Physics and current understanding of
X-ray storage phosphors
DER UNIVERSITÄT-GESAMTHOCHSCHULE PADERBORN
ZUR ERLANGUNG DER LEHRBERECHTIGUNG
(VENIA LEGENDI)
FÜR DAS FACH
EXPERIMENTALPHYSIK
vorgelegte Habilitationsschrift
von
STEFAN SCHWEIZER
MAI 2000 (Eröffnung des Habilitationsverfahrens)
NOVEMBER 2000 (Habilitationsvortrag)
DEZEMBER 2000 (Abschluss des Habilitationsverfahrens)
i
Table of Contents
Introduction 1
1 Experimental fundamentals 5
1.1 Magneto-optical measurement techniques ................................................... 5
1.1.1 Magnetic circular dichroism of the optical absorption ..................... 5
1.1.2 Optical detection of electron paramagnetic resonance (EPR) and
electron nuclear double resonance (ENDOR) .................................. 8
1.1.3 Cross-relaxation spectroscopy ........................................................ 10
1.2 Conventional EPR and ENDOR ................................................................. 11
1.3 Analysis of EPR and ENDOR spectra ........................................................ 11
1.3.1 The spin Hamiltonian ..................................................................... 11
1.3.2 Analysis of EPR spectra ................................................................. 15
1.3.3 Analysis of ENDOR spectra ........................................................... 15
1.3.4 Calculation of powder EPR and ENDOR spectra .......................... 16
2 X-ray storage phosphors 19
2.1 Performance ................................................................................................ 19
2.2 Spatial resolution of X-ray storage phosphor image plates ........................ 21
2.2.1 Image sharpness .............................................................................. 21
2.2.2 Modulation transfer function .......................................................... 21
2.2.3 Description of imaging processes via Fourier transformation ........ 24
2.2.4 Measurement of the spatial resolution and the modulation
transfer function .............................................................................. 25
2.3 Read-out process of X-ray storage phosphor image plates ........................ 27
ii
3 The X-ray storage phosphor BaFBr:Eu2+ 31
3.1 Stoichiometric BaFBr ................................................................................ 31
3.1.1 Generation of electron trap centres ................................................ 31
3.1.2 Generation of hole trap centres ...................................................... 33
3.1.3 PSL active hole trap centres ........................................................... 35
3.2 Non-stoichiometric BaFBr ......................................................................... 38
3.2.1 Photostimulated luminescence ....................................................... 40
3.2.2 Identification of fluorine antisites with MAS-NMR ...................... 41
3.2.3 Identification of electron and hole trap centres with EPR ............. 43
3.2.4 Generation of electron and hole trap centres .................................. 45
3.3 Red-shift of the PSL excitation upon Ca2+ or Sr2+ doping ........................ 49
3.4 Surrounding of the activator Eu2+ .............................................................. 53
3.4.1 EPR and ENDOR of crystalline BaFBr:Eu2+ ................................. 53
3.4.2 EPR and ENDOR of powdered BaFBr:Eu2+ ................................. 54
3.4.3 Influence of the production process ............................................... 56
4 Alkali halides and elpasolites 59
4.1 KBr:In+ ....................................................................................................... 60
4.1.1 Generation of electron and hole trap centres .................................. 62
4.2 RbI:Tl+........................................................................................................ 64
4.2.1 Generation of electron and hole trap centres .................................. 64
4.3 RbBr:Ga+ and CsBr:Ga+ ............................................................................ 66
4.3.1 Sample preparation ......................................................................... 66
4.3.2 Generation of electron and hole trap centres .................................. 67
4.3.3 Generation of (Ga2+)I and (Ga2+)II centres in RbBr:Ga+ ............... 68
4.3.4 PSL experiments with RbBr:Ga+ and CsBr:Ga+ ............................ 71
4.3.5 Optimal activator concentration and Ga+ aggregation ................... 73
4.3.6 PSL fading ...................................................................................... 74
4.3.7 Red-shift of the PSL excitation ...................................................... 76
4.4 RbBr:Eu2+ and CsBr:Eu2+ .......................................................................... 80
4.5 The elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+ ................................ 81
i
ii
5 Glasses and glass ceramics 87
5.1 Production of fluorozirconate glasses and glass ceramics .......................... 87
5.2 X-ray diffraction on fluorozirconate glasses and glass ceramics ............... 88
5.3 Photoluminescence and PSL of fluorozirconate glasses and glass
ceramics....................................................................................................... 89
5.4 Electron and hole trap centres .................................................................... 90
Summary 93
References 97
1
Introduction
In the beginning of radiation diagnostics, conventional photographic films were used for
X-ray imaging. Due to their poor sensitivity to X-rays, high doses had to be used for the
imaging process in medical radiography. Significant progress was achieved by the use
of a combination of intensifying screens coated with a scintillator layer and a photo-
graphic film. The scintillator screen plays the role of an X-ray to optics converter (X-ray
luminescence), and the film is as before the image storing part of the system. The latent
X-ray image becomes visible after appropriate chemical processing of the film.
In contrast to the X-ray scintillators, where the X-ray energy is directly converted into
visible light, X-ray storage phosphors store the radiation image in proportion to its in-
tensity distribution in a storage phosphor screen. The screen is coated with X-ray stor-
age phosphor crystallites which are imbedded in an organic binder. Upon X-irradiation,
complementary defects (electron and hole trap centres) are generated in these crystal-
lites. The electron of the electron trap centre can be stimulated afterwards and it recom-
bines with the complementary hole trap centre. The recombination energy is often trans-
ferred to a doped activator which emits light of a characteristic photon energy. This pro-
cess is called photostimulated luminescence (PSL).
Since the sensitivity of the X-ray storage phosphors exceeds that of the scintillator-film-
system by at least one order of magnitude, the applied X-ray dose can be reduced sig-
nificantly. For the best X-ray storage phosphors the dose dependence shows linearity for
more than five orders of magnitude. Under- and over-exposures caused by the S-shaped
sensitivity characteristics of all the photographic films including the scintillator-film-
system can thus be avoided. The image is digitised directly and can be archived easily.
Due to the scattering effects of the stimulation light during the read-out process, the
spatial resolution of X-ray storage phosphor screens is still inferior compared to that of
scintillator-film-systems.
Introduction
2
For practical use in commercial X-ray imaging systems the requirements for X-ray stor-
age phosphors are somewhat different from those for scintillator-film based systems.
Besides a high X-ray absorption, which can be achieved by the incorporation of ele-
ments with high mass number, the generated complementary defects should be ther-
mally stable at room temperature, to make sure that the image information is stored until
it is read out, possibly several hours later. The wavelength of the stimulation light
should be in a spectral range which is clearly separated from the spectral range of the
emitted light of the activator. Moreover, it would be advantageous if the stimulation
light was in a spectral range where low-cost lasers are available as light sources. The
emission of the activator should be in a spectral range where it can be detected easily
and sensitively. The radiative life time of the activator as well as the bleaching time of
the defects determine the time necessary for the read-out process. The re-usability of
such X-ray storage phosphor screens is an important advantage. It should thus be possi-
ble to erase all defects in the crystallites after the read-out process has been accom-
plished.
