Crystal field strength in C-type cubic rare earth oxides

Crystal field strength in C-type cubic rare earth oxides

Journal of Alloys and Compounds 341 (2002) 82–86 L / locate / jallcom Crystal field strength in C-type cubic rare earth oxides ¨ ¨...

284KB Sizes 0 Downloads 11 Views

Journal of Alloys and Compounds 341 (2002) 82–86

L / locate / jallcom

Crystal field strength in C-type cubic rare earth oxides ¨ ¨ a,b , *, Mika Lastusaari b,c Elisabeth Antic-Fidancev a , Jorma Holsa a

´ de l’ Etat Solide, CNRS, UMR 7574, 11, rue Pierre et Marie Curie, F-75231 Paris Cedex 05, France ENSCP, Laboratoire de Chimie Appliquee b University of Turku, Department of Chemistry, Laboratory of Inorganic Chemistry, FIN-20014 Turku, Finland c Graduate School of Materials Research, Turku, Finland

Abstract Luminescence of the Eu 31 -doped cubic C-type rare earth (RE) oxides, RE 2 O 3 :Eu 31 (RE5Eu, Gd, Lu, Y, In and Sc) powder samples at 77 and 298 K was investigated under UV, argon ion and dye laser excitation. Intense transitions from the 5 D 0 to 7 F 0 – 4 levels of the ground 7 F multiplet were observed between 575 and 720 nm. The complete lifting of the 7 F J level degeneracy as well as the absence of transition selection rules imposed by the group theory are consistent with the C 2 point symmetry of the RE 31 site. Only a few lines originating from Eu 31 ions in the high symmetry S 6 site were observed. The crystal field (c.f.) analysis conducted on the basis of the 18–21 7 F 0 – 4 c.f. levels yielded satisfactory results despite the high number (14) of the B kq and S kq parameters. The strength of the c.f. effect increases in the RE series with decreasing ionic radius of the RE 31 host cation. This results from the increased electrostatic effect of the host lattice on the Eu 31 ion.  2002 Elsevier Science B.V. All rights reserved. Keywords: Luminescence; Crystal field; Rare earth oxides; Europium

1. Introduction The rare earth (RE) oxides, RE 2 O 3 , doped with Eu 31 and Tb 31 ions, are not only important phosphors but also precursors for other, widely used luminescent materials [1,2]. The 4f N energy level structures of the Eu 31 ions doped in RE sesquioxides have been studied rather extensively in the past [3–7] but recently new interest has been aroused because of the need to replace the current luminescent lighting materials with new ones [8]. The present phosphors have been optimized for the use in tricolor low pressure mercury fluorescence tubes and do not necessarily match with the excitation from non-toxic rare gases. The knowledge of the luminescence properties of the RE oxides should thus be complemented with new studies. For theoretical studies, the cubic C-type oxides RE 2 O 3 (RE5Eu 2 Lu, as well as Y, In and Sc) offer a tempting possibility for an assessment of the evolution of luminescence properties because of the extended structural isomorphism [9–11] which is rarely encountered even in the RE series. Also the structure–property relationships as far as the luminescence and the energy level schemes are

*Corresponding author. Tel.: 1358-21-333-6737; fax: 1358-2-3336730. ¨ ¨ E-mail address: [email protected] (J. Holsa).

concerned are of interest because of the high efficiency of these RE oxides as phosphor materials. In this work, the luminescence spectra of the Eu 31 ion in selected C-type rare earth oxides were analyzed in detail. The experimental crystal field (c.f.) energy level schemes were simulated by using a phenomenological model based on the c.f. splittings of the ground 7 F J multiplet. The strength of the c.f. effect was calculated and related to the structural stresses due to doping of the host lattice with the Eu 31 ion.

