime (␮s) of the 3P0levelofPr 3+in transparent oxyfluoride glass and glass–ceramic. The calculated values are obtained using the standard (J–O) and modified (K–K–D) Judd–Ofelt theories [6,7,11] Transitions Glass Glass–ceramic Aexp AJO AKKD,αfixed (α=1×10−5)AKKD,αvar. (α=2.2 ×10−5)Aexp AJO AKKD,αfixed (α=1×10−5)AKKD,αvar. (α=2.4 ×10−5) 3P0→3H422077 22643 21783 20203 19536 25524 24567 22497 3P0→3H51436 0 0 0 1512 0 0 0 3P0→3H69526 7467 10732 16454 8940 8467 12247 20163 3P0→3F22979 543 866 1015 1260 −220 256 575      3P0→3F3,3F42145 0 0 0 1790 0 0 0 4796 4613 4279 5389 5187 4750 3P0→1G4– 821 790 733 – 917 883 808 3P0→1D2–01 1 – 00 0 Lifetime 3P0τexp τJO τKKD (α=1×10−5) τKKD (α=2.2 ×10−5) τexp τJO τKKD (α=1×10−5) τKKD (α=2.4 ×10−5) 19.6 27.6 25.9 23.4 26.4 – 23.2 20.5 for both samples. Although positive values for Ω2have been obtained it seems that they are too large if one realises the large degree of uncertainty in its determination. Some authors [9,11,16] have suggested that the estimation of Ω␭, and ␣, parameters using only data obtained from the optical absorption spectrum contains large errors. Therefore, the standard and modified Judd–Ofelt approximations are applied but using both absorption and emission data. When using the modified method the ␣parameter has been fixed to the expected value (α=1×10−5) or it has been allowed to vary. Results are given in Table 1. The best improvement is that now a positive value of Ω2is obtained using both theories for the glass, although a negative value is still found applying the standard method for the glass–ceramic. Moreover, the values for Ω2are much lower than those obtained by the modified method analysing only the absorption data. The values of Ω4are similar to those obtained before taking into account only absorption data, whereas Ω6increases largely when the modified method is applied. In general, the rms deviation decreases when the Kornienko–Kaminskii–Dunina treatment is applied, giving the best agreement when the α parameter varies (see Table 1). In this sense, the failure of the fit for the oscillator strength of the 3H4→3P2transition is again overcome. However, it is worth noting that applying both theories the agreement is not very good for the emission oscillator strengths. Table 2 lists the experimental spontaneous emission probabilities and those calculated using the intensity parameters given in Table 1. The failure of both theories is reflected in these results since a poor agreement is found, although the experimental lifetimes of the level 3P0are similar to those calculated using Eq. (5). On the other hand, Goldner and Auzel [17] have questioned the fact that the energy of the 4f15d1configuration could be considered as a fitting parameter. The values obtained for the αparameter in the above analysis, around 2.3×10−5for both samples, correspond to a 4f15d1energy of around 22,000cm−1. This value for the αparameter is quite similar to that found by Kornienko et al. [11] although, as it was already pointed out by these authors, the expected value is 1 ×10−5. It is clear that this value is not compatible with the experimental results [18], but it is necessary in order to obtain a good fit [19]. This result may be roughly taken as an evidence of the great percentage of configuration mixing in the wave functions of the Pr3+4f2configuration states, especially for high lying energy levels such as 3P2. Moreover, Auzel et al. [19] analysing Er3+-doped glasses have emphasised that the degeneracy of the 2S+1LJlevels involved in the optical transition has to be taken in a more realistic way in order to compare calculated and experimental oscillator strengths and calculate quantum efficiencies. In the Judd–Ofelt theory the degeneracy of the ground multiplet gives rise to a weighting factor, the (2J+1) factor in Eqs. (2) and (3), that assumes that all the Stark levels of the ground state are equally populated [9]. As pointed out by Auzel et al. [19], the silica glasses have a large maximum Stark splitting (around 700cm−1for the 3H4multiplet in this glass) compared with other oxide and fluoride glasses, thus this assumption does not hold and an alternative statistical weight factor should be introduced. Following the method described by Auzel for Er3+-doped glasses [20,21] but analysing the 3P0→3H4emission of the Pr3+ions in thesesamples, ithas beenfound aneffectiveweighting factor [(2J+1)−1]eff ∼ =1.12 (2J+1)−1for the glass, and a lower correction for the glass–ceramic. On the other hand, it is worth noting that