The so far best-known X-ray storage phosphor is BaFBr doped with Eu2+ as activator
[1]. In spite of many efforts to understand the information storage and read-out proc-
esses, no complete understanding has yet been achieved (see e.g., [2, 3, 4, 5]). The
lower spatial resolution of commercially-used X-ray imaging systems based on BaFBr
is partially due to the matlockite structure of the crystallites in the storage phosphor
layer. During the read-out process the stimulation light is not only scattered due to the
different refractive indices of the organic binder and the phosphor crystallites, but also
due to the statistical distribution of the birefringent crystallites in the screen. Optically
isotropic crystals such as e.g. cubic alkali halides would possibly minimise the scatter-
ing effects in the storage phosphor layer. It is thus a challenge to find optically isotropic
storage phosphor materials which have the same performance as BaFBr:Eu2+.
One of the suitable cubic alkali halides is KBr:In+, which was shown to have storage
and PSL properties [6]. From the practical point of view it will not be a very important
system due to its low X-ray stopping power. RbI:Tl+ has, in principle, very good storage
and PSL properties [7]. However, the stored image fades away at room temperature
within minutes, so that a very fast read-out process has to be initiated immediately after
taking the image. It has been shown that the alkali halides RbBr:Ga+ and CsBr:Ga+ have
excellent figures of merit as X-ray storage phosphors [8]. They are optically isotropic
and therefore very promising. The elpasolite Cs2NaYF6:Ce3+ is also comparable to
BaFBr:Eu2+ with respect to X-ray conversion and necessary stimulation energy [9].
Introduction
3
The observation that upon X-irradiation the electron and hole trap centres are always
created with a spatial correlation to the activator is of particular interest for the func-
tioning of the storage phosphors. It is statistically not very probable that for low medical
doses the photostimulated electron of the electron trap centre and a hole of the hole trap
centre recombine near an activator. A phosphor does not function, unless electron and
hole trap centres are generated near the activator. The reason for this spatial correlation
of the radiation damage centres with the activator is so far not understood and repre-
sents a very interesting, fundamental open question.
In chapter 1 the experimental fundamentals of the measurement techniques used are
described. Chapter 2 deals with the characterisation of X-ray storage phosphors such as
performance and spatial resolution, as well as the experimental setup for the read-out
process.
In chapter 3 one finds a detailed description of the information storage and read-out
process in the X-ray storage phosphor BaFBr:Eu2+ emphasising the generation mecha-
nism and structure of electron and hole trap centres, whilst alternative systems such as
alkali halides and elpasolites, and glasses or glass ceramics are described in chapter 4
and 5, respectively.
Introduction
4
5
Chapter 1
Experimental fundamentals
The main experimental method used to analyse the nature of the generated electron and
hole trap centres, which are the basis of the information storage and read-out processes,
is electron paramagnetic resonance (EPR), because electron and hole trap centres are
paramagnetic before they recombine in the photostimulated luminescence process. For
detailed defect investigations, electron nuclear double resonance (ENDOR) is also
needed. However, because of low defect concentration conventional EPR and ENDOR
are often not sensitive enough. Therefore, optically detected EPR and ENDOR tech-
niques have to be applied. Their basis is the detection of changes of the magnetic circu-
lar dichroism of the optical absorption (MCDA) induced by EPR / ENDOR transitions.
In the following only a brief description can be given of the most important methods.
For further details the reader is referred to e.g. [10].
1.1 Magneto-optical measurement techniques
1.1.1 Magnetic circular dichroism of the optical absorption
The magnetic circular dichroism of the optical absorption (MCDA) is the difference
between the absorption of right and left circularly polarised light. The light is propa-
gated parallel to an external magnetic field B in which the sample is situated. The
MCDA is defined by
εω
=−
d
ckk
2rl
/4 (1.1)
Experimental fundamentals
6
where kr and kl are the absorption indices for right and left circularly polarised light, d
the thickness of the crystal,
ω
the angular frequency of the light and c the speed of light.
The measured quantities are the energy dependent absorption coefficients
α
(E). With
kE c
EE()
=()
2
α
(1.2)
and E=
!
ω
equation (1.1) becomes
εαα
=−
d
4rl
/4
. (1.3)
Assuming an exponential decay of the light intensity within the crystal, according to
Lambert-Beer’s law the measured light intensity is
II d
r,l o r,l
=−
exp
α
05
, (1.4)
where Io is the intensity of the incident light and thus the sum of the incident right and
left circularly polarised light. Transformation of equation (1.4) leads to
α
r,l o
r,l
=1
d
I
I
ln . (1.5)
Insertion of this result into equation (1.3) one obtains
ε
=1
4ln I
Ir
l. (1.6)
With the average intensity III
arl
=+
/4
2 and the intensity difference ∆II I
=−
rl
equation (1.6) becomes
ε
=+
−
1
4
1
1
2
2
ln
∆
∆
I
I
I
I
a
a
. (1.7)
Assuming ∆II<< a, i.e. d
αα
rl
<<1
−
/4, equation (1.7) simplifies to
ε
≈=−
+
∆I
I
II
II4a
rl
rl
2/4
. (1.8)
The MCDA consists of a paramagnetic and a diamagnetic part.
εεε
BT B BT,,
()
=()
+()
dia para (1.9)
1.1 Magneto-optical measurement techniques
7
The paramagnetic part is both temperature and magnetic field dependent. The diamag-
netic part, which is caused by non-resolved Zeeman splittings of the excited states [11],
is temperature independent, but also proportional to the external magnetic field value B.
For S=12 systems the paramagnetic MCDA is proportional to the polarisation P of
the ground state. The polarisation P can be calculated using the Langevin function
εµ
para B
∝= −
+=
−+
−+
Pnn
nn
gB
kT
tanh 2. (1.10)
n+ and n− are the occupation numbers of the mS=±12 states, g is the Landé g-factor,
µ
B the Bohr magneton, B the magnetic field value and T the temperature. For S>12
systems the polarisation P is given by the Brillouin function (see e.g., [10]).
Figure 1.1 schematically represents the circularly polarised transitions for an alkali atom
model, which is often used to describe the MCDA of an F centre (electron trapped in an
anion vacancy) in alkali halides [12]. The excited state is split by the spin-orbit interac-
tion into a J=12 and a J=32 state. The degeneracy is lifted by the applied magnetic
field B. Neglecting the Zeeman splittings of the excited states one absorption band is
observed for right and one for left circularly polarised light. The intensity is equal for
both bands. Both absorption bands, the separation of which is determined by the spin-
orbit interaction, are identical in their respective shapes which are taken to be Gaussians
in this model. Phonon-broadening of the transitions is taken into account via the “rigid
shift” approximation [13]. For the selection rule mJ=±1 and a small spin-orbit inter-
action compared to the half-width of the right and left circular absorption bands this
leads to a derivative-like structure of the resulting MCDA (see e.g., [10]).