2. Experimental

2.1. Sample preparation The polycrystalline Eu 31 -doped rare earth oxides RE 2 O 3 (RE5Gd, Y, Lu, Sc as well as In) were prepared by heating the corresponding oxalate hydrates in air at 1100 8C. The RE 31 host cation was partially replaced by a small amount of the Eu 31 ion, nominally 2 mole %. In order to obtain a homogeneous distribution of the dopant, the RE oxide mixtures were dissolved first in aqueous hydrochloric acid and then co-precipitated as oxalate hydrates by oxalic acid. The distribution of the dopant was assumed random and uniform which, however, doesn’t exclude the statistical possibility of dopant pairs.

0925-8388 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved. PII: S0925-8388( 02 )00073-7

E. Antic-Fidancev et al. / Journal of Alloys and Compounds 341 (2002) 82 – 86

The samples were checked with routine X-ray powder diffraction analysis but no impurity phases were detected. The spectroscopic measurements revealed the complete solid solubility between the Eu 31 dopant and the RE 31 host cation and the absence of impurity phases.

2.2. Spectroscopic measurements The luminescence of all RE 2 O 3 :Eu 31 samples was excited originally by UV radiation emitted by a 150 W Hg lamp centered around 300 nm by a wide band filter. The wavelength region chosen corresponds to the strongly absorbing charge transfer band of the Eu 31 ion. Alternatively, the different blue lines (454.4 and 457.9 nm) of a Spectra Physics 164 CW Ar 1 ion laser were used to excite Eu 31 in Gd 2 O 3 , Y 2 O 3 , Lu 2 O 3 , and Sc 2 O 3 hosts. Selective 5 31 excitation on the D 0 level of the Eu ion (near 580 nm) in the C 2 site was carried out with a Spectra Physics 375 / 376 continuous wave rhodamine 6G dye laser. The emission attributed to the 5 D 0,1 → 7 F 0 – 4 transitions between 515 and 720 nm [12] was dispersed by a 1-m Jarrell-Ash single monochromator equipped with standard photomultiplier detection. All measurements were performed at liquid nitrogen (77 K) and room (298 K) temperatures. The resolution of the equipment was better than 1.0 cm 21 . Exceptionally, emission of Eu 2 O 3 with Ar 1 ion laser excitation at 514.5 nm was recorded at room temperature while the Raman spectra of RE 2 O 3 were studied [13].


phism. In addition to the hexagonal and monoclinic forms, the smaller RE 31 ions from Eu to Lu as well as Y, Sc and In form a cubic polymorph, the C-type RE 2 O 3 , (space group Ia3; no. 206; Z516 [14]). In this structure the RE 31 ions enter into two crystallographically non-equivalent sites with six-fold coordination: one possessing C 2 point symmetry with the RE 31 ion in the center of a distorted cube with two oxygen vacancies on one face diagonal. The high symmetry S 6 site possesses inversion symmetry and, this time, the two oxygen vacancies lie on one body diagonal (Fig. 1).

3. Results and discussion

3.1. Analysis of emission spectra Under conventional, broad band UV excitation to the charge transfer band of the Eu 31 ion, the C-type RE 2 O 3 :Eu 31 powder samples (RE5Gd, Y, Lu, In, and Sc) showed intense red luminescence between 575 and 715 nm (Fig. 2). The luminescence spectra were composed of

2.3. Structural considerations Despite the rather similar chemical properties and ionic radii, the RE oxides (RE 2 O 3 ) show extensive polymor-

Fig. 1. Schematic presentation of the two six-coordinated RE 31 sites in cubic C-type yttrium oxide, Y 2 O 3 .

Fig. 2. Characteristic 5 D 0 → 7 F 0 – 4 transitions of the Eu 31 ion in the RE 2 O 3 :Eu 31 (RE5Gd, Lu, Sc; x Eu 50.02) series.


