Delayed luminescence of Ce3+ doped Y2SiO5

Optical Materials 26 (2004) 107–112 www.elsevier.com/locate/optmat

Delayed luminescence of Ce3þ doped Y2SiO5 Tuomas Aitasalo

a,b,*

, Jorma H€ ols€a a, Mika Lastusaari a, Janina Legendziewicz c, Janne Niittykoski a,b, Fabienne Pelle d

a

d

Department of Chemistry, University of Turku, FIN-20014 Turku, Finland b Graduate School of Materials Research, Turku, Finland c Faculty of Chemistry, University of Wrocław, 14 F. Joliot-Curie, PL-50-383 Wrocław, Poland Materiaux Inorganiques, UMR 7574, CNRS, 1, Place A. Briand, F-92195 Meudon Cedex, France Available online 19 December 2003

Abstract The aim of the present work was to study the luminescence properties of the Ce3þ (xCe ¼ 0:01) doped Y2 SiO5 . The compounds were prepared by a sol–gel reaction between yttrium and cerium nitrate hydrates as well tetraetoxysilane. The X-ray powder diffraction revealed that all samples were of the monoclinic high temperature X2 form. Two strong luminescence bands were observed at 395 and 427 nm with decay times of 43 and 49 ns, respectively. In addition, two other weak bands with fine structure were observed at 484 and 577 nm. The decay times of the latter bands were about 700 and 800 ls, respectively. The excitation spectra had three bands at ca. 270, 300 and 360 nm. The weak emission bands at 484 and 577 nm were tentatively explained to be due to the Ce4þ charge transfer luminescence associated with oxygen vacancies created by the reducing preparation conditions.
2003 Elsevier B.V. All rights reserved. Keywords: Cerium; Yttrium oxyorthosilicate; Charge transfer luminescence

1. Introduction The cerium (Ce3þ ) doped rare earth oxyorthosilicates (R2 SiO5 ) have been studied widely due to their cathodoluminescence, storage phosphor and scintillator properties [1–6]. The parity allowed wide band luminescence of the Ce3þ ion combined with the stable and rigid silicate host lattice offers an excellent combination for efficient luminescence. The Ce3þ doped X2 -type yttrium oxyorthosilicate has been shown to have two strong 5d1 fi 4f1 emission bands with maxima at about 395 (25,300) and 430 nm (23,250 cm1 ) with 40 ns decay times [1,4,5]. The emission band with two maxima the energy difference of which is about 2000 cm1 is due to the splitting of the 2 F term of the ground 4f1 configuration of the Ce3þ ion into two spin–orbit coupled levels, 2 F7=2 and 2 F5=2 [7]. In addition to the regular strong 5d1 fi 4f1 emission at 416 (24,000) and 454 nm (22,000 cm1 ), weak emission bands at 484 and 582 nm have been observed at 372 K for the Ce3þ doped X1 -type Gd2 SiO5 [8]. In this *

Corresponding author. Fax: +358-2-3336700. E-mail address: tukuai@utu.fi (T. Aitasalo).

0925-3467/$ - see front matter
2003 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2003.11.006

host also an exceptionally wide (225 nm) emission band peaking at 430 and 480 nm has been observed with a microsecond range delay time at 77 K [9]. The Ce3þ doped LuSiO5 has been described to have UV and daylight excited afterglow [10,11]. Afterglow and storage phosphor properties have been ascribed to be due to intrinsic traps, which probably are oxygen vacancies in the rare earth oxyorthosilicate lattice [4,11–14]. In this work, the Ce3þ doped high temperature X2 type yttrium oxyorthosilicates (X2 –Y2 SiO5 ) were prepared by a sol–gel method. The luminescence properties of these materials were studied and two delayed emission bands were observed for the first time to the knowledge of the authors. The origin for the delayed luminescence was discussed. 2. Experimental 2.1. Sample preparation The polycrystalline Ce3þ doped yttrium oxyorthosilicates (Y2 SiO5 ) were prepared by a sol–gel method as described earlier [15–18] using the aqueous solutions of

