Crystal structure and luminescence properties of ...

Home

Search

Collections

Journals

About

Contact us

My IOPscience

Crystal structure and luminescence properties of LiYP4O12:Ce3 + phosphor

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2010 J. Phys.: Condens. Matter 22 485503 (http://iopscience.iop.org/0953-8984/22/48/485503) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 91.202.241.133 The article was downloaded on 18/11/2010 at 09:00

Please note that terms and conditions apply.

IOP PUBLISHING

JOURNAL OF PHYSICS: CONDENSED MATTER

J. Phys.: Condens. Matter 22 (2010) 485503 (6pp)

doi:10.1088/0953-8984/22/48/485503

Crystal structure and luminescence properties of LiYP4O12:Ce3+ phosphor T Shalapska1, G Stryganyuk1,2, A Gektin2 , P Demchenko1, A Voloshinovskii1 and P Dorenbos3 1

Ivan Franko National University of Lviv, 8 Kyryla i Mefodiya Street, 79005 Lviv, Ukraine Institute for Scintillation Materials, NAS of Ukraine, 60 Lenin avenue, 61001 Kharkiv, Ukraine 3 Delft University of Technology, 15 Mekelweg, 2629 JB, Delft, The Netherlands 2

E-mail: t [email protected]

Received 22 August 2010, in final form 14 October 2010 Published 17 November 2010 Online at stacks.iop.org/JPhysCM/22/485503 Abstract LiYP4 O12 polyphosphate doped with Ce3+ ions was prepared by the melt solution technique. The crystal structure, interatomic distances, and atom coordination numbers were determined using x-ray powder diffraction. A study of the spectral–kinetic luminescent properties was performed employing excitation with pulsed radiation from a synchrotron (UV–VUV range) and a laboratory x-ray source. The characteristics of Ce3+ luminescence, namely the emission doublet maxima at 3.97 and 3.72 eV and the 4f–5d excitation maxima at 4.20, 5.11, 5.40, 5.65 and 6.55 eV, are discussed in terms of crystal field splitting in a low-symmetry site of the LiYP4 O12 host lattice. The location of the Ce3+ energy levels with respect to the valence and conduction bands of the LiYP4 O12 host is estimated from the temperature dependence of the decay time measured for Ce3+ 5d–4f luminescence. (Some figures in this article are in colour only in the electronic version)

earth ions that prevents concentration quenching of lanthanide luminescence. The present work is devoted to determination of Ce3+ ion energy levels in the LiYP4 O12 matrix based on an analysis of the luminescence properties for LiY0.9 Ce0.1 P4 O12 and on its crystallographic data.

1. Introduction Inorganic phosphates doped with lanthanide ions continually attract the interest of researchers due to the applications of these materials as lighting phosphors, scintillators, lasing media, etc [1]. Among the lanthanide family, Ce3+ ions are the most prominent for scintillation applications due to their fast dipole allowed 5d–4f radiative transitions. Moreover, the knowledge of the location of the Ce3+ energy level in the bandgap of a crystalline matrix makes it possible to estimate the location of energy levels for other rare earth ions in the same matrix using Dorenbos’s empirical relation [2]. A family of ARE1−x Cex P4 O12 compounds (A represents an alkali metal ion and RE stands for a rare earth ion) are rather promising candidates for lighting applications. NaGdP4 O12 :Ce3+ may be a potential scintillator material for x-ray detection [3, 4]. LiPrP4 O12 :Ce3+ can be realized as a compound with an efficient energy transfer for Pr–Ce pairs [5]. In many cases, the high luminescence efficiency is reached in these compounds due to a relatively large distance between rare 0953-8984/10/485503+06$30.00

2. Experimental technique The studies were performed on LiY0.9 Ce0.1 P4 O12 polyphosphate samples grown by a melt solution technique. Li2 CO3 , NH4 H2 PO4 and Y2 O3 , CeO2 oxides were used as starting materials for the synthesis. These reagents were mixed and fired in quartz crucibles at 700 ◦ C for 2 h. X-ray powder diffraction patterns were collected using an automated STOE STADI P diffractometer equipped with a source of Cu Kα1 radiation (monochromatized with curved germanium [1 1 1] crystal) and a linear position sensitive detector (PSD). The diffraction measurements were performed in transmission mode within the 6.000◦ –124.545◦ 2θ -range 1

