Charge transfer transitions in the transition metal

ARTICLE IN PRESS Journal of Luminescence 130 (2010) 1357–1365

Contents lists available at ScienceDirect

Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin

Charge transfer transitions in the transition metal oxides ABO4:Ln3 + and APO4:ln3 + (A¼ La, Gd, Y, Lu, Sc; B ¼ V, Nb, Ta; Ln¼ lanthanide) Andreas H. Krumpel a,, Philippe Boutinaud b, Erik van der Kolk a, Pieter Dorenbos a a b

Faculty of Applied Sciences, Delft University of Technology, Mekelweg 15, 2629 JB Delft, The Netherlands Laboratoire des Mate´riaux Inorganiques—UMR 6002, Universite´ Blaise-Pascal et ENSCCF, Aubie re, France

a r t i c l e in fo

abstract

Article history: Received 17 July 2009 Received in revised form 4 February 2010 Accepted 19 February 2010 Available online 24 February 2010

We have compiled and analyzed optical and structural properties of lanthanide doped non-metal oxides of the form APO4:Ln3 + with A a rare earth and of transition metal oxides with formula ABO4:Ln3 + with B a transition metal. The main objective is to understand better the interrelationships between the band gap energy, the O2 -Ln3 + charge transfer energy, and the Ln3 + -B5 + inter-valence charge transfer energy. Various models exist for each of these three types of electron transitions in inorganic compounds that appear highly related to each other. When properly interpreted, these optically excited transitions provide the locations of the lanthanide electron donating and electron accepting states relative to the conduction band and the valence band of the hosting compound. These locations in turn determine the luminescent properties and charge carrier trapping properties of that host. Hence, understanding the relationship between the different types of charge transfer processes and its implication for lanthanide level location in the band gap is of technological interest. & 2010 Published by Elsevier B.V.

Keywords: Charge transfer Inter-valence charge transfer Lanthanides Transition metal oxides

1. Introduction Many lanthanide (Ln) doped transition metal (d-block) oxides of the form ABO4:Ln3 + (A¼La, Gd, Y, Lu; B ¼V, Nb, Ta; Ln¼La, Ce, y, Lu) are well known for many different applications as laser host crystals [1,2], solar cells [3] or phosphor materials [4]. Their luminescence quantum efficiency depends on the location of the 4f and 5d energy levels of the Ln dopants relative to the valence band (VB) and the conduction band (CB) of the host. The energies of charge transfer (CT) that can be identified in photoluminescence (PL) excitation spectra can be used in order to establish those locations. In the ABO4 transition metal oxides, a CT can be observed from the Ln dopant to a host ion or between the host ions themselves. Fig. 1 exemplifies the three different types of CT that are of main interest throughout this paper: (i) fundamental host transition due to electron transfer from the oxygen VB to the CB which defines the band gap; (ii) inter-valence charge (electron) transfer (IVCT) from a Ln3 + ion to the CB leaving Ln4 + ; and (iii) electron transfer from an O2 ion to the incompletely filled 4f shell of a Ln3 + dopant creating Ln2 + . At the beginning of this article, in Section 2, an overview of the different methods of locating Ln 4f energy levels relative to the host bands by means of observed CT energies will be given. In particular, the method of Boutinaud based on the IVCT energies and the method of Dorenbos based on CT energies will be

Corresponding author. Tel.: + 31 15 27 81954; fax: + 31 15 27 89011.

E-mail address: [email protected] (A.H. Krumpel). 0022-2313/$ - see front matter & 2010 Published by Elsevier B.V. doi:10.1016/j.jlumin.2010.02.035

discussed. One main objective is comparing the different approaches with each other. Another objective is finding hostrelated parameters which help to predict the absolute location of the 4f energy levels in any compound. Section 3 tabulates first the crystallographic and electronic properties of the group 5 transition metal oxides ABO4:Ln3 + (B¼ V, Nb, Ta) and the non-metal oxides APO4:Ln3 + . Next, the observed Eu3 + CT and the Pr3 + IVCT energies in these compounds are presented and discussed. It will be shown that both the O2 -Eu3 + CT and the Pr3 + -B5 + IVCT energies for one specific type or compound like the orthovanadates, AVO4:Ln3 + depend on the electronegativities as well as on the inter-atomic distances of the ions that are involved in the CT process. An expression for the dependence of the Pr3 + -B5 + IVCT energy on these two parameters for different transition metal oxides has been found earlier by Boutinaud et al. [8]. His model and our attempt in this work to find a related expression for the O2 -Eu3 + CT energies are an extension of the optical electronegativity model introduced by Jørgensen [9] four decades ago. The photoluminescence (PL) data that are used in this work were obtained by own measurements or from the literature. The oxygen 2p states form mainly the top of the VB and the d states of the B5 + ions form the bottom of the CB [5–7].

2. Knowledge on CT energies A number of attempts were made in the last few decades to predict the energy of CT transitions. In this work we will only consider electron transfer transitions. The energies of those

ARTICLE IN PRESS 1358

A.H. Krumpel et al. / Journal of Luminescence 130 (2010) 1357–1365

filled, and it is smallest for La and Gd. Expression (1) reflects the trend of CT energies for one specific Ln ion (M¼lanthanide) reasonably well going from one compound family to another (fluorides, chlorides, bromides, etc.). But it does not account for the difference in CT energies between members of, e.g., the oxide family or the fluoride family of compounds [13]. Within one particular group of compounds as for instance the vanadates, AVO4, other parameters as the anion coordination number or the size of the rare earth ion do also have an effect on the CT energies [13]. Eq. (1) can be applied also to express the band gap energy of binary compounds with formula MmXn or compounds containing complexes of the from MXq n [44].

B

B5+5+

(ii)

Ln2

(i)

Ln3 (iii)

OO2 2

Fig. 1. Three different types of CT transitions: (i) fundamental host transition, O2 -B5 + ; (ii) IVCT transition, Ln3 + -B5 + ; (iii) electron transfer from ligand to dopant, O2 -Ln3 + .

transitions involving Ln dopant ions are intimately related to the position of the dopants acceptor or donor state level with respect to the VB and the CB of the host crystal. Two different approaches to predict CT energies that gave the most promising results so far will be reviewed. (1) One of these approaches concentrates on correlations of the CT energy with numerical parameters as, e.g., the optical electronegativity of one of the ions involved in the CT process or inter-atomic distances between the electron donator and the electron acceptor. Its strength is that one formula roughly expresses the CT energies of a large group of different compounds all doped with the same Ln ion. The original approach by Jørgensen [9] has been employed and refined by Boutinaud et al. [8] in order to describe IVCT transitions. (2) The other approach in contrast concentrates on the energetic location of the lowest 4f and the lowest 5d states of the entire series of 14 Ln ions all singly doped in the same compound. It makes use of a characteristic zig-zag curve of the 4f electron binding energies that can be seen when going through the Ln series. The models of Dorenbos [10] and Nakazawa and Shiga [11] are examples of this approach. In the following two sections these two approaches will be presented in more detail.