E. Antic-Fidancev et al. / Journal of Alloys and Compounds 341 (2002) 82 – 86

The evolution of the 5 D 0 → 7 F 0 – 4 emission spectra as far as the peak intensities and positions are concerned was found smooth from one RE 2 O 3 host to another. However, detailed analysis of the 5 D 0 level position revealed that no clear nephelauxetic effect, i.e. a monotonous decrease in the 5 D 0 level energy with the decreasing ionic radius of the host cation, could be observed since the 5 D 0 level position in the In 2 O 3 host deviates from the general trend. A possible explanation to this deviation may involve the 5s 2 lone pair of the In 31 electron configuration which is significantly different from that of the RE 31 ions.

3.2. Simulation of 7 FJ ( J50 – 4) level schemes

Fig. 3. Schematic energy level structure of the Eu 31 ion in the C 2 site of Gd 2 O 3 .

groups of sharp lines that could easily be attributed to transitions from the singlet 5 D 0 level to the c.f. com7 31 ponents of the F 0 – 4 levels of the Eu ion (Fig. 3). Each luminescence spectrum is dominated by one very intense line belonging to the 5 D 0 → 7 F 2 transition group. With the 5 7 exception of an intense magnetic dipole induced D 0 → F 1 31 transition arising from emission of Eu ions in the high symmetry S 6 sites, the emission could be well assigned to originate from a Eu 31 ion in a single well-defined C 2 site. The emission of Eu 31 ions in the C 2 site in several RE oxides is rather well documented [3–7]. However, the construction of the 7 F J energy level scheme for the C 2 site is complicated by the emission, though weak, from the S 6 site as well as weak lines of vibronic origin. Moreover, some weak lines in the range between 515 and 600 nm indicate emission from the 5 D 1 levels. The absence of 5 emission from the higher excited D 2 level (around 21 500 cm 21 ) and the weakness of the emission from the 5 D 1 level (around 19 200 cm 21 ) even at low Eu 31 concentrations is due to efficient multiphonon de-excitation processes. In fact, the IR and Raman studies [13] have established the presence of rather high energy phonons (of the order of 700 cm 21 ). The energy differences between the 5 D 0 and 5 D 1 as well as between the 5 D 1 and 5 D 2 levels amount to some 2000 cm 21 that may easily be covered by three phonons. The single 5 D 0 → 7 F 0 transition has significant intensity since the RE 31 ion occupies a site of C 2 point symmetry for which (in general for C s , C n , and C nv ) the free ion selection rules forbid the 0–0 transition breakdown [15].

The complete c.f. energy level scheme of the Eu 31 ion (4f 6 configuration) consists of 3003 Stark components. The c.f. simulations are, however, usually based on the 49 7 F JM set of levels [7]. The truncation of the basis set from 3003 to only 49 is justified by two facts: first, the c.f. operator mixes only levels with the same multiplicity ( 7 F is the only septet [15]) and second, the energy gap between the ground term 7 F and the next excited term 5 D is large, close to 12 000 cm 21 . Both facts minimize the mixing of wave functions yielding essentially pure 7 F JM wave functions. This truncation also enables to put the 300333003 square matrix down to a 49349 one. No further simplification owing to division to submatrices of lower dimension can be achieved because of the low C 2 point symmetry of the RE 31 site in RE sesquioxides. The one electron c.f. Hamiltonian can be expressed as a sum of products between the real and imaginary (B kq and S qk , respectively) parts of the c.f. parameters and functions (C qk ) related to spherical harmonics (Y qk ) [16]: q5k

O O hB fC 1s 2 1d C k q

H cf 5

k q


k 2q


k q52k

1 iS kq f C kq 2s 2 1d 2q C k2q g j


For C 2 symmetry, the c.f. Hamiltonian H cf (C 2 ) contains 15 parameters [15] — nine real and six imaginary (Table 2 1). However, the S 2 parameter can be fixed to zero with a proper choice of the coordinate axes. From the practical point of view, the c.f. simulations have to be carried out with extreme care since the number of the experimental energy levels is only slightly higher than the number of parameters to be refined. Despite the care taken, it is not impossible that there exist other sets of c.f. parameters giving fitting results with similar figures of merit. The actual fitting of the experimental data was conducted with the aid of the matrix diagonalizing and least squares refinement program IMAGE [17]. The least squares refinement minimized the r.m.s. deviation between the experimental and calculated 7 F 0 – 4 energy level schemes. The c.f. simulations gave satisfactory and consistent