© 2010 IOP Publishing Ltd Printed in the UK & the USA

J. Phys.: Condens. Matter 22 (2010) 485503

T Shalapska et al

Table 1. Crystallographic data for LiY0.9 Ce0.1 P4 O12 polyphosphate. Phase Structure type Pearson symbol, Z Space group–Wyckoff sequence ˚ Cell parameters (A)

LiY0.9 Ce0.1 P4 O12 LiNdP4 O12 m S 72, 15 C 12/c1– f 8 e2 a = 16.2570(3) b = 7.026 98(10) β = 126.0219(14)◦ c = 9.5755(2) ˚ 3) Cell volume (A 884.72(3) Angular range for data collection (deg) 6.000  2θ  124.545 Number of measured reflections 723 Number of refined parameters 54 Half width parameters U , V , W 0.067(4), −0.030(3), 0.0175(5) Reliability factors RB = |Iobs − Icalc |/|Iobs | 0.1095 RF = |Fobs − Fcalc |/|Fobs | 0.0626 Rp = |yi − yc,i |/yi 0.0798 1/2 Rwp = [wi |yi − yc,i |2 /wi yi2 ] 0.1287 2 1/2 Rexp = [n − p/wi yi ] 0.0363 χ 2 = {Rwp /Rexp }2 12.56

Figure 1. X-ray powder diffraction pattern from LiY0.9 Ce0.1 P4 O12 measured with Cu Kα1 radiation. The experimental data (red circles), and the calculated profile (solid line through the circles) are presented together with the calculated Bragg positions (vertical ticks) and difference curve (blue solid line below). Table 2. Atomic position and displacement parameters for LiY0.9 Ce0.1 P4 O12 .

with a 2θ step of 0.015◦ with the following settings: PSD step 0.480◦ 2θ , accumulation time 300 s/step. Preliminary data processing, and phase analyses from XRD patterns were performed using the STOE WinXPOW (version 2.21) program package [6]. The crystal structure was refined from powder diffraction data using the FullProf.2k program (version 4.00) [7] of WinPLOTR software [8] and a pseudo-Voigt profile function. The crystallographic data were standardized with the program Structure Tidy [9]. Measurements of luminescence excitation and emission spectra as well as the decay kinetics were performed upon excitation with synchrotron radiation from the DORIS storage ring (DESY, Hamburg) using the facility of he SUPERLUMI station at HASYLAB [10]. The measurements were carried out at 10 and 300 K. Emission spectra were measured within the 250–600 nm range using an ARC ‘Spectra Pro 308’ 30 cm monochromator-spectrograph in Czerny-Turner mounting equipped with a CCD detector from Princeton Instruments. Luminescence excitation spectra in the 3.72– ˚ by 20 eV range were scanned with the resolution of 3.2 A a primary 2 m monochromator in 15◦ McPherson mounting using a HAMAMATSU R6358P PMT at the secondary ARC monochromator. The luminescence excitation spectra have been corrected for the incident photon flux by means of Na-salycilate. Luminescence decay kinetics curves were accumulated within a 200 ns time range using a single photon counting technique. The temperature dependence of the 5d–4f Ce3+ luminescence decay time in the 300–700 K range was studied using laboratory spectroscopic equipment involving a single photon counting technique. The x-ray excited emission spectra and decay kinetics were studied upon excitation with a laboratory x-ray source (Cu anode voltage of 40 kV, 0.2 mA current) with pulse duration of 2 ns and 100 kHz repetition frequency.

2

Site Wyckoff x

y

Z

˚ ) Biso (A

Li Ya P1 P2 O1 O2 O3 O4 O5 O6

0.699(7) 0.2016(4) 0.1517(8) 0.0566(8) 0.279(2) 0.117(2) 0.381(2) 0.0804(14) 0.4135(15) 0.0106(14)

1/4 1/4 0.1015(7) 0.3353(7) 0.5376(14) 0.2377(13) 0.0865(13) 0.1621(12) 0.0777(14) 0.3066(14)

0.9(6) 1.22(6) 1.9(1) 1.5(1) 1.6(3) 1.5(3) 1.2(2) 0.3(2) 1.5(3) 1.6(3)

a

4e 4e 8f 8f 8f 8f 8f 8f 8f 8f

0 0 0.1482(4) 0.3610(4) 0.1140(8) 0.1262(8) 0.1638(8) 0.2554(9) 0.4273(7) 0.4359(8)

Y ≡ 0.900(10)Y + 0.100(15)Ce.