2.1.2. The model of Boutinaud In Ln doped transition metal oxides a special type of metal to metal CT has been observed where the electron is transferred from the Ln dopant to the transition metal ion of the host. This CT is called inter-valence charge transfer (IVCT). Boutinaud et al. [8,14,15] has developed an empirical model to describe IVCT transitions involving Pr3 + (and Tb3 + ). His model considers not only the optical electronegativity as the model of Jørgensen but also the inter-atomic distances between Pr3 + (Tb3 + ) and the transition metal ion Bq + . Using the position of the IVCT band of Pr3 + (or Tb3 + ) doped closed-shell transition metal oxides, Boutinaud et al. [8] proposed the following linear relationship: EIVCT ðPr3 þ Þ ¼ ð1:961:66wopt ðBq þ Þ=dðPr3 þ Bq þ ÞÞ3:72 eV

ð2Þ

˚ between d(Pr3 + Bq + ) is the shortest inter-atomic distance (in A) Pr3 + and Bq + . Fig. 2 shows the Pr3 + IVCT energies in different compounds together with the straight line given by Eq. (2). The values for both the IVCT energies and the wopt ðBq þ Þ=dðPr3 þ Bq þ Þ ratios inside Fig. 2 are taken from Refs. [8,14,16]. The model of Boutinaud reproduces well the IVCT energies of Pr3 + in different oxides, and shows therefore that the inter-atomic distance has an influence on IVCT transition energies. It will be shown in Section 3.3 in more detail how Eq. (2) is related to Jørgensens formula (1), although here we deal with a metal to metal CT instead of an anion to metal CT. The IVCT bands were identified in the excitation spectra of Pr3 + :1D2 and Pr3 + :3P0 emission in several transition metal oxides [14,17,18]. It was found that the Pr3 + IVCT has about the same energy as the Tb3 + IVCT energy. Fig. 1 then suggests that the location of the Pr3 + ground state (or electron donor state) must be

2.1. CT energy and related numerical parameters 2.1.1. The model of Jørgensen A simple expression for the first Laporte-allowed CT energy, ECT, in an inorganic ionic compound was published by Jørgensen and is based on his definition of the optical electronegativity, wopt [9,12]: ECT ¼ ðwopt ðXÞwopt ðMÞÞ3:72 eV

ð1Þ

X represents here the electron donor (ligand) and M the electron acceptor (metal ion) in the CT process. With a reversed sign, Eq. (1) describes the CT energy for an electron transfer from M to X. The optical electronegativities of the Ln ions are defined such that their values reflect the strength of the binding energy of the electron that is transferred to the Ln 4f shell. The binding energy is largest when the 4f shell becomes half (Eu) or completely (Yb)

Fig. 2. EIVCT(Pr3 + ) plotted against the ratio between the optical electronegativity of the transition metal cation Bq + and the shortest Pr3 + Bn + distance; the numbers for each compound shown inside the figure were taken from Refs. [8,14,16].

ARTICLE IN PRESS A.H. Krumpel et al. / Journal of Luminescence 130 (2010) 1357–1365

1359

at about the same energy as the Tb3 + ground state with respect to the conduction band.

from the 4f ground state of Eu3 + by about 5.7 eV to about 6.7 eV [24].

2.2. The zig-zag curve of binding energies

2.2.2. The model of Nakazawa On the basis of Jørgensen’s [9] theoretical work on the variation of the lowest 4f–5d and CT transition energies of Ln ions as a function of the number of electrons in their 4f shell, Nakazawa published a theoretical approach that concentrates on the 4f–5d and CT transition energies using the examples of YPO4:Ln3 + [25] or LaPO4:Ln3 + [11]. Nakazawa started with a representation of the transition energies by means of Slater, Condon and spin–orbit interaction parameters and he derived a simple expression of the form

2.2.1. The model of Dorenbos Analyzing the lowest 4f–5d transition energies of Ln ions as dopants in different compounds, Dorenbos found that for all Ln ions when on the same site in the same compound this energy decreases as compared to its value for the free gaseous ion by the same host-specific amount. This energy decrease was named redshift or depression D (q+ ,A); q+ stands for the Ln ionic charge and A for the compound name. From experimental techniques like ultraviolet photoelectron spectroscopy (UPS), X-ray photoelectron spectroscopy (XPS), photoconductivity measurements, excited state absorption (ESA), or luminescence quenching studies the energy difference between the delocalized VB or CB levels and a localized Ln 4f GS or 5d level can be determined [19,20,21,22]. Dorenbos combined information from these techniques with information on the energy for charge transfer from an anion ligand to trivalent Ln ions in compounds to locate the 4f ground state of lanthanides relative to the top of the VB. The results have been depicted by him in energy level diagrams [10,23]. When going through the lanthanide series from La to Lu, the 4f GS locations relative to the VB and the CB of the host compound form a characteristic zig-zag or double-seated curve similar to the one shown in Fig. 3. The curve has minima when the 4f shell is half and fully filled with electrons and maxima when it is filled with one or eight electrons. We point out that the diagrams proposed by Dorenbos are not based on the calculated energy levels of a quantum mechanical system but show energy differences that were measured or expected on grounds of simple, empirical relations. The shape of the zig-zag curve appears to be almost the same in all types of compounds, even though its location relative to VB and CB changes. This equality of shape means that when the energy of CT from the VB to only one Ln ion, for instance of Eu3 + , is known, the CT energies to all other lanthanides when doped on the same lattice site in the same compound are known as well: ECT ðnÞ ffiECT ð6Þ þ DECT ðnÞ

ð3Þ

n is the number of electrons in the 4f shell of the trivalent lanthanide ion where n ¼6 for Eu3 + , and the host independent constant DECT(n) is compiled in Ref. [10]. Depending on the band gap of the host material, the 4f ground state of Eu2 + is separated

ECT ðnÞ ¼ VCT0 þvCT nDUðn þ 1Þ

where VCT0 and vCT are empirical parameters describing the single-electron-scheme binding energies in (4) and DU(n + 1) is the many-electron correction term given for each n in Ref. [11]. ECT can then be obtained for all lanthanide ions by fitting few experimental values with formula (4). In Fig. 3, the zig-zag curve for CT energies in YPO4:Ln3 + as proposed by Dorenbos is compared to the one predicted by the model of Nakazawa [25]. Apart from few variations (at most 0.56 eV for Gd) the two curves are similar. As mentioned in Section 2.1.1, the values of the optical electronegativities of the Ln ions are related to the binding energy of the electron added to the Ln 4f shell. When going trough the Ln series from La, Ce to Lu, the binding energies form the same zigzag curve as shown in Fig. 3. It is noted that the DECT(n) values of Eq. (3), the DU(n + 1) values of Eq. (4), and the wopt(Ln) of Eq. (1) are closely related to each other, and they are all three related to the redox potentials of the lanthanides. 2.3. Relationship between IVCT energy and 4f electron binding energy Dorenbos stated that the Ln3 + CT energies shown in Fig. 3 for YPO4:Ln3 + give approximately the 4f ground state locations of the divalent Ln dopant ions (Ln2 + ) relative to the VB of the host [10]. The zig-zag curve shown in Fig. 3, however, was not based exclusively on measured CT energies. It is also based on an assumed or alleged location of the lowest 5d states relative to the bottom of the CB. Then, by subtracting the predicted 4f–5d energy differences using the appropriate value for the redshift, one obtains also the Ln2 + :4f GS energy relative to the top of the VB. It was proposed in Ref. [10] that the lowest 5d level locations of the trivalent Ln dopants behave in a similar fashion as those for the divalent ones. One then obtains a similar curve for the 4f ground state energies of the Ln3 + ions relative to the VB as the ones shown in Fig. 3. It appeared that the zig-zag curve for the trivalent lanthanides as proposed by Dorenbos in Ref. [10] contradicted for some lanthanides with the results from the IVCT model of Boutinaud. The IVCT data of Boutinaud (B) (see Section 2.1.2) revealed that the energy difference between the 4f GS energies of Pr3 + and Tb3 + relative to the VB of the host, DE (Pr3 + , Tb3 + ), is negligible, whereas the model of Dorenbos (D) assumed the 4f GS energy of Tb3 + to be about 0.73 eV higher than the one of Pr3 + :

DEB ðPr3 þ ,Tb3 þ Þ ffi0 eV o DED ðPr3 þ ,Tb3 þ Þ ¼ 0:73 eV

Fig. 3. ECT(n) for YPO4:Ln3 + determined from the models of Dorenbos (K) and Nakazawa (J) [25]; for both models ECT (6) 5.56 eV (223 nm) was chosen as reference; n¼number of electrons in the Ln3 + 4f shell.