E. Antic-Fidancev et al. / Journal of Alloys and Compounds 341 (2002) 82 – 86


Table 1 Crystal field parameters for RE 2 O 3 :Eu 31 (in cm 21 units) Parameter

Gd 2 O 3


Lu 2 O 3

In 2 O 3

Sc 2 O 3

B 20 B 22 B 40 B 42 S 42 B 44 S 44 B 60 B 62 S 62 B 64 S 64 B 66 S 66 NV

2231(23) 2596(12) 21249(42) 21559(36) 2287(115) 1221(30) 251(116) 29(62) 448(42) 227(84) 786(32) 285(65) 155(30) 129(48) 4160

2253(30) 2651(16) 21381(52) 21487(49) 2132(105) 1329(38) 209(107) 390(69) 512(51) 288(105) 745(37) 282(83) 216(43) 174(75) 4286

2219(17) 2714(9) 21356(33) 21711(26) 2540(40) 1196(26) 448(54) 274(56) 458(31) 222(49) 1018(26) 443(40) 165(30) 261(30) 4678

254(17) 2743(8) 21400(25) 21803(21) 2370(50) 1273(19) 552(35) 398(38) 382(25) 271(68) 959(2) 442(45) 23(26) 254(40) 4818

2110(16) 2843(9) 21563(28) 21760(24) 2370(42) 1414(19) 567(47) 482(38) 523(25) 194(74) 942(24) 452(42) 103(26) 246(31) 5055

results despite the high number of c.f. parameters. Moreover, no large discrepancies between each individual experimental and calculated energy level value pair could be observed. The r.m.s. deviation assumed reasonable values for all the oxide hosts studied. The sets of c.f. parameters (Table 1) are characterized by very high B 40 , B 42 and B 44 values which indicates the importance of the middle-range c.f. interactions [18] in the RE 2 O 3 lattice. In contrast, the long- and short-distance c.f. interactions are of minor importance. The deviation of the RE 31 site symmetry from the next higher one, i.e. C 2v , is clear as the significantly non-zero values of the imaginary parts (S kq ) indicate. Due to the rather restricted set of 7 F JM levels and, especially, the lack of the 7 F 5,6 levels, the reliability of the 6 th rank parameters (B 60 , B 62 , S 62 , B 64 , S 64 , B 66 , and S 66 ) as well as, presenting a more general problem, that of the imaginary parts, S 42 and S 44 , is less than that of the 2 nd and 4 th rank parameters. This can be seen in the higher esd values obtained for these parameters after the least squares fitting procedure (Table 1). The c.f. strength parameter N v (calculated according to [19]) increases regularly with the decreasing ionic radius of the RE 31 (RE5Gd–Lu, Y, In and Sc) host cation (Fig. 4). The conventional wisdom predicts — based on the extended electrostatic point charge (PCEM) model (Eq. (2)) [20] — correctly the present experimental observation. According to the PCEM model, the B kq (and S kq ) parameters can be expressed as products between the radial integrals kr k l [21] and the lattice sums A kq which depend only on the doping ion and the host lattice, respectively. B kq 5 t 2k s1 2 s kd A qk kr k l

Eu 31 ion in the RE 2 O 3 series, the B qk (and S qk ) parameters follow the strong increase of the lattice sums A kq with decreasing ionic radius of the host cation. Similar evolution of the A kq lattice sums was found for the RE oxychloride series [22].