3. Results 3.1. X-ray powder diffraction The XRD pattern from a polycrystalline LiY0.9 Ce0.1 P4 O12 sample (figure 1) shows the presence of two phases in the studied sample. The main phase is a polyphosphate of refined LiY0.9 Ce0.1 P4 O12 composition, while the composition and structure of the additional phase could not be determined due to its low concentration in the sample. Crystallographic data for LiY0.9 Ce0.1 P4 O12 are given in table 1, while the final atomic and displacement parameters after Rietveld refinement are listed in table 2. Interatomic distances and coordination numbers are given in table 3. Comparison of experimental and calculated powder patterns is presented in figure 1. The studied LiY0.9 Ce0.1 P4 O12 has the same crystal structure as LiNdP4 O12 [11], and our crystallographic data on LiY0.9 Ce0.1 P4 O12 are in agreement with data for the parent LiYP4 O12 compound [12]. The Y3+ (Ce3+ ) cation is located in the center of a polyhedron of the dodecahedral form with eight fold coordination number. 3.2. Emission and luminescence excitation spectra Upon excitation of the powdered LiY0.9 Ce0.1 P4 O12 sample within the Ce3+ 4f → 5d absorption range at T = 10 K, its 2

J. Phys.: Condens. Matter 22 (2010) 485503

T Shalapska et al

Figure 2. Emission (1— E exc = 5.10 eV, 3— E exc = 8.90 eV) and luminescence excitation (2— E em = 3.87 eV, 4— E em = 2.90 eV) spectra of LiY0.9 Ce0.1 P4 O12 (1, 2) and LiYP4 O12 (3, 4) at T = 10 K. Table 3. Selected interatomic distances (δ ) and coordination numbers (CN) for LiY0.9 Ce0.1 P4 O12 . Atoms

˚ δ (A)

CN

Atoms

˚ δ (A)

CN

Li–2O6 –2O5 Ya –2O2 –2O1 –2O5 –2O6 P1–O5 –O4

1.95(4) 2.02(4) 2.205(15) 2.304(10) 2.434(11) 2.600(12) 1.478(11) 1.561(15)

4

O1–P2 O2–P1 O3–P2 –P1 O4–P2 –P1 O5–P1 O6–P2

1.556(14) 1.561(17) 1.608(15) 1.650(12) 1.542(10) 1.561(15) 1.478(11) 1.432(18)

1 1 2

–O2 –O3 P2–O6 –O4 –O1 –O3

1.561(17) 1.650(12) 1.432(18) 1.542(10) 1.556(14) 1.608(15)

Li–Li Y–Y Li–Y Li–Y Li–Y

5.54(4) 5.563(2) 3.50(5) 3.53(5) 4.838(7)

a

8

4

4

2 1 1

Figure 3. Decay kinetics curves of Ce3+ emission ( E em = 3.75 eV) from LiY0.9 Ce0.1 P4 O12 upon UV (1) ( E exc = 4.2 eV) and band-to-band excitation (2) ( E exc = 17.7 eV) at T = 300 K.

Y ≡ 0.900(10)Y + 0.100(15)C.