ð4Þ

ð5Þ

To repair this shortcoming, a new set of DECT(n) values (see Eq. (3)) for the divalent and trivalent lanthanides was proposed in Ref. [26]. We restrict the following discussion to the transition metal compounds with formula ABO4 (B¼ V, Nb, Ta). Furthermore, we confine ourselves to a simple ionic picture in which we single out group with an adjacent Ln3 + ion that we refer as the a BO3 4



3 system fLn3 þ ,BO3 4 g with initial state jiS. The lowest BO4 group transition (LGT) is treated to be a CT from O2 to B5 + , which changes the system as follows (Fig. 4a): 3þ jiS-jf S1 ¼ ^ fLn3 þ ,B5 þ O8 ,B4 þ O7 4 g-fLn 4 g

ðLGTÞ

ð6Þ

Eq. (6) describes by definition an electron transfer from the top of the VB to B5 + . The IVCT will be treated as a CT from Ln3 + to B5 + , which changes the system as follows (Fig. 4b): 4þ ^ fLn3 þ ,B5 þ O8 ,B4 þ O8 jiS-jf S2 ¼ 4 g-fLn 4 g

ðIVCTÞ

ð7Þ

In principle, the transferred electron is still bound by the Coulomb potential of Ln4 + or of the hole in the VB. One still needs to add the electron hole binding energy to obtain a free electron in the CB [27]. Let us assume that in both cases, Eqs. (6) and (7), the electron is transferred into the same electronic state that involves somehow the B4 + cation. In order to relate energetically the LGT to the IVCT within one energy level diagram, we have adopted the following method [26]. As a rule of thumb the bottom of the conduction band is assumed at 8% higher energy than the energy of the maximum of the first host excitation band, i.e., this is the energy needed to remove the transferred electron from the B4 + cation in the fLn3 þ ,B4 þ O7 4 g state. This rule of thumb has been motivated in Ref. [13]. In the fLn4 þ ,B4 þ O8 4 g state, the hole is more distant from the transferred electron than in the fLn3 þ ,B4 þ O7 4 gstate, and consequently less energy is required to excite the electron to the conduction band. We will assume that this energy is 4% of the energy at the maximum of the first host excitation band. The IVCT data and the model of Boutinaud provides us then with the energy difference between the 4f ground state of Ln3 + (for e.g., Pr3 + and Tb3 + ) and the CB (Fig. 4c). The CT data and the Dorenbos model on the other hand provide us the energy difference between the lowest 4f state of Ln2 + and the top of the VB. The Boutinaud model and the Dorenbos model can now be united in order to provide the locations of the lowest 4f energy levels for all tri- and divalent lanthanides in the energy difference scheme of one compound. As mentioned in Section 2.1.2, the IVCT energies for Pr3 + and Tb3 + in the same compound appear almost similar. This means that their respective ground state energies are also at the same location with respect to the CB. It happens that then also the emitting 3P0 and 5D4 states of Pr3 + and Tb3 + are at the same absolute location, as illustrated in Fig. 5. The thermal quenching of the Pr3 + :3P0 and the Tb3 + :5D4 emission is attributed to intersystem crossing via the IVCT state or the CB as illustrated in the configurational coordinate diagram on the right hand side

a f

b

c cb ex

i

i

ELGT

2

EIVCT

f

IVCT

LGT

1

Ln3+ GS

vb

3þ Fig. 4. Energy level schemes depicting the fLn3 þ ,B5 þ O8 ,B4 þ O7 4 g-fLn 4 g LGT 4þ ,B4 þ O7 (a), and the fLn3 þ ,B5 þ O8 4 g-fLn 4 g IVCT (b). In the scheme that combines the two CT transitions (c) the energy differences are depicted relative to a common ionization level (dotted horizontal line); ELGT ¼LGT energy; EIVCT ¼ IVCT energy; vb¼ valence band, cb¼ conduction band, ex ¼ exciton state.

Fig. 5. Energy level diagram (left side) and configurational coordinate diagram (right side) of GdVO4:Ln3 + (Ln¼Pr, Tb); arrows indicate the IVCT transition as used for energy level location.

in Fig. 5. It was found for AVO4 (A ¼La, Gd, Y) [17,18] and for CaNb2O6 and YNbO4 [8] (see also Section 3.4) that the activation energies for thermal quenching of the Pr3 + :3P0 and Tb3 + :5D4 emission are about the same which is in line with the level scheme.

3. Characterization and analysis of the ABO4:Ln3 + system This chapter compiles relevant crystallographic and electronic properties of the ABO4:Ln3 + system. The O2 -Eu3 + CT and the Pr3 + -B5 + IVCT will be discussed and a formula will be presented that relates the CT energy to the distance between Eu3 + and O2 . It will be shown how the band gap energy can be expressed by means of the O2 -Ln3 + CT and the Ln3 + -B5 + IVCT energies. 3.1. The crystal structures of the ABO4 compounds Table 1 compiles structural information on the ABO4 compounds studied in this work. The orthovanadates AVO4 (A ¼Ce, y, Lu) have a tetragonal zircon-type structure with space group I41/amd (Nr. 141), whereas LaVO4 has a monoclinic monazite-type structure with space group P21/n (Nr. 14) (see Table 1). Despite the difference in structure, in all orthovanadates of rare earths the vanadium ions are coordinated by 4 oxygen atoms. Nevertheless, the band gap of group transition energy, EðVO3LaVO4, i.e., the lowest VO3 4 4 Þ, is about 0.46 eV larger than that of all the other orthovanadates (see Table 2). The structure of the rare earth orthoniobates ANbO4 is temperature dependent. At low temperatures they have monoclinic scheelite (fergusonite)-type structure with space group I2/a (Nr. 15), whereas at higher temperatures (500–850 1C [53]; for instance: T4495 1C for LaNbO4 and T4725 1C for NdNbO4 [54]) it transforms into the tetragonal scheelite-type structure with space group I41/a (Nr. 88). The orthotantalates ATaO4 crystallize in three different structures: (1) the monoclinic fergusonite- or M-type with space group I2/a (Nr. 15, for A¼Nd, y, Er), (2) the structurally related M0 -type with space group P2/a (Nr. 13, for A ¼Nd, y, Er) and (3) the P21/c type (Nr. 14, for A¼ La, Ce, Pr). Table 1 compiles the average inter-atomic distances, dðTa5 þ O2 Þ, in M-ATaO4 (A¼Gd, Y, Lu) which were estimated from the two shortest Ta5 + –O2 distances given in the cited references for the M0 -type of these


1361

Table 1 A¼ f-block elements or P; B ¼d-block elements; r ¼ effective ionic radius [28]; the space group is represented by its Hermann Mauguin symbol and number, as recorded in the International Tables for Crystallography [29]; I5 ¼ 5th ionization potential [30]; wP ¼ electronegativity (Pauling scale) [31,32]; d(R O) (R ¼ A, B) ¼average inter-atomic distance in polyhedra of the ABO4 structures (values in parenthesis are estimated); CN¼ coordination number (same reference as space group or inter-atomic distance). ABO4

r(A3 + ) (pm)

r(B5 + ) (pm)