Experimental fundamentals
8
1.1.2 Optical detection of electron paramagnetic resonance (EPR) and
electron nuclear double resonance (ENDOR)
As described above the paramagnetic part of the MCDA is proportional to the spin po-
larisation of the ground state. Perturbing the thermal equilibrium occupation of the
Zeeman levels at a fixed optical wavelength by applying a resonant and partly saturating
electron paramagnetic resonance (EPR) transition, the spin polarisation is changed and,
Figure 1.1 Schematic diagram of the energy levels in the alkali atom model to explain
the MCDA of an F centre [12]. The ∆mJ = +1 transitions absorb right, and
the ∆mJ = −1 transitions, left circularly polarised light. The energy differ-
ence between the S and the P states amounts to several eV, the spin-orbit
splitting ∆LS to several meV up to 1 eV depending on the nature of the ani-
ons. The numbers on the transition arrows indicate the relative transition
probabilities.
1.1 Magneto-optical measurement techniques
9
according to equation (1.10), the paramagnetic MCDA decreases (figure 1.2). In an op-
tically detected EPR experiment the changes of the detected MCDA are measured. A
fixed optical wavelength and a fixed saturating microwave field are used and the exter-
nal magnetic field is swept through the resonance conditions. EPR lines due to defects
are then observed as changes in the magnetic field dependent MCDA signal (MCDA-
detected EPR) at this optical wavelength. This technique, the optical detection of EPR,
can increase the sensitivity considerably depending though on the optical transition
probability (oscillator strength) and the spin-orbit splitting of the excited states. Moreo-
ver, the coupling of MCDA and EPR allows the assignment of optical transitions to
paramagnetic defects. In particular, it is possible to separate superimposed MCDA
bands, which are due to different defects, by their distinct MCDA-detected EPR spectra.
When recording the wavelength-dependent ODEPR effect under resonant conditions,
the MCDA “excitation” spectrum of a defect is obtained. This procedure is known as
“tagged” MCDA (see e.g., [10]).
Figure 1.2 Schematic diagram of the energy levels of a paramagnetic centre having
S=12 and I=12
to explain optically detected EPR and ENDOR.
Experimental fundamentals
10
In addition to the EPR transitions the ground state spin polarisation can be affected by
applying simultaneously a resonant radio frequency (rf) field to induce NMR transi-
tions. The NMR transitions between hyperfine (hf) or superhyperfine (shf) Zeeman lev-
els also change the spin polarisation. Therefore also electron nuclear double resonance
(ENDOR) can be detected optically via the change of the MCDA. Information about the
defect structure can be obtained from the measured hf and shf interaction (see e.g.,
[10]).
1.1.3 Cross-relaxation spectroscopy
For the functioning of the storage phosphors it is necessary that the X-ray induced elec-
tron and hole trap centres are generated not too far from another and they are also spa-
tially correlated to the activators. To investigate this experimentally one can use cross-
relaxation spectroscopy. The cross-relaxation probability between two spin systems
assuming a dipole-dipole interaction is given by [14]
Rg
ij ij
=−
22
&
α
β
with &ij i j ij ij
gg r
2224 2 2 6
3=−
βϑ
( cos ) (1.11)
and ggg
α
β
α
β
ννδνννν
=′′′′
−′′ ′ ′′
H() ( )( )dd ,
where g
αβ
is the overlap integral of the shape function of the EPR lines of both defects
types,
α
and
β
, which can be determined experimentally from the EPR spectra. The in-
dices i and j characterise the individual defect of each type taking part in the cross-
relaxation (CR), with a separation rij between defects i and j and an angle
ϑ
ij between
the ij connecting line and the magnetic field.
β
is the gyromagnetic ratio in the CGS
system and gi and gj are the electronic g factors.
According to equation (1.11) the cross-relaxation probability depends significantly on
the separation between the two interacting spin-systems. Thus, it is possible to deter-
mine the distribution of the separations between different paramagnetic defects, how-
ever, only in case of diluted spin-systems. The dynamical behaviour of the spin polari-
sation of a paramagnetic defect can then be calculated by a set of rate equations de-
scribing the occupations of the Zeeman levels. The occupations are influenced by EPR
transitions and spin-lattice relaxations. If two spin systems are coupled by CR, addi-
1.3 Analysis of EPR and ENDOR spectra
1
1
tional terms due to CR enter the rate equations, which become non-linear. For a high
concentration of paramagnetic defects the cross-relaxation effects are mainly deter-
mined by the average distance between the interacting spin-systems. In this case one
does not obtain any further information about the distribution of separations of the dif-
ferent spin-systems. For details see e.g., [15].
1.2 Conventional EPR and ENDOR
In contrast to optically detected EPR, the EPR transitions in conventional experiments
are detected via microwave absorption. In ENDOR the detection of NMR transitions is
achieved via the change of the partly saturated EPR effect. To observe an ENDOR ef-
fect, the EPR transition has to be saturated partly by applying sufficiently high micro-
wave power. The rf-induced NMR transition leads to a partial desaturation of the EPR,
which is compensated by an increased microwave absorption (figure 1.3). The latter is
detected (see e.g., [10]). In stationary ENDOR experiments [16] one can make use of a
cross-relaxation probability Tx
−1, which allows the stationary observation of the rf-
induced desaturation of the EPR transition.
1.3 Analysis of EPR and ENDOR spectra
1.3.1 The spin Hamiltonian
For the analysis of EPR and ENDOR spectra an appropriate spin Hamiltonian is used
(see e.g., [10]), which allows the calculation of the energy levels of the spin system. The
spin Hamiltonian operator & is the sum of several operators: The electron Zeeman op-
erator &EZ, the fine-structure operator &FS, the hyperfine structure operator &HF, the
superhyperfine structure operator &SHF, the nuclear Zeeman operator &NZ, and the nu-
clear quadrupole operator &Q.
&=⋅⋅+ + ⋅⋅−⋅⋅+⋅⋅
==−
∑∑ ∑
µµ
Bnn
SB S I S I BI I
gAgQ
k
S
k
qk
q
qk
k
iii i i iii
i
bO
2
2() ,
:?
&EZ &FS &HF/SHF &NZ &Q(1.12)
with Bmagnetic field vector,
µ
BBohr magneton,
Experimental fundamentals
12
µ
nnuclear magneton,
Selectron spin operator,
Iinuclear spin operator of nucleus i,
gelectron g tensor,
gn,inuclear g tensor of nucleus i,
bk
qStevens parameter,
Ok
qStevens operator,
Aihyperfine or superhyperfine structure tensor of nucleus i,
Qiquadrupole tensor of nucleus i.