bands for Ce3+ luminescence from LiY0.9 Ce0.1 P4 O12 measured at T = 300 K are located at the same energies but the bands are broadened as compared with the case of T = 10 K. Figure 2 (curves 3 and 4) shows the emission and excitation spectra for the intrinsic luminescence attributed to self-trapped excitons in the pure LiYP4 O12 host at T = 10 K [13]. The excitation maximum at E ex = 8.61 eV is attributed to the excitation of its phosphate groups. For comparison, the energy of the excitation maximum for the phosphate group in YPO4 excitation is observed at 8.55 eV [14]. The onset in the excitation spectrum at E fa = 8.3 eV reveals the fundamental absorption edge of LiYP4 O12 ; it defines the optical band gap. To estimate the energy of the mobility edge or bottom of the conduction band we have to add the exciton binding energy to E ex . Based on [15] this energy is about 8% of E ex and then an energy of 9.3 eV is obtained for the bottom of the conduction band. Curves 2 and 4 in figure 2 show clearly a reduced efficiency of Ce3+ excitation in the energy range of exciton creation. This indicates a low probability for energy transfer from excitons, i.e. bound electron and hole pairs, to Ce3+ . The efficiency of Ce3+ excitation increases

luminescence spectrum shows a doublet profile with maxima at 3.97 and 3.72 eV (312 and 333 nm, curve 1 in figure 2). Such a doublet is characteristic for Ce3+ luminescence and corresponds to the transitions from the lowest Ce3+ 5d excited state to the 2 F5/2 and 2 F7/2 spin-orbitally split ground states of the Ce3+ 4f configuration. The energy difference between the doublet maxima is 0.25 eV which is the usual energy of spin– orbit splitting for the Ce3+ 2 F state. Curve 1 in figure 2 shows the luminescence excitation spectrum of the 3.87 eV emission band of LiY0.9 Ce0.1 P4 O12 . It reveals the structure of a typical distribution of Ce3+ 5d energy levels for Ce3+ in a low-symmetry crystal field. In the energy range of 4–7 eV, figure 2, curve 2 shows the maxima of excitation efficiency at 4.20 eV (295 nm), 5.11 eV (242 nm), 5.40 eV (231 nm), 5.65 eV (219 nm), and 6.55 eV (189 nm). The decay kinetics of Ce3+ 5d → 4f luminescence shows an exponential profile with the same decay constant τ of 18.6 ns (figure 3, curve 1) upon the excitation in each of these five bands. Thus, the observed bands are attributed to transition of Ce3+ from the 2 F5/2 ground state to the five crystal field split 5d excited states. The maxima of emission and excitation 3

J. Phys.: Condens. Matter 22 (2010) 485503

T Shalapska et al

Figure 4. Temperature dependence of decay time constant for Ce3+ luminescence from LiY0.9 Ce0.1 P4 O12 . Full circles show the experimental values and the solid line presents the fitting curve.

Figure 5. Normalized x-ray excited emission spectrum (1) and the emission spectrum upon UV excitation of LiY0.9 Ce0.1 P4 O12 (2) with decay kinetics profile for x-ray excited emission (inset) at T = 300 K.

steadily above the mobility edge, shows a plateau in the 17– 20 eV range and increases again above 20 eV (not shown in figure 2). The energy of 20 eV is about 2.5 times higher than the mobility edge and electron multiplication processes start here. Apparently, free electrons and holes are more efficient in exciting Ce3+ luminescence than excitons. Upon the excitation of LiY0.9 Ce0.1 P4 O12 with energy above the bottom of conduction, the decay kinetics of Ce3+ luminescence shows a non-exponential profile (figure 3, curve 2) with a time constant of 15.6 ns for its main decay component. The temperature dependence of the time constant for the decay of Ce3+ 5d–4f luminescence is presented in figure 4. The decay time starts to decrease at around 550 K. A slight increase of the decay time constant is observed in the temperature range from 300 to 500 K. A similar increase in decay time has been reported in [16] where it was attributed to the self-absorption and re-emission of Ce3+ luminescence. The observed temperature dependence of the decay time can be described by a Mott’s formula:

τ=

τ0 , 1 + τ0 0 exp(−E a /kT )

the excitation pulse. The x-ray excited luminescence spectrum does not show any resolved emission band at 3.97 eV like under UV excitation. Only a shoulder band is observed. This is attributed to a larger x-ray penetration depth into the crystal than that for UV light. Also, the emitted photons have to travel a larger distance to escape the sample with a higher probability of reabsorption. The decay kinetic curve of Ce3+ 5d–4f luminescence upon x-ray excitation shown in the inset of figure 5 reveals a dominant fast component with decay time constant of 24.7 ns The increase of the decay time constant upon x-ray excitation (as compared with UV excitation) is attributed to the time required for the free carriers to migrate to Ce3+ centers. Luminescence caused by self-trapped excitons or defect stabilized excitons was not observed from LiY0.9 Ce0.1 P4 O12 upon x-ray excitation. This can testify to the fact that fast recombination of hot electron–hole pairs with cerium ions dominates over exciton formation in LiY0.9 Ce0.1 P4 O12 .