Space group (no.)

d(O A) (pm)

d(O B) (pm)

I5(B) (eV)

wP(B)

CN(A3 + )

CN(B5 + )

LaVO4 GdVO4 YVO4 LuVO4 LaNbO4 GdNbO4 YNbO4 LuNbO4 LaTaO4 M-GdTaO4 M0 -GdTaO4 M-YTaO4 M0 -YTaO4 M-LuTaO4 M0 -LuTaO4 LaPO4 GdPO4 YPO4 LuPO4 ScPO4

121.6 105.3 101.9 97.7 116 105.3 101.9 97.7 116 105.3 105.3 101.9 101.9 97.7 97.7 121.6 110.7 101.9 97.7 87

35.5 35.5 35.5 35.5 48 48 48 48 64 64 64 64 64 64 64 17 17 17 17 17

P21/n (14) [33] I41/amd (141) [34] I41/amd (141) [35] I41/amd (141) [36] I2/a (15) [37] I2/a (15) [40] I2/a (15) [40,41] I2/a (15) [40] P21/c (14) [43,44] I2/a (15) [45] P2/a (13) [45] I2/a (15) [46,47] P2/a (13) [45] I2/a (15) [48] P2/a (13) [48] P21/n (14) [49], 50 P21/n (14) [50,51] I41/amd (141) [50,52] I41/amd (141) [50,52] I41/amd (141) [50,52]

259.7 248.3 236.5 233.1 250.5 239.4 236.3 232.3 254.4 241.1 241.1 235.5 235.5 235.3 235.3 257.3 246.9 233.7 230.3 220.6

170.9 [33] 160.4 [34] 170.88 [36] 170.67 [36] 209.8 [39] 207.3 [39] 207.5 [41] 205.7 [42] 199 [44] (193.25) [44] 202.3 [44] (190.4) [46] 201.1 [46] (191.2) [44] 199.7 [44] 153.4 [49] 153 [51] 154.3 [52] 153.4 [52] 153.4 [52]

65 65 65 65 51 51 51 51 45 45 45 45 45 45 45 65 65 65 65 65

1.63 1.63 1.63 1.63 1.6 1.6 1.6 1.6 1.5 1.5 1.5 1.5 1.5 1.5 1.5 2.19 2.19 2.19 2.19 2.19

9 8 8 8 8 8 8 8 8 8 8 8 8 8 8 9 9 8 8 8

4 4 4 4 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 +2 4 4 4 4 4

Table 2. 3+ EðBO3 ) ¼ CT energy of Eu3 + , 4 Þ¼lowest transition energy in BO4 group; ECT(Eu where no reference is given, the values are taken from own measurements. ABO4

EðBO3 4 Þ (eV)

ECT(Eu3 + ) (eV)

LaVO4 GdVO4 YVO4 LuVO4 ScVO4 LaNbO4 GdNbO4 YNbO4 LuNbO4 ScNbO4 LaTaO4 M-GdTaO4 M0 -GdTaO4 M-YTaO4 M0 -YTaO4 M-LuTaO4 M0 -LuTaO4 ScTaO4 LaPO4 GdPO4 YPO4 LuPO4 ScPO4

4.26 [17] 3.9 [18] 3.87 [55,84], 3.76 [57] 3.83 [18] 3.72 [59], 3.69 [55] 4.84 [60] (NbO4), 4.13 [60] (NbO6) 4.58 [61] 4.68 [62], 4.4 [63] – 4.77 [59,65] 4.77 [66] 5.32 [67], 5.39 [68] 5.41 [67] 5.51 [66] – – – 4.77 [70] 8 [71], 7.8 [72] 7.75 [72] 8.16 [72] 8.55 [72] 7.47 [58], 7.21 [74]

3.41 [17], 3.94 – 4.77 [58] – – 4.61 – 5.1 [64], 4.84 [64] 5.17 [48] – 4.43 [66] 5 [67] 5 [67], 4.96 [45] 5.06 [66] 4.88 [69] – 5.17 [48] – 4.84 [73], 11 5 [11] 5.66 [11], 5.56 [11,25,55] 5.74 [11] 6.05 [58], 5.96 [55]

compounds. The average distances between A3 + and O2 , dðA3 þ O2 Þ are assumed to be approximately the same for the M- and M0 -type of the ATaO4 compounds [56]. The crystal structures of the orthophosphates APO4 are similar to that of the orthovanadates, AVO4:LaPO4 and CePO4 are isostructural to LaVO4, i.e., they have the monoclinic monazitetype structure with space group P21/n (Nr. 14), whereas APO4 (A ¼Sm, Eu, Gd, Y, Sc, Lu) have a tetragonal zircon-type structure with space group I41/amd (Nr. 141). 3.2. Energy of charge transfer from the valence band to Eu3 Table 2 compiles the Eu3 + and band gap CT energies for the ABO4 compounds. For LaVO4 we compiled another value for the

[33] [34] [36] [36] [38] [39] [41] [42] [44] [44] [44] [46] [46] [44] [44] [49] [51] [52] [52] [52]

O2 Eu3 + CT energy, ECT(Eu3 + ), than in our earlier work [17]. After a new analysis, a shoulder in the excitation spectrum of LaVO4:Eu3 + at about 315 nm has been assigned to the Eu3 + CT. The O2 -Eu3 + CT is an anion ligand to metal CT and can, in principle, be expressed by Eq. (1). Eq. (1), however, does not account for variations in the value for the CT energy within, for example, the oxides family of compounds or within any other family of compounds. In this section, we will propose a modification to Eq. (1) in order to reproduce the A-dependent (A ¼La, Gd, Y, Lu) Eu3 + CT energies for the transition metal oxides with formula ABO4:Eu3 + (B ¼V, Nb, Ta) and for the nonmetal oxides APO4:Eu3 + . In Eu3 + CT, oxygen is the electron donor and europium, the lanthanide, the electron acceptor. Thus, from lanthanide’s standpoint, the Eu3 + CT is the opposite of the IVCT, meaning the lanthanide accepts an electron instead of donating one. This suggests that the IVCT model of Boutinaud (see Section 2.1.2) should also be applicable to the Eu3 + CT energy. In fact, a correlation has been observed between the Eu3 + CT energy and the ionic radius of the rare earth, which is being replaced by the Eu3 + dopant in a rare earth oxide [13]. The Eu3 + CT increases non-linearly with decrease in size of the host cation. Translating the ideas of Boutinaud about IVCT, we may express the Eu3 + CT energy, ECT (Eu3 + ), given in Table 2, as a function of the average inter-atomic distance, dðO2 Eu3 þ Þ, between the Eu3 + dopant ion and its oxygen ligands in a compound with formula ABO:Eu3 + . In order to account for anion relaxation that occurs when Eu3 + substitutes for a host rare earth, A3 + , with a different ionic radius we assume that the average O2 Eu3 + distance is expressed as dðO2 Eu3 þ Þ ffi dðO2 A3 þ Þ þ 0:5ðrðEu3 þ ÞrðA3 þ ÞÞ:

ð8Þ

Fig. 6 shows ECT (Eu3 + ) plotted against d1 ðO2 Eu3 þ Þ for ABO4:Eu3 + (A ¼La, Gd, Y, Lu; B¼V, Nb, Ta) and APO4:Eu3 + (A¼ La, Gd, Y, Lu, Sc). We observe for each type of compound that the O2 -Eu3 + CT energy increases with decrease in size of the substituted lattice rare earth cation. Such a relationship between the Eu3 + CT energy and the decreasing lattice site has been observed earlier [55]. In order to get a better understanding of this relationship we consider Eq. (1) that expresses the difference between two energy values: one related to the electron donor and the other to the electron acceptor. Eq. (1) can principally be



O

2−

B 5+

B5+

O2−

O 2−

B 5+

e−

O2−

A3+

Eu 3+

O2−

2−

O

A3+

Fig. 6. ECT(Eu3 + ) plotted against the reciprocal of the average interatomic distance dðO2 Eu3 þ Þ for the vanadates (curve 1, m), the niobates (curve 2, ]), the tantalates (curve 3, E) and the phosphates (curve 4, K); host cation A is specified inside the figure.