108

T. Aitasalo et al. / Optical Materials 26 (2004) 107–112

yttrium and cerium nitrate hydrates (R(NO3 )3 Æ xH2 O, R ¼ Y, Ce) as well ethanol solution of tetraetoxysilane (Si(OC2 H5 )4 ). The mixed solution was evaporated until a clear gel appeared. The gel was subsequently dried at 110 C for 16–20 h. The dried gel was annealed at 1250 C for 3 h in the dynamic N2 + 12 volume per cent H2 gas sphere. The nominal concentration of the Ce3þ ion was 1 mol % of the total amount of the rare earth ions. The sol–gel method of preparation was preferred since in this way the homogeneous dopant distribution in the host lattice would be achieved most easily. The phase and structural purity of all samples was checked routinely by X-ray powder diffraction. All the samples were of the X2 structure as should be expected due to the rather high final annealing temperature. 2.2. Spectroscopic measurements The low resolution emission and excitation spectra were measured at 77 and 300 K with a Perkin Elmer LS5 spectrometer and an Aminco SPF-500 spectrophotometer. Wide band UV excitation was provided by a 8.3 and 300 W xenon lamps centered at 300 and 360 nm with band widths of 15 and 20 nm, respectively. The delayed luminescence in the ms range was excited by UV radiation (a pulsed 8.3 W xenon lamp) or by the third harmonic (355 nm) of a 501-DNS 720 Nd:YAG laser (BM Industrie). The spectra were collected by the Perkin Elmer spectrometer (UV excitation) with a F/3 MonkGillieson type monochromator or by a 0.460 m Jobin– Yvon HR460 monochromator with a EMI 9558 QBM photomultiplier (laser excitation). The emission decay times were measured using the latter setup. 2.3. Structural considerations The oxyorthosilicates of the large rare earth ions (La– Tb) possess a monoclinic X1 type structure with the

space group P21 /c (#14) [19], Z ¼ 8 whereas the small ones (Dy–Lu, Sc) have a X2 type structure with the space group B2/b (#15) [19], Z ¼ 8 [20]. Y2 SiO5 crystallizes in the X1 structure when annealed at a temperature below 1190 C and in the X2 structure above this temperature [21]. Both structures have two different rare earth sites of a very low (C1 ) point symmetry. The sites have a rather different coordination to the oxide ions (Table 1, Fig. 1). Not all oxygen ions are bonded to both yttrium and silicon as shown in Table and, moreover, the number of free (bonded only to yttrium) and bridging oxygens (bonded also to silicon) is different in these two structures. The possible defect structure, especially the oxygen vacancies needs to be surveyed more closely since a reducing atmosphere was used in the preparation of yttrium oxyorthosilicates. Similar to zinc borates, Zn4 B6 O13 , [22] the oxygens will be more tightly bound to the tetravalent silicon (trivalent boron) than to trivalent yttrium (divalent zinc) due to the strongly covalent interactions within the SiO4 group. Accordingly, the vacancies are more easily created into the oxygen sites bonded only to the yttrium ions. In addition, similarly to a-alumina there will be a possibility for two different oxygen vacancy centres containing one (a Fþ centre) or two electrons (a F centre) [23,24]. For the electrostatic reasons, the energy of the Fþ centre is higher than the energy of the F centre. Table 1 The rare earth coordination to the oxide ions in oxyorthosilicates, R2 SiO5 [20]

Total coordination number Number of silicon bonded oxygens Number of free oxygens

X1 -structure

X2 -structure

A1 site

A2 site

B1 site

B2 site

9 8

7 4

7 5

6 4

1

3

2

2

Fig. 1. The Y1-O7 and Y2-O6 units of Y2 SiO5 forming a monocapped trigonal antiprism and a distorted octahedron as a coordination polyhedron around the Y3þ ion, respectively.

T. Aitasalo et al. / Optical Materials 26 (2004) 107–112 10

3.1. X-ray powder diffraction

77 K, λex=360 nm

Intensity / Arb. unit

8

The X-ray powder diffraction revealed that all the samples were of the high temperature monoclinic X2 type yttrium oxyorthosilicate form (Fig. 2). Small, but observable, amounts of Y2 Si2 O7 and/or Y2 O3 were observed depending on the preparation conditions. The calculated diffraction patterns used in the phase identification were obtained by the PowderCell for Windows Version 2.4 computer program [25] using structural data from [26].