(1)

4. Discussion The experimental data available on all five Ce3+ 5d level energies in LiY0.9 Ce0.1 P4 O12 and the knowledge about the lattice structure allow us to calculate the appropriate crystal field parameters that influence the energy levels of Ce3+ on the yttrium site. Because the same crystal field parameters will apply for all other trivalent lanthanides if at the same yttrium site, one may use the results obtained for the Ce3+ ion to predict the energy levels for other trivalent lanthanides. The relevant parameters are the total crystal field splitting (εcfs ), the centroid shift (εc ), and the red shift (D). The values are obtained from the spectroscopic data in this work and are listed in table 4. The value of εcfs is defined as the energy difference between the lowest and highest 5d levels, the centroid shift is the shift of the average energy of all five 5d states with respect to the value (6.35 eV) for the free Ce3+ ion, and the redshift is the energy difference between the first 5d excitation energy in

where E a is an activation energy for the thermal quenching, τ0 is the radiative decay time, and 0 is the attempt rate. A least squares fit of the experimental data (figure 4) yields E a = 0.75 eV, 0 = 3.1 × 1013 Hz, and τ0 = 18.9 ns. The value for 0 corresponds with 0.128 eV or 1030 cm−1 . This value is equal to the maximum phonon energy of the phosphate group in YPO4 and we expect that the value in LiYP4 O12 is not too much different. 3.3. X-ray luminescence The spectrum of x-ray excited luminescence from LiY0.9 Ce0.1 P4 O12 at T = 300 K is presented as curve 1 in figure 5 together with the spectrum of UV excited emission (curve 2). All decay and afterglow components contribute to the x-ray excited luminescence whereas the UV excited emission was registered within a time window of 0–100 ns after 4

J. Phys.: Condens. Matter 22 (2010) 485503

T Shalapska et al

Table 4. Spectroscopic and crystallographic parameters of LiY0.9 Ce0.1 P4 O12 , LiGd0.9 Ce0.1 P4 O12 [4] and YPO4 [14]. CN and Rav are the anion coordination number and average distance to those anions, respectively. The type of polyhedron (poly) at the Ce3+ site is dodecahedral (ddh) and is given as well. Compound

CN

Rav (pm)

poly

λ5, λ4, λ3, λ2, λ1 (nm)

εcfs (cm−1 )

εc

D (cm−1 )

LiY0.9 Ce0.1 P4 O12 LiGd0.9 Ce0.1 P4 O12 YPO4

8 8 8

238.6 242.7 234.0

(ddh) (ddh) (ddh)

189, 219, 231, 242, 295 187, 219, 229, 241, 294 203, 225, 238, 250, 323

19 000 18 500 18 200

7820 7550 9560

15 500 15 300 18 950

the compound and that (6.12 eV) for the free Ce3+ ion [14, 17]. For comparison, we have also compiled the values for Ce3+ in LiGdP4 O12 and in YPO4 that have similar crystal structure providing very similar 8-fold coordinated dodecahedral (ddh) yttrium sites. The crystal field splitting for Ce3+ 5d in LiYP4 O12 is 19 × 103 cm−1 which is quite similar to that in the other two compounds in table 4. Figure 6 is partly reproduced from [17] and shows the value of the total crystal field splitting in various compounds (see [17]) as a function of the average distance between Ce3+ and the nearest anions. The crystal field splitting can be analyzed in terms of the type and the size of the anion coordination polyhedron around Ce3+ . The size dependence can be described by [17] −2 εcfs = β Rav ,

(2) Figure 6. Spectroscopic polarizability in oxide and fluoride compounds.