Fig. 7. Simplified Eu3 + CT transition scheme including the nearest neighbors for a compound with formula ABO4:Eu3 + ; the arrow pointing from the O2 ion to the Eu3 + ion indicates the direction of electron transfer.

modified in the following way so as to reproduce the trend shown in Fig. 6. ! ( ) b 3þ w ðEu Þ 3:72 eV ECT ðEu3 þ Þ ¼ awopt ðO2 Þ 1 opt dðO2 Eu3 þ Þ

Table 3 Fitting parameters of the Eu3 + CT energy in ABO4:Eu3 + expressed by Eq. (9);

( ¼

awopt ðO2 Þwopt ðEu3 þ Þ þ

bwopt ðEu3 þ Þ 2

dðO

3þ

Eu

) Þ

3:72 eV

ð9Þ

To motivate this equation, let us consider two compounds, viz. LaBO4:Eu3 + and LuBO4:Eu3 + (B¼V, Nb, Ta, P). The ionic radius of La3 + varies, depending on the coordination number, from 18.3 to 23.9 pm. The radius is larger than that of Lu3 + , and consequently the O2 to Eu3 + distances are larger in LaBO4:Eu3 + than in LuBO4:Eu3 + , see Eq. (8). The binding energy of the electron transferred to Eu3 + that is involved in the CT process will be affected by the Coulomb interaction with the neighboring negative anions. A closer distance decreases the electron binding at Eu and consequently the energy of CT for Eu3 + on a Lu3 + site tends to be higher than on a La3 + site. In Eq. (9) this is accounted for by the correction term 1b=dðO2 Eu3 þ Þ. Also the binding of the electron on the donating oxygen ligand may depend on the anion to rare earth distance. A smaller size rare earth cation can have two effects: (i) a small ionic radius is accompanied with a higher value for the electronegativity. It leads to slightly stronger binding of the oxygen ligand and (ii) a smaller distance to the positive rare earth leads to stronger Coulomb binding of the electron. Fig. 7 displays schematically the situation. The acceptor ion (Eu3 + ) is coordinated by at least six oxygen whereas the donating ion (O2 ) is mostly coordinated by both Ln and B cations We then expect that the binding energy of O2 ligands that are involved in the CT process will not be affected too much when going from LaVO4:Eu3 + to LuVO4:Eu3 + ; the strongest binding is provided anyway by the B5 + ions. Table 3 compiles the fitting parameters that were used in Eq. (9) in order to get the straight lines shown in Fig. 6. Eq. (9) expresses an inverse proportional relationship between the CT energy and the inter-atomic distances in the lattice. A related inverse proportionality exists between the direct band gap energy and the inter-atomic distance in the simple binary compounds with formula A1O1, such as the alkaline earth metal oxides (A ¼Ba, Sr, Ca, Mg) [13,75], the group 10 element oxides (A ¼Ni, Pd) [76], or the group 12 element oxides (A ¼Zn, Cd) [77].

C1 bwopt ðEu3 þ Þ; the second parameter C2 awopt ðO2 Þwopt ðEu3 þ Þ is specified at d 1(O2 Eu3 + ) ¼0.4 A˚ 1. B

C1

C2

V Nb Ta P

8.22 8.22 8.22 8.22

1.13 1.20 1.22 1.34

In the case of the alkaline earth oxides the valence bands are mainly formed by oxygen p states, whereas the bottom of the conduction band consists mainly of metal s and d states [78,79]. Although the particular electronic configuration of each metal influences the width of the band gap, we might assume that the physical process that causes the inverse proportional relationship between the CT energy and the inter-atomic distance is similar both in the alkaline earth metal oxides and in the Eu3 + doped transition metal oxides. Indeed, it is known for many compounds that the band gap is inversely proportional to the size of the host cation, the states of which form the CB [13]. It has also been observed that the band gap is directly proportional to the Eu3 + CT energy, the constant of proportionality being approximately 3 [13]. 3.3. The energy of the Pr3 + B5 + inter-valence charge transfer In the ABO4:Ln3 + (A ¼La, Gd, Y, Lu; B¼V, Nb, Ta, P) compounds the B cation always has oxidation state 5+ with a noble gas electron configuration: V5 + [Ar], Nb5 + [Kr], Ta5 + [Xe], P5 + [Ne]. Column 3 in Table 1 shows that the P5 + ion has a much smaller effective ionic radius and a larger electronegativity than the other cations. This is related to the stronger binding between the oxygen ligands and phosphor, and consequently the band gap in the orthophosphates is larger than the one of the transition metal oxides as can be seen in Table 2. In the small band gap transition metal oxides no 5d–4f emission can be observed. This can be explained by assuming that all Ln 5d states are located above the (B¼V, Nb, Ta) state. In the phosphates the lowest excited BO3 4 lowest 5d state is energetically located inside the band gap and 5d–4f emission is observed [80]. If both the Ln3 + 4f ground state


and the lowest 5d energy level are energetically located inside the band gap, the first intense Ln3 + 4f–5d excitation band is at a lower energy than the weak Ln3 + -B5 + IVCT band. Consequently, the IVCT band is completely overwhelmed and not observable in excitation spectra. If on the other hand the Ln3 + 4f ground state is located inside the band gap but the lowest 5d state is above it then an Ln3 + B5 + IVCT can be observed. This explains why IVCT transitions have only been observed in Ln doped transition metal oxides and not in APO4 phosphates. In YTaO4:Ln3 + the lowest energy 4f–5d excitation band of Tb3 + emission is expected at the same wavelength of about 275 nm as the Pr3 + and Tb3 + IVCT band [15,62]. Lammers and Blasse [68] assigned the weak excitation band at 295 nm of Tb3 + emission in GdTaO4:Tb3 + to the spin-forbidden transition from the Tb3 + :4f8 GS to the 4f75d1:9D level. The model of Boutinaud et al. [8] predicts the Tb3 + Ta5 + IVCT band at about 298 713 nm. In those situations it is difficult to assign the observed band either to 4f–5d or to IVCT transitions. Another interesting feature was observed for the vanadates, AVO4:Pr3 + , and the titanates, ATiO3:Pr3 + [57]: the IVCT energy increases linearly with increasing Pr3 + B5 + (B¼V, Ti) distance itself, where the latter one is affected by the size of the Aq + cation. This is implied in Eq. (2) although it does not seem to be the case for the niobates and the tantalates [57]. In order to account for the observed linear dependence of the Pr3 + IVCT energy on the Pr3 + B5 + distance, we can again modify Eq. (1) in order to approximate the Pr3 + IVCT energy within the orthovanadates or within the tantalates by ! ( ) a 3þ 5þ 1 w ðPr Þb w ðB Þ 3:72 eV EIVCT ðPr3 þ Þ ¼ opt opt dðPr3 þ B5 þ Þ ( ¼

wopt ðPr3 þ Þbwopt ðB5 þ Þ

3þ

awopt ðPr

Þ

dðPr3 þ B5 þ Þ

)