The electron Zeeman term &EZ describes the magnetic interaction between the electron
spin S and the external magnetic field B, the nuclear Zeeman term &NZ the interaction
between the nuclear spin I and the magnetic field B. The fine-structure term &FS is a
sum of Stevens operators [17] and describes the spin-spin interaction within a spin sys-
tem having S>12. Due to time reversal symmetry the sum indices q are even num-
bers. The use of Stevens operators in the fine-structure term has the advantage that they
obey certain symmetry restrictions, i.e. some of the Stevens parameters vanish for cer-
tain point group symmetries. For S≤3 2 the fine-structure expression can be reduced
to SS
⋅⋅
D. The interaction between the electron spin S and the nuclear spin I of the
Figure 1.3 Schematic diagram explaining the EPR and ENDOR effect of a simple
system having S=1 2 and I=12
.
The relative occupation of levels 1-4
is marked by their lengths considering a non-saturated EPR transition (fig-
ure 1.3a) and the stationary occupation for saturated EPR and NMR tran-
sitions, including possible relaxation mechanisms (figure 1.3b).
1.3 Analysis of EPR and ENDOR spectra
1
3
central nucleus or a neighbour nucleus is described by the hyperfine (hf) or superhyper-
fine (shf) structure term &HF / SHF. The quadrupole term &Q contains the interaction
between the nuclear quadrupole moment of a nucleus having I>12 and the electric
field gradient at its site.
The hf or shf tensor A can be split into an isotropic part a and a traceless anisotropic
part B.
AEB
=⋅+a, (1.13)
where E is the 3×3 unit matrix. Since the trace of the anisotropic hf or shf tensor is zero,
the tensor can be described in its principal axes system by the two independent interac-
tion parameters
bBzz
=2(1.14)
and bByy
'Bxx
=−
2.
The scalar a, the so-called Fermi contact interaction constant, is proportional to the un-
paired spin density at the site of a nucleus in a one-particle approximation for the centre
wave function
ψ
(r) [13].
agg=⋅
()
2
302
µµµψ
oeBn n , (1.15)
where
µ
o is the permeability of a vacuum. The matrix elements of the anisotropic part B
are given by [13]
Bgg
xx
rr
ik ik ik
=⋅−
$
%
&'
(
)()
H
1
4
3
53 2
πµµ µ δψ
oeBn n drr. (1.16)
The quadrupole tensor Q, which does not vanish for nuclei with spin I>1 2, is defined
by
QeQ
II
V
xx
ik ik
r
=−⋅∂
∂∂ =
22 1
2
0
() (1.17)
with eelementary charge,
Qquadrupole moment of a nucleus,
Velectric potential at the site of a nucleus.
Experimental fundamentals
14
The quadrupole tensor Q is also traceless and can be described analogously to the ani-
sotropic hf or shf tensor in its principal axes system by the two parameters
qQzz
=′′
2(1.18)
and qQyy
'Qxx
=−
′′ ′′
2.
The z axis of the hf, shf or quadrupole tensor’s principal axes system is to be aligned
along the direction of the largest interaction by definition. Consequently, b and q yield
the axially symmetric parts of each tensor, whereas ′
b and ′
q describe the deviation
from axial symmetry.
Since the principal axes system of the interaction tensors (x, y, z) does often not coin-
cide with the crystalline axes system or the laboratory system (a, b, c), the relative posi-
tion of the tensor principal axes system can be described by Euler angles (figure 1.4).
The exact determination of the energy levels of a spin system requires a full diagonali-
sation of the appropriate spin Hamiltonian. For a rough estimation perturbation theory
of first or second order is often helpful, provided that one term of the Hamiltonian
dominates. For example if &EZ >> &SHF the interaction for a simple S=12 system
with isotropic electronic and nuclear g-factors and one neighbour nucleus is given in
first order perturbation theory by
Figure 1.4 Definition of the Euler angles describing the tensor orientations. a, b, and c
are the crystal axes, whereas x, y, and z are the principal axes of the tensor.
The ab plane cuts the xy plane along line S.
1.3 Analysis of EPR and ENDOR spectra
1
5
EmgBmmW mg B Wm II
SSII I
=+ − +−
+
()
$
%
&'
(
)
µµ
BSHFnnQ
1
2
1
3
2(1.19)
with Wab b
SHF =+⋅ −+⋅ ⋅ ()
31 2
22
cos sin cos
ϑϑφ
27
'
and Wq q
Q=⋅ −+⋅ ⋅()
33 13 2
22
cos sin cos
ϑϑφ
''''
27 .
ϑ
and ′
ϑ
are the angles between the z axis of the respective principal axes system and
the magnetic field vector,
φ
and ′
φ
the angles between the x axis and the projection of
the magnetic field vector into the xy plane of the respective principal axes system (see
e.g., [10]).
1.3.2 Analysis of EPR spectra
EPR spectra are analysed by determining the energy differences between the levels for
which transitions obeying the selection rules ∆mS=±
1 and ∆mI=0 occur. For a
S=12 system equation (1.19) yields the energy positions of the allowed EPR transi-
tions according to first order perturbation theory.
hgBmW
I
νµ
EPR B SHF
=+ (1.20)
In an EPR experiment the microwave frequency is fixed while the magnetic field is
swept. Thus equation (1.20) has to be solved for the corresponding resonant magnetic
fields. The interaction matrices can be obtained by rotating the magnetic field vector in
two perpendicular planes (or rotating the crystal relative to the fixed magnetic field).
The angular dependencies usually yield much more resonances than needed for the de-
termination of the spin Hamiltonian parameters. Therefore, the interaction tensors are
adjusted to the observed resonances in an iterative procedure where the weighted sum of
the deviation squares of the measured and the calculated resonances is minimised.
1.3.3 Analysis of ENDOR spectra
ENDOR transitions obey the selection rules ∆mS=0 and ∆mI=±1. According to
equation (1.19) first order theory yields
hmWgBmW
SQ
νµ
ENDOR SHF n n Q
=−+ (1.21)
Experimental fundamentals
16
with mmm
QII
=+'
/4
2.
mI and mI' represent the nuclear spin states between which the ENDOR transition oc-
curs. By analogy to the EPR analysis also here the corresponding interaction matrices
are obtained by recording ENDOR angular dependencies for a rotation of the magnetic
field vector in two perpendicular planes and fitting the interaction parameters to the
experimental data afterwards.
According to their separation from the defect centre, neighbouring nuclei are classified
in different shells. The symmetry of a shell is determined by symmetry elements which
transform the shell’s nuclei into each other. The origin of each symmetry operation is
the point group symmetry of defect centre and a centre of inversion, i.e., is determined
by the Laue class of the centre (site). Each nucleus of a shell leads to a certain ENDOR
line and thus to an ENDOR branch in the angular dependence.