where β depends on the shape of the anion polyhedron. Rav is defined as N  Rav = 1/N (Ri − 0.6 R) (3)

known crystal structure with

i=1

αsp =

and presents the average distance to the coordinating ligands in the relaxed lattice structure around Ce3+ . Ri are the bound lengths to the N coordinating anions in the crystal lattice. R is the difference of the ionic radius between Ce3+ and an Y3+ cation that Ce3+ substitutes. The value of β for different types of polyhedra has been obtained empirically from analysis of experimental data on a series of crystals [17]. In the case of 6-fold octahedral coordination (octa) βocta = 1.35 × 109 pm2 cm−1 , for 8-fold cubal and dodecahedral coordination (ddh) and for 9-fold tricapped trigonal prism coordination (3cpt) the value of β is a factor of 0.89, 0.79, and 0.42 times that for octahedral coordination, respectively. For Ce3+ in LiYP4 O12 and LiGdP4 O12 with dodecahedral coordination, we then predict crystal field splittings of 18 700 cm−1 and 18 100 cm−1 , respectively, which are in excellent agreement with the experimental values in table 4. This confirms a correct interpretation of the excitation bands in figure 2 as belonging to the 5d1 · · · 5d5 bands. Whereas the crystal field splitting depends on the shape and size of the anion coordination polyhedron, the centroid shift depends on other parameters like the polarizability of the anion ligands and the related covalency between those anions and Ce3+ . Figure 6 is adapted from [17] and [18], and demonstrates a linear relationship between the so-called spectroscopic polarizability αsp and the inverse square of the average electronegativity χav of the cations in the compound. αsp is calculated from the observed centroid shift and the

6.9 × 10−18 εc , − 0.6 R)−6

N i− 1 (Ri

(4)

where Ri , R , and N have the same meaning as in equation (3). The χav values are 2.018 and 2.013 for LiYP4 O12 and LiGdP4 O12 , respectively. Figure 6 shows that the data for LiYP4 O12 and LiGdP4 O12 agree very nicely with the empirical relationship between αsp and χav . Note also that among the oxide compounds the polyphosphates have a very small value for αsp which reflects that the oxygen ligands are strongly bonded within the polyphosphate framework leading to the small polarizability and the small centroid shift. In the orthophosphate compound YPO4 the oxygen ligands are only bonded to one phosphate atom leading to larger polarizability and large centroid shift. The longer wavelength of emission and larger redshift value for the YPO4 (see table 4) are then entirely caused by a larger centroid shift. The crystal field splitting is almost the same as for the polyphosphates. The spectroscopy of Ce3+ is important to predict the spectroscopy of all the other trivalent lanthanides. What is also important for application and for understanding of luminescence properties is the knowledge of the location of lanthanide energy levels with respect to the conduction band and valence band of the host compound. Once information on Ce is known, the level locations of other lanthanides can be predicted as well. The studies on pure LiYP4 O12 showed that the fundamental excitation of the phosphate group is at 8.61 eV from which the mobility edge was estimated at 9.3 eV. 5

J. Phys.: Condens. Matter 22 (2010) 485503

T Shalapska et al

5. Conclusions The melt solution synthesis technique provides a reproducible production of polycrystalline LiY0.9 Ce0.1 P4 O12 . Luminescence spectroscopy studies allowed us to interpret the five 5d1 · · · 5d5 bands as those corresponding to the components of the excited Ce3+ 5d state split by the crystal field due to the low symmetry at the dodecahedrally coordinated yttrium site in LiY0.9 Ce0.1 P4 O12 . The crystal field splitting and the centroid shift of Ce3+ 5d levels agree well with empirical predictions based on the crystal structure and composition. An energy barrier of 0.75 eV between the relaxed 5d1 state and the bottom of the conduction band was determined from the Arrhenius behavior of decay time shortening as a function of temperature. Under x-ray excitation, luminescence of self-trapped excitons and excitons localized near host defects were not observed, but only the 5d–4f Ce3+ emission was revealed.

Acknowledgments The work is partially supported by the Ukraine Ministry of Science and Education (project No.0109U002075). The authors also appreciate the support of the Superlumi experiment by HASYLAB (DESY, Hamburg).

Figure 7. Energy levels scheme for Ce3+ in LiYP4 O12 .