1363

EIVCT La

A = Lu

d1 + d2 EIVCT ∝ (a − b/(d1 + d2))⋅ ))⋅k ⋅

e−

V 5+ V 5+

d1

O 2− O 2−

d2

Ln 3+ Ln 3+

e−

ECT

3:72 eV ð10Þ

As can be seen from column 6 in Table 1, the O2 to B5 + distance is not affected very much by the rare earth cation of the host. That means that the Pr3 + to B5 + distance is determined mainly by the O2 to Pr3 + distance. Therefore, the relationship between the binding energy of the B5 + cation and the Pr3 + to B5 + distance has been neglected in Eq. (10). Eq. (10) explains the change in Pr3 + IVCT energy when going from LaVO4:Pr3 + to LuVO4:Pr3 + . It does not explain the difference in IVCT energies when going through the transition metal oxides as done by Boutinaud. In that latter case Eq. (2) holds. The difference between Eq. (9) and Eq.(10) is schematically depicted in Fig. 8 using the vanadates as an example. The upper box summarizes the relationship between the Ln3 + -V5 + IVCT energy and the Ln3 + to V5 + distance, and the lower box shows the dependence of the O2 -Ln3 + CT energy on the O2 to Ln3 + distance. 3.4. The band gap energy and its relationship with electronegativity and ionization potentials For the transition metal oxides, the band gap energy, EðBO3 4 Þ, increases with decrease in value for the 5th ionization potential, I5(B), listed in column 7 of Table 1, and with decreasing value for the Pauling electronegativity, wP(B), of the metal cation B, listed in column 8 of Table 1. A large 5th ionization potential, as for the vanadates, implies a relatively strong binding when an electron is transferred to the B5 + ion, and therefore the band gap energies are relatively low. Besides the properties of the B5 + cation also the size of the A3 + cation within an ABO4 group of compounds (B¼ V,

A = Lu La

d2 ECT ∝ (a/d2 − b)⋅k

Fig. 8. One-dimensional CT transition scheme for the orthovanadates, AVO4:Ln3 + ; the arrows pointing from one ion to another one indicate the direction of electron transfer between the ions. Inside the two boxes the dependence of the O2 -Ln3 + CT (lower box, Ln¼ Eu) energy and the Ln3 + -V5 + IVCT (upper box, Ln¼Pr) energy on the inter-atomic distances d1 and d2 is shown; k is a constant.

Nb, Ta, P) affects the band gap as can be seen in Table 2. This latter observation will be discussed in more detail in the next section.

3.4.1. The band gap of transition metal oxides In the AVO4 rare earth metal orthovanadates it was observed earlier by Blasse and Bril [59,81] and Bril [82] that the maximum of the vanadate (VO3 4 ) emission band depends on the size of the rare earth cation A3 + : the smaller its effective ionic radius, r(A3 + ), with a certain coordination number is, the more the maximum of emission band is shifted towards longer wavelengths. the VO3 4 Also the band gap decreases with decrease in size of A3 + , as can be seen in Table 2. The same correlation was observed for the alkaline earth tungstates, AWO4 (A ¼Ba, Ca, Cd, Cu, Pb, Sr, Zn) [7]: here too the band gap energy depends approximately linearly on the effective ionic radius of the A2 + ions, viz., the smaller the



r(A2 + ), the lower the band gap energy is. Lacomba-Perales et al. argue that the decrease in band gap energy is caused by an increase of the crystal-field splitting of the O2 2p states (it raises the top of the VB) and of the W6 + 5d states (it lowers the bottom of the CB), in combination with a hybridization of these oxygen and tungsten states with the p, d, or f valence electrons of the A2 + cations. A similar argument could hold for the orthovanadates where the CB consists mainly of V5 + 3d states. The orthoniobates and orthotantalates differ from the vanadates and tungstates [7] by the fact that they all have a monoclinic structure, independent of the size of the A3 + cation. In the case of 3+ ) is the tantalates, the correlation between EðTaO3 4 Þ and r(A comparable with the orthophosphates [83]: the smaller r(A3 + ), the larger EðBO3 4 Þ (see Table 2). ScBO4 (B¼Nb, Ta, P) appears to be a 3+ is special case. The relation between EðNbO3 4 Þ and the size of A not clear in the orthoniobates due to the inconsistent assignment of band gap values in the literature (see Table 2). In both the niobates and the M0 -type tantalates, however, the B5 + (B¼Nb, Ta) ions are coordinated by six oxygen atoms. The special structure of the orthoniobates and orthotantalates may indicate a different type of interaction between the A3 + ions and the surrounding oxygen ions. It might be that the binding energy of the oxygen atoms, which is affected by the Coulomb potential and therewith the size of the A3 + ions, might be the dominant parameter in BO4 group transitions. This thought will be explained in more detail for the orthophosphates in Section 3.4.2. In Fig. 9, the observed trends in Eu3 + CT and Pr3 + IVCT energies mentioned in Sections 3.2 and 3.3. are depicted in one energy level scheme. Fig. 9 illustrates that the energy of CT from the VB to a trivalent Ln dopant ion tends to increase as A3 + decreases. The lowest BO3 4 group transition energy, which has been expressed within a simple ionic picture by Eq. (6) in Section 2.3, decreases for B¼V and increases for B¼Ta (see Table 2) with decrease in size of A3 + . Fig. 9 clearly shows why the Pr3 + IVCT energy decreases with decrease in size of A3 + for the orthovanadates, while the Eu3 + CT energy increases simultaneously. In view of Fig. 9, the Pr3 + IVCT energy is expected to show less dependence on the size of A3 + in the tantalates in comparison to the vanadates. This is indeed the case when considering the different Pr3 + IVCT energies for the tantalates and vanadates that have been collected by Boutinaud et al. [57]. The difference between the Pr3 + IVCT energy in LaTaO4:Pr3 + and the one in YTaO4:Pr3 + is negligible, whereas the difference between the corresponding energy in LaVO4:Pr3 + and the one in YVO4:Pr3 + is about 0.31 eV [16]. When we neglect the effect of lattice relaxation after an electron transfer on the location of Ln 4f ground states, the lowest BO3 4 group

A3+ = La3+

Gd3+

Lu3+ Ta5+ V5+

IVCT

ΔE1

Ln32+

Ln12 +

Ln12+

Ln12+

ΔE2 3+

Ln2

ΔE3 Ln32+

CT

Energy

B5+

O22−− Fig. 9. Energy level diagram for the transition metal oxides ABO4:Ln3 + (A ¼La, Gd, Lu), showing the change in O2 -A3 + CT energy and O2 -B5 + CT energy for different host cations, and the effect this change has on both the O2 -Ln31 þ CT and the Ln32 þ -B5 þ IVCT energies, where B ¼V (1) and B ¼Ta (2); the DE indicate the energy difference between the 4f ground states of the Ln21 þ and the Ln32 þ ions.

Table 4 Band

gap

energies,

EA ðBO4 Þ,

and

energy

differences,

DEA ðEu2 þ ,Pr3 þ Þ ¼

3þ Þ þ EIVCT ðPr3 þ ÞEA ðBO3 ECT 4 Þ (see Eq. (9)) for the vanadates, niobates and A ðEu A tantalates. The energies are given in eV.