6

2

em= 4

λ

em= 3

27 nm

95 nm

0.01 0.1

0.2

0.3 0.4 Time / µs

0.5

77 K, λex=300 nm 300 K, λex=300 nm

0 400

450

500

550

600

650

Wavelength / nm

3.2. Luminescence 1

1

The strong parity allowed 5d fi 4f emission bands of the Ce3þ ion under UV excitation were observed at 395 (25,300) and 427 nm (23,420 cm1 ) with decay times of 43 and 49 ns, respectively (Fig. 3). The emission is similar with both 300 and 360 nm excitation. The two bands with an energy difference of 1900 cm1 were resolved much better at 77 K indicating decrease in the homogeneous line broadening and a decrease in the electron lattice phonon interaction. The fine structure between 445 and 500 nm on the spectra is from the xenon lamp. The two observed emission bands are due to the 5d1 fi 4f1 transition of one Ce3þ centre [1,4,5] and the splitting of the main band into two bands is due to the splitting of the 4f1 ground configuration of the Ce3þ ion by the spin–orbit coupling into the 2 F7=2 and 2 F5=2 levels. This interpretation is supported by the observed energy difference being close to 2000 cm1 [7] which value is practically independent of the host lattice. The emission observed in this study has been ascribed to

400 Measured Calculated

300

Intensity / Counts

λ

0.1

4

350

200

100 *

0 20

1

Intensity / Arb. unit

3. Results and discussion

109

24

28

32

36

40

Diffraction Angle 2θ / Degrees

Fig. 2. The experimental and calculated [23,24] X-ray powder patterns of the monoclinic yttrium oxyorthosilicate X2 –Y2 SiO5 . The reflection corresponding to the Y2 O3 impurity phase is marked with an asterisk.

Fig. 3. The UV-excited luminescence spectra of X2 –Y2 SiO5 :Ce3þ at 77 and 300 K (kex ¼ 300 and 360 nm). The inset shows the decay times of the emission at 427 and 395 nm fitted with a first order exponential function.

originate from the Ce3þ B1 site [4] with about 40 ns decay time [5]. The present results are in good agreement with the previous reports. Another emission band, presumably originating from the Ce3þ B2 site has been shortly noted earlier [5] to situate at 500 nm at 298 K with a decay time of 60 ns, but emission spectrum or other details as the splitting of this band due to the spin– orbit split 4f1 ground state has not been presented. The possibility of impurities or defect emission has not been exploited in this previous study, either. A conceivable explanation to the absence of the emission from the other Ce3þ site may be due to the fact that the energies of the 5d-levels of the B2 site are higher than those of the B1 site and owing to an efficient energy transfer from the B2 site to B1 only the emission from the B1 site is observed. On the other hand, the two Y3þ sites in Y2 SiO5 have quite different coordination to oxide ligands although the CNs are 6 and 7, since the average Y–O distances are  respectively. Since the Ce3þ ion is much 2.28 and 2.43 A, larger than the Y3þ host cation it prefers to occupy exclusively the Y3þ site with higher coordination number, i.e. the site with longer R–O bond distances, at least at such a low doping level used in this work. The excitation spectra of the regular emission from the B1 site had three peaks at about 270, 300, and 360 nm, i.e. at 37,000, 33,300, and 27,800 cm1 , respectively. The Ce3þ luminescence occurs at 23,800 cm1 . Since the 5d1 configuration should be split into five different components there is one 5d level missing. The probable explanation to the absence of this band may be either an accidental overlap with another band since the band width of the observed bands is large or the missing 5d level situates at higher energy than that of the highest level observed, i.e. above 37,000 cm1 . Since the overall splitting of the 5d1 configuration is already larger than

T. Aitasalo et al. / Optical Materials 26 (2004) 107–112

10,000 cm1 (usually around 13,000 cm1 in aluminates) [27] in agreement with this study, the former explanation may be a more plausible one. The barycentre of the 5d1 configuration seems to be at quite a low energy, however. The interpretation of the excitation spectra is complicated even further since if there were an efficient energy transfer from the B2 site not observed in emission, the excitation bands corresponding to this site should be visible in the excitation spectrum anyhow. If only one Y3þ site were occupied by Ce3þ , no additional bands could be observed in the excitation spectra as was confirmed experimentally. The distribution of the Ce3þ ions between the two Y3þ sites in Y2 SiO5 needs to be established by further studies, however.