1.3.4 Calculation of powder EPR and ENDOR spectra
With the data known from a single crystal, the corresponding powder EPR spectrum can
be calculated. For the individual EPR lines and inhomogeneous broadening, one can
assume a Gaussian line shape ΩEPR according to
ΩEPR EPR
o
EPR
exp() (,)
BBB
=−
−
$
%
&
'
(
)
1
2
1
2
2
2
σπ
ϑφ
σ
/4 (1.22)
with
σ
EPR =∆B
22 2ln
whereby ∆B is the full width at half maximum of the EPR line. Since the EPR line po-
sitions Bo are angular dependent, one has to integrate over all possible orientations of
the magnetic field.
For the calculation of powder ENDOR spectra, one has to consider that the intensity of
an ENDOR line at a certain magnetic field is proportional to the intensity of an EPR
line at that field. Therefore, the intensity of the ENDOR signals at the applied magnetic
field BENDOR has to be weighted with the intensity of the EPR line at this field. An
ENDOR frequency depends on the orientation of the magnetic field with respect to the
principal axes of the interaction tensors. Therefore, in order to calculate the intensity of
1.3 Analysis of EPR and ENDOR spectra
1
7
an ENDOR line for a certain frequency
ν
at a magnetic field BENDOR and assuming a
Gaussian line shape, one obtains
ΩENDOR EPR
o ENDOR
EPR
exp () ,
νσπ
ϑ
φ
σ
=−
()
−
$
%
&
'
(
)×
1
2
1
2
2
2
BB
/4
1
2
1
2
2
2
σπ
νϑ
φ
ν
σ
ENDOR
ENDOR ENDOR
ENDOR
exp −−
$
%
&
'
(
)
,,B
/4/4
(1.23)
where the polar angle
ϑ
and the azimuthal angle
φ
describe the orientation of the static
magnetic field Bo with respect to the crystal axes. In order to obtain the powder ENDOR
spectrum as a function of the rf frequency
ν
, one has to integrate equation (1.23) over
all angles of
ϑ
and
φ
(see e.g., [18]).
Experimental fundamentals
18
19
Chapter 2
X-ray storage phosphors
2.1 Performance
To determine the performance of an X-ray storage phosphor a read-out experiment must
be performed (figure 2.1). After X-irradiation at RT the sample is excited continuously
with the appropriate stimulation light. A photomultiplier detects the resulting pho-
tostimulated luminescence (PSL) versus time. For detailed information see e.g. [19].
Continuous photostimulation leads to a decrease of the number of the PSL-active cen-
tres. Consequently, the PSL intensity decays under continuing stimulation. The area
below the PSL curve is proportional to the absorbed X-ray dose. In a commercial sys-
tem a laser beam is used to read out the information stored in the phosphor screen point
by point and line by line. The information should be read-out in the shortest time possi-
ble. To describe the characteristics of an X-ray storage phosphor the quantities “conver-
sion efficiency” (CE) and “stimulation energy” (SE) determining these requirements are
introduced.
CE is defined as the released photon energy (Etot) per absorbed X-ray dose
CE tot
=E
absorbedX-raydose , (2.1)
where Etot is the area below the PSL curve.
In the simplest case the time dependence of the PSL can be described by a monoexpo-
nential decay, i.e.
X-ray storage phosphors
20
It I t( ) exp( )=⋅ −
0
τ
. (2.2)
Io is the amplitude of the PSL at the beginning of the stimulation experiment, t the
stimulation time and
τ
a time constant depending on the power P of the stimulation light
(see definition of stimulation energy) and limited by the characteristic decay time of the
activator luminescence.
Thus, Etot is obtained by integrating I(t) from t = 0 to t = ∞,
Etot =⋅ −
∞
H
Itt
0dexp( )
τ
0
=⋅⋅ − − − ∞
It t
0
ττ τ
exp( ) exp( )
0
:?
=⋅I0
τ
. (2.3)
Finally, CE is determined by
CE absorbedX-raydose
=⋅I0
τ
. (2.4)
In the experiment the time constant
τ
, after which the PSL intensity is decreased to 1e
of its initial value, as well as the initial PSL intensity I0 are measured.
Figure 2.1 Typical PSL decay curve of an X-ray storage phosphor under continued
stimulation.
2.2 Spatial resolution of X-ray storage phosphor image plates
2
1
The stimulation energy (SE), i.e. the energy required to reduce the PSL intensity to 1e
of its initial value, is defined as
SE =⋅
τ
P, (2.5)
where P is the power of the applied stimulation light beam.
The sensitivity of an X-ray storage phosphor is proportional to the stored energy and
proportional to the inverse of energy required for the read-out. Therefore, the sensitivity
is given by the ratio CE / SE.
2.2 Spatial resolution of X-ray storage phosphor image
plates
2.2.1 Image sharpness
The image of an infinitely sharp edge obtained by an X-ray imaging system based on
storage phosphor screens should have a sharp step in the detected luminescence inten-
sity from zero to a certain luminescence intensity level proportional to the absorbed X-
ray dose (figure 2.2a (left)). Unfortunately, this is only valid in an ideal case whereas in
reality, the image does not yield such a sharp step, but a region of unsharpness in which
the luminescence signal changes steadily from zero to the corresponding luminescence
intensity (figure 2.2b (left)). The sharpness of such an imaging system can either be
described by imaging of an infinitely sharp edge or by imaging of an infinitely narrow
slit. An ideal image of such a slit would be an infinitely small square function for the
luminescence intensity (figure 2.2a (right)). In reality one gets a bell-shaped curve the
half width of which determines the sharpness of the system (figure 2.2b (right)).
2.2.2 Modulation transfer function
One way to assess the image quality is to express it in terms of the resolving power of
the imaging system, i.e. the smallest separation of a pair of linear objects at which the
images do not merge. A very useful tool to describe the image sharpness of an X-ray
imaging system is the concept of the modulation transfer function (MTF). This is based
on the ideas of a Fourier analysis, for detailed information see e.g. [20].
X-ray storage phosphors
22
Starting at the object, at any stage in the imaging process all the available information
can be expressed in terms of spatial frequencies. The idea of spatial frequency can be
understood by considering two ways of describing a simple object consisting of a set of
equally spaced parallel lines. The usual convention would be to say the lines were
equally spaced 0.2 mm apart. Alternatively, one could say that the lines occur with a
frequency in space (spatial frequency) of five per mm. In general, the finer the detail the
greater the intensity of high spatial frequencies in the spatial frequency spectrum. Thus,
fine detail, or high resolution, is associated with high spatial frequencies [21].
The imaging process converts the two-dimensional radiation image g(x, y) into a two-
dimensional visible image b(x, y). For the sake of simplicity the radiation image is taken
here to change only in one dimension. The imaging process can then be described by
gx bx() ()→. (2.6)
A linear imaging system is characterised by the fact that a sinusoidal radiation image
gx A k kx( ) ( ) sin( )=⋅
in (2.7)
Figure 2.2 Imaging of a sharp edge (left) and a narrow slit (right) by an X-ray imag-
ing system based on X-ray storage phosphors.