ABO4

EA ðBO4 Þ

DEA ðEu2 þ ,Pr3 þ Þ

LaVO4 YVO4 LaNbO4 YNbO4 LaTaO4 YTaO4

4.26 3.87 4.84 4.4 4.77 5.5

3.25 4.16 3.6 4.86 4 3.83

transition energy of a transition metal oxide with formula ABO4:Ln3 + (A¼rare earth; B¼V, Nb, Ta) can be expressed as 3þ CT IVCT EA ðBO3 ðLn32 þ ÞDEA ðLn21 þ ,Ln32 þ Þ 4 Þ ¼ EA ðLn1 Þ þ EA

ð11Þ

where DEA ðLn21 þ ,Ln32 þ Þ is ground states of Ln21 þ and

the energy difference between the 4f Ln32 þ . Table 4 compiles both the band gap energy, EA ðBO4 Þ, and the found energy difference DEA ðEu2 þ ,Pr3 þ Þ for the vanadates, the niobates and the tantalates. EA ðBO4 Þ was taken from Table 2, and DEA ðEu2 þ ,Pr3 þ Þ was derived with the help of Eq. (11), whereby 3þ Þ from Table 2 and the values we have taken the values of ECT A ðEu 3þ ðPr Þ from Ref. [16]. of EIVCT A When we compare two compounds with formulas A1BO4:Ln3 + and A2BO4:Ln3 + , where A1 and A2 indicate two different rare earth host cations, then we find the following relations from Table 4: 3 EA1 ðBO3 4 Þ 4 EA2 ðBO4 Þ3DEA1 o DEA2

ð12Þ

Relation (12) is an observation and will necessarily be influenced by the unavoidable variations in the assignment of the CT excitation bands that have to be taken into consideration. 3.4.2. The band gap of non-metal orthophosphates Similar to the orthoniobates and the orthotantalates, also in the orthophosphates the band gap shows an inverse proportionality to the size of A3 + . Both in the monazite and in the xenotime structure of the orthophosphate with formula APO4, oxygen atoms coordinate to two A atoms and one P atom [50]. This means that not only the P5 + ions but also the A3 + ions have, due to their size and charge, an effect on the stabilization of the oxygen ions, which are involved in the PO3 4 group transition: the smaller the A3 + ions and the higher their charge, the larger the electrons binding energy within the O2 ions. This explains the correlation between band gap and size of A3 + as shown in Table 2. But we also have observed another interesting aspect in the orthophosphates that seems to contradict the former explanation: in the VUV excitation spectra of APO4:Ce3 + ,Ln3 + (A ¼Y, Lu; Ln¼ Sm, Dy, Ho, Er, Tm), monitoring Ln3 + emission, we could notice a linear excitation energy and the relation between the lowest PO3 4 effective ionic radius of the Ln3 + co-dopant [83], similar to AVO4 (A¼V, W): the smaller the radius, the lower the PO3 4 transition energy. The local lattice distortion around an activator A2 in a compound with formula A1 PO4 : A32 þ does therefore bring about a fundamental different effect than the same ion would cause in a compound with formula A2PO4.

4. Conclusion Any charge transfer (CT) energy in an inorganic ionic compound can basically be expressed as a difference between two parameters, one referring to the electron donor and the other to the electron acceptor. Jørgensen approximated the CT energy


between a ligand and a central atom with the help of the so-called optical electronegativities of these two ions. His simple expression describes reasonably well the trend in CT energies when going from one family of compounds to another (fluorides, chlorides, bromides, etc.). The CT energy, however, is not constant but varies between different members of these compound families. Boutinaud et al. found a linear relation between the Pr3 + -Bq + (B¼transition metal) inter-valence charge transfer (IVCT) energy and the quotient between the optical electronegativity of the Bq + cation and the Pr3 + Bq + distance. His formula reflects the trend of IVCT energies inside various transition metal oxides. Following the idea of including the inter-atomic distances, we found a linear relation between the O2 -Eu3 + CT energies and the O2 Eu3 + distances in both the transition metal oxides, ABO4:Eu3 + , and the phosphates, APO4:Eu3 + . Within the energy level diagrams that were introduced by Dorenbos a few years back, the lowest BO3 4 group transition energy, i.e., the band gap, should be expressible in the case of the transition metal oxides as the sum of the Eu3 + CT and the Pr3 + IVCT energies minus the difference between the Eu2 + and the Pr3 + ground state energies. With this, it becomes apparent that the electronic properties of the BO3 4 ion depend on the size of the (rare earth) cation adjacent to it. Yet, the band gap of the orthophosphates APO4:Ln3 + cannot be expressed in this manner because they do not show any IVCT excitation bands. Although within the phosphates the band gap energy also shows a dependency on the size of the A3 + ion; their generally larger band gap in comparison to the transition metal oxides cannot be explained with the same parameters that explain the difference in band gap energies between the latter types of oxides, meaning for instance between the vanadates and the tantalates. The fact that the diverse parameters in Table 1, describing structural and electronic properties, are unable to connect the orthophosphates with the transition metal oxides points to the complexity of the nature of CT transitions.

Acknowledgement This work was supported by the Dutch Technology Foundation (STW). References [1] L. Zhang, Z. Hu, Z. Lin, G. Wang, J. Cryst. Growth 260 (2004) 460. [2] N.P. Barnes, M.E. Storm, P.L. Cross, M.W. Skolant, J. Quantum Electron. 26 (1990) 558. [3] J. Liu, Q. Yao, Y. Li, Appl. Phys. Lett. 88 (2006) 173119. [4] M. Yu, J. Lin, S. Wang, Appl. Phys. A: Mater. Sci. Proc. 80 (2005) 353. [5] F.W. Kutzler, D.E. Ellis, Phys. Rev. B 29 (1984) 1008. [6] M. Machida, S. Murakami, T. Kijima, J. Phys. Chem. B 105 (2001) 3289. [7] R. Lacomba-Perales, J. Ruiz-Fuertes, D. Errandonea, D. Martıńez-Garcıá, A. Segura, EPL 83 (2008) 37002. [8] P. Boutinaud, E. Cavalli, M. Bettinelli, J. Phys.: Condens. Matter 19 (2007) 386230. [9] C.K. Jørgensen, Modern Aspects of Ligand Field Theory, North-Holland Publishing Company, Amsterdam, London, 1971. [10] P. Dorenbos, J. Phys.: Condens. Matter 15 (2003) 8417. [11] E. Nakazawa, F. Shiga, Jpn. J. Appl. Phys. 42 (2003) 1642. [12] A.P.B. Lever, Inorganic Electronic Spectroscopy, Elsevier, Amsterdam, 1968. [13] P. Dorenbos, J. Lumin. 111 (2005) 89. [14] P. Boutinaud, R. Mahiou, E. Cavalli, M. Bettinelli, Chem. Phys. Lett. 418 (2005) 181. [15] P. Boutinaud, P. Putaj, R. Mahiou, E. Cavalli, A. Speghini, M. Bettinelli, Spectrosc. Lett. 40 (2007) 209. [16] P. Boutinaud, E. Pinel, M. Oubaha, R. Mahiou, E. Cavalli, M. Bettinelli, Opt. Mater. 28 (2006) 9. [17] A.H. Krumpel, E. van der Kolk, P. Dorenbos, P. Boutinaud, E. Cavalli, M. Bettinelli, Mater. Sci. Eng. B 146 (2008) 114. [18] A.H. Krumpel, E. van der Kolk, E. Cavalli, P. Boutinaud, M. Bettinelli, P. Dorenbos, J. Phys.: Condens. Matter 21 (2009) 115503. [19] S. Sato, J. Phys. Soc. Jpn. 41 (1976) 913.