10000

Intensity / Arb. unit

110

λ

em

λ

=4 84 n

em

1000

0.0

0.5

=5 77 n

1.0

m

m

1.5

2.0

2.5

Time / ms

3.3. Delayed luminescence

Fig. 5. The decay times of the UV-excited luminescence bands of X2 – Y2 SiO5 :Ce3þ at 484 and 577 nm at 300 K fitted with a first order exponential function (kex ¼ 360 nm).

In addition to the very fast ns scale emission from a regular Ce3þ site presented above, also two very weak emission bands at 484 (20660) and 577 nm (17330) under UV excitation at 360 nm were observed with delay times in the millisecond range (Fig. 4). The energy difference (3330 cm1 ) between the bands is too large to be due to the Ce3þ ground 4f1 state splitting [7] and thus any emission from one regular intrinsic Ce3þ site, i.e. from the B2 site, should be excluded. With 40 ls delay time both the regular Ce3þ emission at 427 nm as well as the weak emission at 484 and 577 nm were observed. With increasing delay time the intensity of the long wavelength emission increased until with a delay time of 80 ls this emission was the only one to be observed despite the huge difference in the original intensities. Accordingly, the decay times of UV excited delayed emission bands were 726 (kem ¼ 484 nm) and 846 ls (kem ¼ 577 nm) (Fig. 5) which are too different to originate from the same centre. These decay times are also much too long to correspond to those of the 5d1 fi 4f1 transitions of

any regular Ce3þ sites [5]. The decay time values measured with Nd:YAG laser excitation (kex ¼ 355 nm) were quite similar to the values measured with the UV excitation. The excitation spectra of the delayed luminescence had three peaks at about 270, 300 and 360 nm (Fig. 6) with all delay times and this indicates an efficient energy transfer between the regular Ce3þ site discussed above and the delayed luminescence centre. The origin of this delayed luminescence is thus still unknown but it may be clear that emission from any Ce3þ site may be excluded. A careful survey of the literature published on the rare earth oxyorthosilicate systems revealed that similar, even at the quite same wavelengths emission bands at 484 and 582 nm have been observed from the Ce3þ doped Gd2 SiO5 at 372 K [8]. Although no structural data was given, it may be assumed that the Gd2 SiO5 samples used in that study are of the high temperature monoclinic X2 oxyorthosilicate form because of the high 12

300 0.1 ms

λ em

Intensity / Arb. unit

Intensity / Arb. unit

1.0 ms 100

1.0 ms

2.0 ms

1.5 ms

1.5 ms

2.0 ms

200

nm

0.5 ms

8

75

0.5 ms

=5

0.1 ms 200

λe

m

4

=4 85 n

m 100

0

λem=430 nm

0 400

450

500

550

600

650

Wavelength / nm

Fig. 4. The UV excited (kex ¼ 360 nm) delayed luminescence spectra of X2 –Y2 SiO5 :Ce3þ with delay times of 0.1–2.0 ms at 300 K.

240

280

320

360

400

440

Wavelength / nm

Fig. 6. The excitation spectra of X2 –Y2 SiO5 :Ce3þ at 300 K (kem ¼ 430 nm: no delay time; kem ¼ 484 and 577 nm: 100 ls delay time).