in glasses there is a continuous distribution of local structures for the lanthanide ions that covers from weak to strong crystal-fields environments and, even though for larger Stark splitting the Boltzman distribution gives rise to non-equally populated levels, the number of strong crystal-field environments is much less than the medium and weak ones, for which more equally populations exist. However, it is also true that the larger the (odd) crystal-field the larger the transition probability. As a conclusion, from these facts and since the corrections are within the interval of error 172 R.T. G´enova et al./Journal of Alloys and Compounds 380 (2004) 167–172 of the Judd–Ofelt parameters, they have not been used in this study. Finally, it is worth noting that taking into account absorption and emission data the sequence Ω2<Ω 4<Ω 6 is held for every fitting process, as found in several hosts [9]. Moreover, some empirical correlations of the intensity parameter and the local structure of the lanthanide ions have been stated [9]. As a general conclusion, the Ω2parameter increases with the asymmetry of the local structure and with the degree of covalency of the lanthanide–ligand bonds, whereas the Ω6parameter decreases with the degree of covalency. The Ω4parameter is related to bulk properties of the samples [9]. For the matrices involved in this study no clear conclusions can be extracted from the Ω2 parameter, since its large uncertainty, but its slight decrease after the thermal treatment of the precursor glass together with the slight increase of the Ω6parameter could be understood as due to a change of the lanthanide local structure towards a more ionic environment, as it is found in fluoride nanocrystals. 5. Conclusions Applying basic optical spectroscopic techniques to Pr3+-doped oxyfluoride glass and glass–ceramic the experimental oscillator strengths in absorption and emission have been measured. The intensity parameters have been calculated by using both the standard Judd–Ofelt theory and the modified theory developed by Kornienko, Kaminskii and Dunina. As a general conclusion, the modified theory gives a better agreement when the energy difference between configurations is allowed to vary, although the fit of the spontaneous emission probabilities is not very good when applying both theories. Moreover, the change in the values of the intensity parameters from the glass to the glass–ceramic seems to confirm that the lanthanide ions are incorporated in the fluoride nanocrystals after the thermal treatment. Acknowledgements The authors are indebted to Dr. J.J. Romero (Dpto. de F´ ısica de Materiales, Univ. Autónoma de Madrid) for providing the lifetime measurements. This research has been supported by Ministerio de Ciencia y Tecnolog´ ıa (MAT 2001-3363). References [1] M.J. Dejneka, MRS Bull. 23 (1998) 57, and references therein. [2] W. Höland, G. Beall, in: Glass–Ceramic Technology, The American Ceramic Society, 2002. [3] P.A. Tick, N.F. Borrelli, I.M. Reaney, Opt. Mater. 15 (2000) 81. [4] J. Méndez-Ramos, V. Lav´ ın, I.R. Mart´ ın, U.R. Rodr´ ıguez-Mendoza, V.D. Rodr´ ıguez, A.D. Lozano-Gorr´ ın, P. Núñez, J. Appl. Phys. 89 (2001) 5307. [5] J. Méndez-Ramos, V. Lav´ ın, I.R. Mart´ ın, U.R. Rodr´ ıguez-Mendoza, V.D. Rodr´ ıguez, A.D. Lozano-Gorr´ ın, P. Núñez, J. Appl. Phys. 94 (2003) 2295. [6] B.R. Judd, Phys. Rev. 127 (1962) 750. [7] G.S. Ofelt, J. Chem. Phys. 37 (1962) 511. [8] R.D. Peacock, Struct. Bond. 22 (1975) 83. [9] C. Görller-Walrand, K. Binnemans, Handbook on the Physics and Chemistry of Rare Earths, vol. 25, 1998, p. 101; and references therein. [10] F. Auzel, S. Hubert, P. Delamoye, J. Lumin. 26 (1982) 251. [11] A.A. Kornienko, A.A. Kaminskii, E.B. Dunina, Phys. Status Solidi (b) 157 (1990) 267. [12] P.A. Tick, N.F. Borrelli, L.K. Cornelius, M.A. Newhouse, J. Appl. Phys. 78 (1995) 6367. [13] R.S. Quimby, P.A. Tick, N.F. Borrelli, L.K. Cornelius, J. Appl. Phys. 83 (1998) 1649. [14] M.J. Weber, J. Chem. Phys. 48 (1968) 4774. [15] F. Auzel, J. Alloys Compd. 275–277 (1998) 692. [16] M.A. Buñuel, R. Cases, M.A. Chamarro, R. Alcalá, Phys. Chem. Glasses 33 (1992) 16. [17] P. Goldner, F. Auzel, J. Appl. Phys. 79 (1996) 7972. [18] G.H. Dieke, in: Spectra and Energy Levels of Rare Earth Ions in Crystal, Wiley, New York, 1968, p. 52. [19] F. Auzel, D. Meichenin, A. Mendioroz, R. Balda, J. Fernández, J. Lumin. 72–74 (1997) 152. [20] F. Auzel, Doctoral Thesis, Faculté des Sciences de Paris, 1968. [21] F. Auzel, Ann. Télécom. (Paris) 24 (1969) 199.