2.2 Spatial resolution of X-ray storage phosphor image plates
2
3
is converted into a sinusoidal visible image
bx A k kx( ) ( ) sin( )=⋅
out . (2.8)
Hereby, the parameter k, which has the dimension of a reciprocal length, denotes the
spatial frequency. The modulation transfer function (MTF) is defined as
MTF out in
() () ()kAkAk=. (2.9)
The modulation transfer function describes the transfer conditions of an imaging sys-
tem. It defines how the image contrast decreases with higher spatial frequencies. The
highest spatial resolution is determined by the point of intersection between the MTF
and a line which represents a smallest detectable contrast. Another characteristic value
often used is the spatial frequency, where the MTF(k) is decreased to 50% (figure 2.3).
The two MTF curves of figure 2.3 show that the spatial resolution characterises the per-
formance of an imaging system insufficiently. The imaging system represented by
curve 1 has a significantly higher spatial resolution as system 2, but for average spatial
frequencies the modulation transfer and thus the image contrast for objects having these
spatial frequencies is much worse than that of curve 2.
Figure 2.3 Imaging of a sinusoidal radiation image and two arbitrary modulation
transfer functions (MTF).
X-ray storage phosphors
24
2.2.3 Description of imaging processes via Fourier transformation
The mathematical derivation of the modulation transfer function (MTF) assumes a sinu-
soidally varying object which is very difficult to realise practically. Thus, the relation
between the MTF and the imaging of other objects has to be clarified. The imaging pro-
cess of an arbitrary object function g(x) via a system having a modulation transfer func-
tion MTF(k) can be described by the following:
The object function g(x) can be Fourier transformed into its corresponding frequency
spectrum G(k) by
Gk gx x
kx
() ()=⋅
−
−∞
+∞
Hed
i2
π
(2.10)
According to equation (2.9) the spatial frequency spectrum of the image can be obtained
by a multiplication with the modulation transfer function MTF(k).
B(k) = MTF(k) · G(k) (2.11)
Inverse Fourier transformation leads to the image function
bx Bk k k Gk k
kx kx
() () () ()=⋅ = ⋅⋅
−∞
+∞
−∞
+∞
HH
ed MTF ed
ii
22
ππ
. (2.12)
For the Dirac delta function
δ
(x) as object function, that means an infinitely fine detail,
one obtains with equation (2.10) a spatial frequency spectrum of G(k) = 1. Equation
(2.12) simplifies to
bx k k x
kx
() () ()=⋅=
−∞
+∞
HMTF e d LSF
i2
π
. (2.13)
The image function b(x) of the Dirac delta function
δ
(x) is called the line spread func-
tion LSF(x). The integral over the line spread function LSF(x) yields the edge spread
function ESF(x). The relations between LSF(x), ESF(x) and MTF(k) are given by
LSF d
dESF() ()xxx=(2.14)
MTF LSF e d() ()kxx
kx
=⋅
−
−∞
+∞
Hi2
π
(2.15)
2.2 Spatial resolution of X-ray storage phosphor image plates
2
5
Figure 2.4 shows a theoretical example for the transformations described above. In
practice one measures the ESF(x) of an infinitely sharp edge and determines the MTF(k)
via numerical differentiation and subsequent fast Fourier transformation.
2.2.4 Measurement of the spatial resolution and the modulation trans-
fer function
The assessment of image quality is obtained practically by means of special test objects.
The spatial resolution is usually determined with a lead grid, where the spatial fre-
quency (line pairs per mm) changes stepwise from one line pair to the next (figure 2.5a).
Here, a spatial frequency of one (two, four, five) line pair(s) per mm corresponds to a
set of 500 µm (250 µm, 125 µm, 100 µm) fine lines which are spaced 500 µm (250 µm,
125 µm, 100 µm) apart. The highest spatial frequency which is detectable upon imaging
of such a lead grid yields the spatial resolution of the investigated imaging system.
Figure 2.4 Theoretical example for the determination of the MTF(k) of a given
ESF(x).
X-ray storage phosphors
26
The spatial resolution does not characterise an imaging system sufficiently. It is often
useful to determine the modulation transfer function. This requires, however, much
more experimental effort. There are, in principle, two ways to determine the MTF:
One method is based on a quantitative analysis of the image of the lead grid (figure 2.5).
Figure 2.5c shows that the measured contrast modulation depends on the spatial fre-
quency k. The contrast can be described with the luminescence intensities I1 and I2 by
CII II
=− +
()()
12 12
. (2.16)
For the ideal image of the lead grid (figure 2.5b), the intensities I1 and I2 do not change
upon increasing the spatial frequency. Thus, the resulting contrast function Cideal(k) is
constant. For the real image (figure 2.5c), I1 and I2 change upon increasing the spatial
frequency k and the value of the contrast function Creal(k) decreases. The analysis of this
dependence yields the contrast transfer function (CTF) which is defined as
CTF const
real ideal real
() () () ()kCkC kCk
==. (2.17)
Figure 2.5 Determination of the contrast modulation and thus the contrast transfer
function by imaging of a lead grid.
2.3 Read-out process of X-ray storage phosphor image plates
2
7
The CTF (figure 2.6a) can afterwards be transformed into the modulation transfer func-
tion (MTF) (figure 2.6b). Since the MTF assumes a sinusoidal varying object, the
square wave function of the grid (figure 2.5b) is approximated by a series of sinusoidal
functions.
The other technique is the exact analysis of a line spread function LSF(x) and subse-
quent Fourier transformation. Due to experimental reasons one measures often the edge
spread function ESF(x) and gets the LSF(x) via differentiation.
2.3 Read-out process of X-ray storage phosphor image
plates
During the read-out process the X-ray storage phosphor image plate is optically stimu-
lated pixel by pixel and line by line by means of a laser beam, e.g. a HeNe laser
(632.8 nm) for the commercially used BaFBr:Eu2+ phosphor screen. The photostimu-
lated luminescence light is collected globally with a waveguide and detected by a pho-
tomultiplier. The stimulation light is cut off by means of an appropriate optical filter
combination which is placed between the waveguide and the photomultiplier (not indi-
cated in figure 2.7). The detected luminescence intensity during stimulation of a pixel is
Figure 2.6 (a) Measured data of the contrast transfer function (CTF) and (b) modula-
tion transfer function (MTF), derived from the data of (a), for a commer-
cially-used image plate (AGFA MD 10) [22].
X-ray storage phosphors
28
a measure for the absorbed X-ray dose in that pixel. The analogue signal of the photo-
multiplier is converted by an analogue / digital (A / D) converter to a digital signal. The
scanning procedure is carried out by moving the laser with a rotating mirror and / or the
image plate with the roller plate (figure 2.7).