T. Aitasalo et al. / Optical Materials 26 (2004) 107–112

temperatures involved in the preparation method. The preparation method used in that study involves the use of air atmosphere and a very fast cooling rate. Both of these suggest that there is a considerable amount of lattice defects as well as tetravalent cerium present in those samples. In [8] the emission bands at 484 and 582 nm have been explained by a charge transfer from O2 to Ce3þ resulting in the formation of the Ce2þ ion. This explanation is impossible since the energy required for that kind of process would be much higher than what is available. It is well known that the ligand to Ce3þ charge transfer is among the highest in the R3þ series whereas the energy required to promote one electron from Ce3þ to create the Ce4þ ion is definitely the lowest one. Moreover, the formation of the Ce2þ ion is incompatible with the Gd2 SiO5 lattice which due to the size effect between the larger Ce3þ and the much smaller Gd3þ one should prefer inclusion of the even smaller Ce4þ ion into the lattice [28]. The similarities in the Ce3þ emission in Gd2 SiO5 [8] and in Y2 SiO5 at 484 and 582 nm and at 484 and 577 nm, respectively, suggest that these emissions are somehow related to the lattice defects created in the preparation. Although it may be considered certain that in Gd2 SiO5 there are some tetravalent Ce4þ ions present, in our case, the creation of the Ce4þ ions in Y2 SiO5 is more complicated. Our Ce3þ doped Y2 SiO5 samples were prepared in reducing gas sphere in order to prevent the oxidation of Ce3þ to Ce4þ . It is thus safe to assume that some oxygen vacancies are created in the Y2 SiO5 lattice. Since these vacancies are very positive when compared to the regular O2 site they can electrostatically attract electrons from the nearby ions and form the positive Fþ and neutral F centres. Similar defect centres are known to exist e.g. in alumina [23,24]. The only source of electrons available in Ce3þ doped Y2 SiO5 are the Ce3þ ions which are known to exist in the tetravalent state. The charge imbalance is stabilized by a transfer of electrons from the Ce3þ centres resulting in a formation of the Ce4þ centres. By this mechanism the creation of both lattice defects and Ce4þ ions in Y2 SiO5 can be realized. The two bands at 484 and 577 nm can be related to the two types of defects present, the Fþ and F centre. As for the long emission decay times of the delayed luminescence is close to the ms range, these times are not compatible with the usually very short emission of the colour centre. The origin of the emission should be searched from other sources. According to literature available, a wide delayed emission band peaking at 430 and 480 nm has also been observed from the Gd2 SiO5 :Ce3þ at 77 K in a microsecond range [9]. The changes in the shape of the spectra (at 430 nm) were explained by emission from two different Ce sites. However, the results are inconsistent with the usually observed short decay time for the regular 5d1 –4f1 transitions of Ce3þ and the shape of the

111

emission band. The reasons for the delayed emission were not discussed in detail [9], however. There is not much published about the charge transfer luminescence of the Ce4þ ion. So far detailed information is available only for the strontium cerate, Sr2 CeO4 [29]. In this compound the Ce4þ emission was observed as a lone band at 465 nm with a life time of 65 and 35 ls at 6 and 298 K, respectively. The decay time of the Ce4þ emission was observed to increase somewhat with lowering temperature. Since the presence of the Ce4þ centres in Y2 SiO5 is more than probable, the other source of delayed luminescence may be the Ce4þ centres. The decay times of delayed luminescence observed in the present investigation, 700 and 800 ls are about an order of magnitude longer than those observed in the fully concentrated Ce4þ system, Sr2 CeO4 , but the Ce4þ concentration in Y2 SiO5 is only a fraction of the original Ce3þ concentration (1 mol %) and thus the decay times are expected to be much longer. The apparent two band structure in the luminescence spectra of the delayed luminescence may be explained to be due to the two principal types of emitting Ce4þ centres: those Ce4þ centres close to the F defect centres and the others close to the Fþ defect centres. The former Ce4þ centres should be related to the emission at lower energy, i.e. at 577 nm, whereas the emission band at higher energy at 484 nm results from the Ce4þ centres close to the Fþ defects. This reasoning is based on the energies required to produce the charge transfer Ce4þ fi Ce3þ luminescence by an electron transfer from a neutral and positively charged defect.

4. Conclusions The delayed luminescence in the Ce3þ doped Y2 SiO5 is due to the Ce4þ charge transfer luminescence in relation with two different oxygen vacancies (F and Fþ centres) created by the preparation at reducing conditions. The charge imbalance is stabilized by capturing of electrons from the Ce3þ centres which results in the formation of the Ce4þ centres. For electrostatic reasons the defect and Ce4þ centres are located close to each other though the distance between these centres is not always the same. This variation explains both the apparent two band structure with fine structure. The defect structure in the Ce3þ doped Y2 SiO5 will be studied in the near future by thermoluminescence, EPR and high-resolution laser excited photoluminescence methods.