The energy difference between the lowest 5d state and the mobility edge can be determined from the temperature dependence of the luminescence intensity or decay time. Analysis of the data in figure 4 with equation (1) yields an energy barrier of E a = 0.75 eV for the thermal quenching of Ce3+ 5d–4f luminescence. Studies on the quenching of Ce3+ emission in Lu2 SiO5 and GdAlO3 revealed a clear correlation between thermal quenching and photoconductivity, and it seems to be a general mechanism for Ce3+ 5d–4f emission quenching in compounds [19, 20]. We therefore regard E a as the energy distance between the 5d1 state and the bottom of the conduction band. Consequently the 4f ground state will be at 4.35 eV, and the excited 5di (i —1 · · · 5) states will be at 8.55, 9.46, 9.70, 10.00, 10.90 eV relative to the top of the valence band as shown in figure 7. The 5d2 –5d5 levels are located within the conduction band. On excitation in each of these 5d states we observe decay kinetics curves that are exponential with the same decay time constant as upon excitation in 5d1 . This evidences that the relaxation from the states within the conduction band to the 5d1 state must be extremely rapid, i.e. faster than 1 ns. Only when the auto-ionization speed of the 5d electron to conduction band states is of the same magnitude, may a considerable fraction of those 5d electrons indeed delocalize leaving Ce4+ behind. Delayed re-trapping of the conduction band electron by Ce4+ may then lead to the appearance of slow decay components [21]. However, we did not observe such a slow component. This indicates either that: (1) only a small fraction of 5d electrons delocalize because of too low auto-ionization rate, (2) the recombination with Ce4+ is radiationless, or (3) recombination takes place on a timescale much longer than a microsecond.

References [1] Laroch M, Girard S, Margerie J, Moncorge R, Bettinelli M and Cavani E 2001 J. Phys.: Condens. Matter 13 765 [2] Dorenbos P 2000 J. Lumin. 91 91–106 [3] Zhong J, Liang H, Su Q, Dorenbos P and Birowosuto D M 2007 Chem. Phys. Lett. 445 32 [4] Shalapska T, Stryganyuk G, Demchenko P, Voloshinovskii A and Dorenbos P 2009 J. Phys.: Condens. Matter 21 445901 [5] Shalapska T, Stryganyuk G, Trots D, Demkiv T, Gektin A, Voloshinovskii A and Dorenbos P 2010 J. Lumin. 130 1941 [6] Stoe WinXPOW , version 2.21 Stoe & Cie GmbH, Darmstadt 2007 [7] Rodriguez-Carvajal J 2001 IUCr Newsletter 26 12 [8] Roisnel T and Rodriguez-Carvajal J 2000 EPDIC 7: Mater. Sci. Forum, Proc. Eur. Powder Diffr. Conf. (Barcelona) p PB33 [9] Gelato L M and Parth´e E 1987 J. Appl. Crystallogr. 20 139 [10] Zimmerer G 2007 Radiat. Meas. 42 859 [11] Koizumi H 1976 Acta Crystallogr. B 32 266 [12] Jouini A, Ferid M and Trabelsi-Ayadi M 2003 Mater. Res. Bull. 38 437 [13] Voloshinovskii A, Sol’skii I, Struganyuk G, Romanyshyn Yu, Khapko Z and Antonyak O 2003 HASYLAB Annual Report 237 [14] Dorenbos P 2001 Phys. Rev. B 64 125117 [15] Dorenbos P 2005 J. Lumin. 111 89 [16] Lyu L-J and Hamilton D S 1991 J. Lumin. 48–49 251 [17] Dorenbos P, Pierron L, Dinca L, van Eijk C W E, Kahn-Harari A and Viana B 2003 J. Phys.: Condens. Matter 15 511 [18] Dorenbos P 2002 Phys. Rev. B. 65 235110 [19] van der Kolk E, Basun S A, Imbusch G F and Yen W M 2003 Appl. Phys. Lett. 83 1740 [20] van der Kolk E, Dorenbos P, de Haas J T M and van Eijk C W E 2005 Phys. Rev. B 71 045121 [21] Nikl M, Begnamini A M, Jary V, Niznansky D and Mihokova E 2009 Phys. Status Solidi (RRL) 3 293

6