4. Conclusions The 7 F 024 energy level schemes of the Eu 31 -doped cubic C-type rare earth (RE) oxides, RE 2 O 3 :Eu 31 (RE5 Eu, Gd, Lu, Y, In and Sc), were successfully simulated with a phenomenological model despite the high number of c.f. parameters (14) and low number of experimentally determined energy level values (18–21). The c.f. parameter sets obtained are coherent and follow the predictions of the extended electrostatic point charge model. Future studies will concentrate on explaining in detail evolution of c.f. parameters.

(2) k

Since the radial integrals kr l (subject to corrections for the expansion of the 4f wave functions as well as for the shielding of the 6s, 5d and 5p electrons in solid state with the factors t and s k , respectively) are constants for the

Fig. 4. Evolution of the crystal field strength parameter N V in the RE 2 O 3 :Eu 31 series.


E. Antic-Fidancev et al. / Journal of Alloys and Compounds 341 (2002) 82 – 86

Acknowledgements Financial support from the Graduate School of Materials Research (Turku) and the University of Turku (M.L.) as well as from the Academy of Finland (project 48645) (J.H.) is acknowledged.

[8] [9] [10] [11] [12] [13] [14]

References [1] G. Blasse, B.C. Grabmaier, Luminescent Materials, Springer, Berlin, 1994. [2] S. Shionoya, W.M. Yen (Eds.), The Phosphor Handbook, CRC Press, Boca Raton, FL, 1999. ˆ [3] O.L. Malta, E. Antic-Fidancev, M. Lemaıtre-Blaise, A. MilicicTang, M. Taibi, J. Alloys Comp. 228 (1995) 41. ˆ [4] E. Antic-Fidancev, J. Aride, M. Lemaıtre-Blaise, P. Porcher, M. Taibi, J. Alloys Comp. 188 (1992) 242. [5] M. Faucher, J. Dexpert-Ghys, Phys. Rev. B24 (1981) 3138. [6] J. Dexpert-Ghys, PhD Thesis, Universite´ de Paris-sud, 1979. ¨ [7] C. Gorller-Walrand, K. Binnemans, in: K.A. Gschneidner, L. Eyring (Eds.), Rationalization of Crystal-Field Parameterization, Handbook Phys. Chem. Rare Earths, Vol. 23, Elsevier, Amsterdam, 1996, Chapter 155.

[15] [16] [17]

[18] [19] [20] [21] [22]

¨ T. Justel, J.-C. Krupa, D.U. Wiechert, J. Lumin. 93 (2001) 179. M. Faucher, J. Pannetier, Acta Crystallogr. B 36 (1980) 3209. M. Marezio, Acta Crystallogr. 20 (1966) 723. A. Fert, Bull. Soc. Fr. Miner. Crist. 85 (1962) 267. G.H. Dieke, Spectra and Energy Levels of Rare Earth Ions in Crystals, Wiley Interscience, New York, 1968. E. Antic-Fidancev, A. Milicic-Tang, 1995 (unpublished observations). K. Lonsdale (Ed.), International Tables for X-ray Crystallography, Vol. 3, Kynoch, Birmingham, 1962, p. 163. J.L. Prather, NBS Monograph (U.S.), 19 (1961). B.G. Wybourne, Spectroscopic Properties of Rare Earths, Interscience, New York, 1965. P. Porcher, Computer Program IMAGE for the Simulation of d N and f N Configurations Involving the Complex Crystal Field Parameters, Meudon, France, 1989 (unpublished). T.A. Metcalf, D.H. Hopkins, F.S. Richardson, Inorg. Chem. 34 (1995) 4868. F. Auzel, O.L. Malta, J. Phys. (Paris) 44 (1983) 201. R.P. Leavitt, C.A. Morrison, D.E. Wortman, Report TR-1673, Harry Diamond Laboratories, Adelphi, MD, 1975. A.J. Freeman, J.P. Desclaux, J. Magn. Magn. Mater. 12 (1979) 11. ¨ ¨ R.-J. Lamminmaki, ¨ ´ J. Holsa, M. Lastusaari, P. Porcher, R. Saez Puche, J. Alloys Comp. 300–301 (2000) 45.