The spatial resolution of X-ray storage phosphor screens is still inferior to that of con-
ventional X-ray films. One of the reasons for this is certainly the light scattering of the
scanning laser beam which is used for the read-out. The structural reasons for this are
explained in figure 2.8: The image plate consists of fine phosphor grains imbedded in an
organic binder. The phosphor / binder layer is deposited on an organic substrate and
protected by a thin foil. During the read-out the stimulating light of the scanning laser
beam is scattered within the phosphor / binder layer due to the different refractive indi-
ces of the two components. The scattered stimulating light can then excite the electron
trap centres in the X-irradiated areas and cause photostimulated luminescence (PSL)
although the scanning laser beam is not in the right position yet. The luminescence light
is detected globally on the side of the stimulation. The sharp edges of the X-irradiated
Figure 2.7 Setup for the read-out procedure of X-ray storage phosphor image plates
[22].
2.3 Read-out process of X-ray storage phosphor image plates
2
9
areas are spread and the contrast modulation is reduced. For a contrast modulation be-
low the smallest detectable contrast the two X-irradiated areas in figure 2.8 cannot be
resolved any more. Here, the detection of the luminescence light is done globally on the
stimulation side of the phosphor screen and not focussed on the laser beam position in
the layer. The spatial resolution is thus determined by the scattering region, i.e. the ex-
citation volume of the stimulating light.
Figure 2.8 Scattering of the scanning laser beam during the read-out process of a stor-
age phosphor image plate.
X-ray storage phosphors
30
31
Chapter 3
The X-ray storage phosphor
BaFBr:Eu2+
The so far best and world-wide commercially used X-ray storage phosphor is BaFBr
doped with Eu2+ as activator [1]. Upon X-irradiation, room temperature (RT) stable
electron and hole trap centres are generated. The electron trap centres are photostimula-
ble and the electron recombines upon photostimulation with the hole trap centre under
light emission. Although the principle of information storage and read-out mechanism is
simple, it is not clearly known how the recombination energy is transferred to the acti-
vator Eu2+ which emits at 3.18 eV (390 nm). The PSL decay time is 750 ns [1, 23]. In
many years of thorough research and optimisation BaFBr:Eu2+ has already reached a
very high level of performance. However, there is still space for some improvements
such as e.g. a better conversion to PSL-active centres [24], a higher stability of the PSL
active centres, and a better erasability of the generated defects after the read-out proc-
ess. The spatial resolution can probably not be improved. For recent reviews see e.g. [2,
3, 4, 5, 25].
3.1 Stoichiometric BaFBr
3.1.1 Generation of electron trap centres
There is general agreement that X-irradiation generates F centres as electron trap cen-
tres. In BaFBr, which has the matlockite structure [26, 27], two F centres are possible,
F(Br−) and F(F−) centres, where electrons are trapped at Br− vacancies or F− vacancies,
The X-ray storage phosphor BaFBr:Eu2+
32
respectively. Their generation mechanism, however, is controversial. In order to form
an F centre after creating electron-hole pairs by X-irradiation, one either needs to have a
halide vacancy present in the crystal or one must generate it during the radiation dam-
age process.
In [28] it was assumed that Br− vacancies are already present in the material. However,
no experimental evidence was given for this assumption. In [29] it was shown that
photostimulable centres can be created by using vacuum UV light instead of X-rays and
concluded that the decay of self-trapped excitons into F and H centres (Br2
−centres on a
halide lattice site, see e.g. [30]), provide the necessary mechanism to create F centres.
H centres were not detected, though, in spite of an intense search for them at 1.5 K by
EPR and optically detected EPR using the magneto-optical method (MCDA-EPR). Had
there been H centres, they would have been detectable using MCDA-EPR. It was shown
in [31] that H centres created in KBr can be measured in this way.
When measuring the photostimulation as a function of photon energy, two peaks are
usually observed: One at 2.15 eV and one around 2.65 eV. The two peaks are clearly
resolved when using single crystals and polarised light (e.g. E ⊥ c) [32, 33]. The spectral
shape of the excitation spectrum of the photostimulated luminescence (figure 3.1b)
agrees very well with the optical absorption band of F(Br−) centres (peak at 2.15 eV for
E ⊥ c) and F(F−) centres (peak at 2.65 eV for E ⊥ c) (figure 3.1a). The absorption band
for E ⊥ c was clearly identified for each F centre by magneto-optical techniques [34].
Although it was argued that only F(Br−) centres are photostimulable [24], it could be
clearly shown that both F centres contribute to the PSL [32, 33]. Yet it was interesting
to observe that F(F−) centres are only generated by X-irradiation above 200 K. Their
formation apparently requires a thermally activated process. When observing the gen-
eration of both centres as a function of the X-ray dose, it turns out, however, that F(F−)
centres are not generated as a secondary product of F(Br−) centres. F(Br−) centres are
generated first and then saturate, whilst F(F−) centres continued to be generated, but not
at the expense of the F(Br−) centres. The process has not yet been understood [15].
When no F-H centre mechanism is responsible for the generation of F(Br−) centres,
which are found even at the lowest temperature (1.5 K) upon X-irradiation together with
VK centres (hole shared between two adjacent Br− [32]) as “anti-centres” (see fig-
ure 3.2a), it must indeed be assumed that BaFBr does contain Br− vacancies (VBr−).
3.1 Stoichiometric BaFBr
3
3
3.1.2 Generation of hole trap centres
It was long overlooked that BaFBr produced by firing stoichiometric mixtures of BaF2
and BaBr2 or by growing single crystals from the melt of such mixtures with the
Bridgman method are all contaminated with oxygen, irrespective of the manufacturer.
All attempts to eliminate oxygen completely have failed. Very careful avoidance of
oxygen could only reduce the oxygen contamination [35, 36]. Oxygen can be incorpo-
Figure 3.1 (a) Optical absorption spectrum of F(Br−) and F(F−) centres in BaFBr and
(b) PSL excitation spectrum of Eu2+ in BaFBr. Both spectra were recorded
at 10 K for E ⊥ c after X-irradiation at RT [32].
The X-ray storage phosphor BaFBr:Eu2+
34
rated in many ways: It was found as O2− in the fluorine (OF
2−) [37] as well as in the
bromine sublattice (OBr
2−) [38, 39]. There was also a molecular oxygen centre ((OF-
OBr)3−) observed where one oxygen is placed on a fluorine site and the other one on a
neighbouring bromine site [39].
For charge compensation an anion vacancy is needed. It was shown by the study of the
generation of F centres, VK centres and O− centres at low temperature using magneto-
optical (MCDA) and MCDA-EPR methods that these vacancies are indeed Br− vacan-
cies [15, 35, 36]. Upon X-irradiation at temperatures below 120 K V Br
K()
2
− centres
and F(Br−) centres are formed, the latter being near to the OF
2− centres. Above 120 K
the V Br
K()
2
− centres become mobile, react with the OF
2− centres and form OF
− hole
trap centres. Above 200 K the F(Br−) centres can diffuse away and become isolated [37,
40]. The microscopic structure of the OF
− defect (figure 3.2b) was established by EPR
and ENDOR. Figure 3.3 shows the EPR spectrum of 17 F
O− centres in BaFBr. The mag-
netic oxygen isotope 17O has a nuclear spin of I=5 2 and hence shows a hyperfine
splitting into 6 lines. The isotope substitution of 17O for 16O proved that the defect cen-
tre of figure 3.3, always produced upon X-irradiation of BaFBr, is due to an oxygen
contamination [37].