1365

[20] C.W. Thiel, H. Gruguel, H. Wu, Y. Sun, G.J. Lapeyre, R.L. Cone, R.W. Equall, R.M. Macfarlane, Phys. Rev. B 64 (2001) 085107. [21] C. Pedrini, F. Rogemond, D.S. McClure, J. Appl. Phys. 59 (1986) 1196. [22] J.K. Lawson, S.A. Payne, J. Opt. Soc. Am. B 8 (1991) 1404. [23] P. Dorenbos, J. Phys.: Condens. Matter 15 (2003) 2645. [24] P. Dorenbos, E. van der Kolk, Appl. Phys. Lett. 89 (2006) 061122. [25] E. Nakazawa, J. Lumin. 100 (2002) 89. [26] P. Dorenbos, A.H. Krumpel, E. van der Kolk, P. Boutinaud, M. Bettinelli, E. Bettinelli, Opt. Mater., in press online, doi:10.1016/j.optmat.2010.02.021. [27] S. Mahlik, M. Grinberg, E. Cavalli, M. Bettinelli, P. Boutinaud, J. Phys.: Condens. Matter 21 (2009) 105401. [28] R.D. Shannon, Acta Crystallogr. A 32 (1976) 751. [29] T. Hahn (Ed.), International Tables for Crystallography, vol. A, 2005, DOI: 10.1107/97809553602060000100. [30] J.E. Huheey, E.A. Keiter, R.L. Keiter, Inorganic Chemistry. Principles of Structure and Reactivity, Addison Wesley, 1993. [31] L. Pauling, J. Am. Chem. Soc. 54 (1932) 3570. [32] L. Pauling, The Nature of the Chemical Bond, 3rd ed., Cornell University Press, Ithaka NY, 1960. [33] C.E. Rice, W.R. Robinson, Acta Crystallogr. B 32 (1976) 2232. [34] S. Mahapatra, A. Ramanan, J. Alloys Compd. 395 (2005) 149. [35] X. Wang, I. Loa, K. Syassen, M. Hanfland, B. Ferrand, Phys. Rev. B 70 (2004) 064109. [36] B.C. Chakoumakos, M.M. Abraham, L.A. Boatner, J. Solid State Chem. 109 (1994) 197. [37] S. Tsunekawa, H. Takei, Phys. Status Solidi A 50 (1978) 695. [38] J.W.E. Mariathasan, Acta Crystallogr. B 41 (1985) 179. [39] V.K. Trunov, L.N. Kinzhibalo, Dokl. Akad. Nauk SSSR 263 (1982) 348. [40] S. Tsunekawa, T. Kamiyama, K. Sasaki, H. Asano, T. Fukuda, Acta Crystallogr. A 49 (1993) 595. ¨ [41] H. Weitzel, H. Schrocke, Z. Kristallogr. 152 (1980) 69. [42] A.G. Beda, A.A. Bush, A.F. Volkov, V.F. Meshcheryakov, Kristallografiya 47 (2002) 357. [43] T.A. Kurova, V.B. Aleksandrov, Dokl. Akad. Nauk SSSR 201 (1971) 1095. ¨ [44] I. Hartenbach, F. Lissner, T. Nikelski, S.F. Meier, H. Muller-Bunz, T. Schleid, Z. Anorg. Allg. Chem. 631 (2005) 2377. [45] X. Liu, X. Xu, M. Gu, L. Xiao, K. han, R. Zhang, Appl. Surf. Sci. 253 (2007) 4344. [46] L.H. Brixner, H.-Y. Chen, J. Electrochem. Soc. 130 (1983) 2435. [47] R.B. Chen, Can. Mineral. 6 (1957) 72. [48] G. Blasse, G.J. Dirksen, L.H. Brixner, M.K. Crawford, J. Alloys Compd. 209 (1994) 1. [49] D.F. Crawford, D.A. Grossie, Inorg. Chim. Acta 95 (1984) 231. [50] Y. Ni, J.M. Hughes, A.N. Mariano, Am. Mineral. 80 (1995) 21. [51] D.F. Mullica, D.A. Grossie, L.A. Boatner, Inorg. Chim. Acta 109 (1985) 105. [52] W.O. Milligan, D.F. Mullica, Inorg. Chim. Acta 60 (1982) 39. [53] L.H. Brixner, J.F. Whitney, F.C. Zumsteg, G.A. Jones, Mater. Res. Bull. 12 (1977) 17. [54] H. Takei, S. Tsunekawa, J. Cryst. Growth 38 (1977) 55. [55] G. Blasse, A. Bril, Philips Tech. Rev. 31 (1970) 304. [56] G.M. Wolten, Acta Crystallogr. 23 (1967) 939. [57] P. Boutinaud, R. Mahiou, E. Cavalli, M. Bettinelli, J. Lumin. 122 (2007) 430. [58] L. van Pieterson, A. Meijerink, J. Alloys Compd. 300 (2000) 426. [59] G. Blasse, A. Bril, J. Chem. Phys. 50 (1969) 2974. [60] Y.J. Hsiao, T.H. Fang, Y.S. Chang, Y.H. Chang, C.H. Liu, L.W. Ji, W.Y. Jywe, J. Lumin. 126 (2007) 866. [61] X.M. Liu, J. Lin, J. Lumin. 122 (2007) 700. [62] W.J. Schipper, M.F. Hoogendorp, G. Blasse, J. Alloys Compd. 202 (1993) 283. [63] S.K. Lee, H. Chang, C.H. Han, H.J. Kim, H.G. Jang, H.D. Park, J. Solid State Chem. 156 (2001) 267. [64] X. Xiao, B. Yan, J. Non-cryst. Solids 351 (2005) 3634. [65] G. Blasse, G.J. Dirksen, Mater. Chem. Phys. 14 (1986) 485. [66] G. Blasse, A. Bril, J. Lumin. 3 (1970) 109. [67] B. Li, Z. Gu, J. Mater. Sci. 32 (2000) 1139. [68] M.J.J. Lammers, G. Blasse, Mater. Res. Bull. 19 (1984) 759. [69] M.H. Hwang, Y.J. Kim, Ceram. Int. 34 (2008) 1117. [70] L.H. Brixner, J. Chem. Edu. 57 (1980) 588. [71] U. Sasum, M. Kloss, A. Rohmann, L. Schwarz, D. Haberland, J. Lumin. 72 (1997) 255. [72] E. Nakazawa, F. Shiga, J. Lumin. 15 (1977) 255. [73] W. van Schaik, S. Lizzo, W. Smit, G. Blasse, J. Electrochem. Soc. 140 (1993) 216. [74] A. Trukhin, L.A. Boatner, Mater. Sci. Forum 239 (1997) 573. [75] A.R. Protheroe, A. Steinbrunn, T.E. Gallon, Surf. Sci. 126 (1983) 534. [76] S.F. Matar, G. Campet, J. Phys. Chem. Solids 68 (2007) 331. [77] J.A. Duffy, J. Phys. C: Solid State Phys. 13 (1980) 2979. [78] O.E. Taurian, M. Springborg, N.E. Christensen, Solid State Commun. 55 (1985) 351. [79] M. Springborg, O.E. Taurian, J. Phys. C: Solid State Phys. 19 (1986) 6347. [80] P. Dorenbos, J. Lumin. 91 (2000) 91. [81] G. Blasse, A. Bril, J. Chem. Phys. 48 (1968) 217. [82] G. Blasse, Philips Res. Rep. 23 (1968) 344. [83] A.H. Krumpel, A.J.J. Bos, A. Bessie re, P. Dorenbos, yy. [84] S.Z. Toma, D.T. Palumbo, J. Electrochem. Soc. 117 (1970) 236.