Acknowledgements The work was funded by the Academy of Finland, the University of Turku (Turku, Finland), the Graduate

112

T. Aitasalo et al. / Optical Materials 26 (2004) 107–112

School of Materials Research (Turku, Finland), and the European Union Marie Curie program.

References [1] A.H. Gomes de Mesquita, A. Bril, Mater. Res. Bull. 4 (1969) 643. [2] J. Shmulovich, G.W. Berkstresser, C.D. Brandle, A. Valentino, J. Electrochem. Soc. 135 (1988) 3141. [3] J. Reichardt, M. Stiebler, R. Hirrle, S. Kemmler-Sack, Phys. Status Solidi A 119 (1990) 631. [4] A. Meijerink, W.J. Schipper, G. Blasse, J. Phys. D: Appl. Phys. 24 (1991) 997. [5] H. Suzuki, T.A. Tombrello, C.L. Melcher, J.S. Schweitzer, Nucl. Instrum. Methods Phys. Res., Sect. A 320 (1992) 263. [6] G. Blasse, Chem. Mater. 6 (1994) 1465. [7] G.H. Dieke, Spectra and Energy Levels of Rare Earth Ions in Crystals, Wiley Interscience, New York, 1968. [8] C. Shi, Z. Han, S. Huang, G. Zhang, G. Zimmerer, J. Beker, M. Kamada, L. Lu, W.M. Yen, J. Electron. Spectrosc. Relat. Phenom. 101–103 (1999) 633. [9] J. Flor, A.M. Pires, M.R. Davolos, M. Jafelicci Jr., J. Alloys Comp. 344 (2002) 323. [10] P. Dorenbos, C.W.E. van Eijk, A.J.J. Bos, C.L. Melcher, J. Phys.: Condens. Matter 6 (1994) 4167. [11] P. Dorenbos, A.J.J. Bos, C.W.E. van Eijk, J. Phys.: Condens. Matter 14 (2002) L99. [12] D.W. Cooke, B.L. Bennett, R.E. Muenchausen, K.J. McClellan, J.M. Roper, M.T. Whittaker, J. Appl. Phys. 86 (1999) 5308.

[13] D.W. Cooke, B.L. Bennett, K.J. McClellan, J.M. Roper, M.T. Whittaker, J. Lumin. 92 (2001) 83. [14] D. Meiss, J. Reichardt, W. Wischert, S. Kemmer-Sack, Phys. Status Solidi A 142 (1994) 237. [15] J. Wang, S. Tian, G. Li, F. Liao, X. Jing, Mater. Res. Bull. 36 (2001) 1855. [16] J. Lin, Q. Su, J. Alloys Compd. 210 (1994) 159. [17] D.V. Arkhipov, T.I. Khristov, N.V. Popovich, S.S. Galaktionov, N.P. Soshchin, Inorg. Mater. 32 (1996) 459. [18] M. Yin, C. Duan, W. Zhang, L. Lou, S. Xia, J.-C. Krupa, J. Appl. Phys. 86 (1999) 3751. [19] N.F.M. Henry, K. Lonsdale, in: International Tables for X-Ray Crystallography, Vol. 1, Kynoch, Birmingham UK, 1952. [20] J. Felsche, Struct. Bond. 13 (1973) 99. [21] J. Ito, H. Johnson, Am. Mineral. 53 (1968) 1940. [22] A. Meijerink, G. Blasse, M. Glasbeek, J. Phys.: Condens. Matter 2 (1990) 6303. [23] M. Benabdesselam, P. Iacconi, E. Papin, P. Grosseau, B. Guilhot, J. Therm. Anal. 51 (1998) 913. [24] K.H. Lee, G.E. Holmberg, J.H. Crawford, Solid State Commun. 20 (1976) 183. [25] W. Kraus, G. Nolze, PowderCell for Windows, Version 2.4, Federal Institute for Materials Research and Testing, Berlin Germany, 2000. [26] C. Michel, G. Buisson, E.F. Bertaut, C. R. Hebd. Seances Acad. Sci. B 264 (1967) 397. [27] P. Dorenbos, J. Lumin. 99 (2002) 283. [28] R.D. Shannon, Acta Cryst. A 32 (1976) 751. [29] L. van Pieterson, S. Soverna, A. Meijerink, J. Electrochem. Soc. 147 (2000) 4688.