Figure 3.2 Model of (a) the VBr()
−
and (b) the O−
centre in BaFBr.
3.1 Stoichiometric BaFBr
3
5
It can be assumed that the BaFBr used by [29] also contained oxygen and thus the UV
production of F(Br−) centres becomes understandable by the process
VO FBrVBrO
Br F
2UV, X-ray KF
2
−+ →++
−−−−
() ()
2(3.1)
>−−
→+
120 K F
FBr O() .
Thus, the generation of F(Br−) centres in oxygen-contaminated, stoichiometric BaFBr is
understood, whereas the mechanism leading to F(F−) centres remains still unclear. The
microscopic structure of the two F centres was established by detailed ENDOR investi-
gations [41].
3.1.3 PSL active hole trap centres
The nature of the hole trap centre taking part in the PSL process is still unclear and
controversially discussed. In [28, 42] it was claimed that upon electron and hole crea-
tion by X-rays holes are trapped at Eu2+ and form Eu3+. Upon photostimulation of F
centres, the F electrons move through the conduction band and recombine with Eu3+,
exciting Eu2+ to its excited state, from which the 390 nm luminescence occurs. In this
model the PSL process requires thermal activation, since the excited F centres have re-
Figure 3.3 EPR spectra of the OF
−
centre in 17O-doped BaFBr after X-irradiation at
RT, recorded for B || c at T = 5 K applying a microwave frequency of
9.42 GHz. The bars indicate the 17O hyperfine splittings [37].
The X-ray storage phosphor BaFBr:Eu2+
36
laxed excited states below the conduction band. The thermal activation energy was de-
termined to be 37 meV for F(Br−) and 1.3 meV for F(F−) [43].
This simple pair model for the PSL mechanism was questioned by several authors. The
EPR signal of Eu2+ did not change upon prolonged X-irradiation [32], nor could a
change be observed in the Eu2+ luminescence [44]. Furthermore, the observation of an
almost temperature independent PSL effect by stimulating F(Br−) centres from 4.2 K to
RT [23] contradicts the model that electrons move through the conduction band upon
photostimulation of F centres.
However, it has been proposed on the basis of several different experimental findings
that the recombination between electrons and holes takes place via tunnelling and that a
kind of “aggregate” between F centres, hole trap centres and the activator Eu2+ must be
formed during X-irradiation (called triple aggregate centres [32]). It was found that the
decay of the photostimulated luminescence under continued stimulation is temperature
independent [45] and that the increase in the PSL intensity is proportional to the X-ray
dose [23]. This latter result implies that retrapping of electrons after photostimulation
does not occur, which makes electron-hole recombination via tunnelling more likely
than via the conduction band. Tunnelling, however, requires a spatial correlation be-
tween the F centre and the activator. Direct evidence for a spatial correlation between F
centres, OF
− centres and Eu2+ was given with cross-relaxation spectroscopy using mag-
neto-optical techniques (figure 3.4) [46].
Figure 3.4 Schematic model for the spatial correlation between F(Br−), VBr
K()
2
−,
O2− and Eu2+ after low temperature X-irradiation [15].
3.1 Stoichiometric BaFBr
3
7
It could be shown by the cross-relaxation spectroscopy within the MCDA-detected EPR
spectroscopy (see section 1.1.3) that there is a spatial correlation between F centres and
Eu2+ as well as between OF
− and Eu2+ after X-irradiation at RT. When X-irradiating at
low temperature, VK centres also show this correlation with the other centres [40, 46].
This leads to the speculation that an exciton decay occurs at the Eu2+ activator forming
electron and hole trap centres and that there is a spatial correlation between the OF
2−-
anion vacancy pairs and the Eu2+ activators generated during the production of the ma-
terial (figure 3.5).
Indirect evidence for the formation of “triple aggregate” centres was also found in the
so-called replenishment effect. When the phosphor is stimulated at 4.2 K and the PSL
exhausted, it can be replenished by annealing to temperatures above 200 K [32]. It
seems that after exhaustion of the aggregates formed first, new aggregates can be
formed by thermally activated motion of either hole trap or electron trap centres, or
both. The size of the effect depends on the amount of oxygen contamination. The mi-
gration of F(Br−) centres was recently investigated at RT by [47]. An investigation of
the crystal size dependence on the F(Br−) centre stability yielded that in small crystal-
lites the F(Br−) centres seem to migrate to be stabilised near the surface of the crystal-
Figure 3.5 EPR spectrum detected in the MCDA band of the VBr
K()
2
− centre in
BaFBr after X-irradiation at 4.2 K, recorded for B || c at T = 1.5 K applying
a microwave frequency of 24 GHz. The spectrum shows cross-relaxations
to the F(Br−) centre and to Eu2+ [46].
The X-ray storage phosphor BaFBr:Eu2+
38
lites, whereas in larger crystals they only migrate to recombine with the hole trap cen-
tres.
The role of OF
2− centres is certainly that it captures holes and forms OF
− centres. The
question is, though, whether it is also the hole trap centre active in the PSL process.
When bleaching into the F centre band, the number of F centres can be drastically de-
creased to practically zero, while that of the OF
− centres is hardly affected: The decrease
is at most by 20%-30% [15]. Hence, it is not very likely that the OF
− centres are those
active hole trap centres in the triple aggregates described above. On the other hand, it
was found that the variation of the oxygen content influences the PSL intensity. A low
oxygen content results in a low PSL intensity. Also, the stimulation energy needed for
the read-out process is higher for oxygen-poor BaFBr. Thus, it seems that oxygen is
involved in two ways in the PSL mechanism: It provides vacancies for the F(Br−) gen-
eration and it somehow seems to influence favourably the photostimulation of the triple
aggregate centres in that less stimulation energy is needed for read-out and a high PSL
intensity results after short stimulation.
3.2 Non-stoichiometric BaFBr
In the previous section it was shown that BaFBr powders which are normally formed by
firing intimate mixtures of BaF2 and BaBr2 are always contaminated with oxygen. The
role of the oxygen contamination is both beneficial and detrimental, and it is difficult to
control its concentration such as to optimise the performance of these storage phos-
phors. There is a different method to produce BaFBr powders [22]. An intimate mixture
of BaF2 and NH4Br in the ratio of 1:1 is fired instead of a mixture of BaF2 and BaBr2.
When doped with Eu2+ the resulting material was found to be an excellent storage
phosphor, although this material showed some properties significantly different from
that produced by firing a mixture of BaF2 and BaBr2.
Chemical analysis of the material yielded that it is non-stoichiometric with a fluorine
excess of about 10%±2% [48]. The non-stoichiometric BaF1.1Br0.9 still has the mat-
lockite structure as does the stoichiometric BaFBr including that no change of the lattice
parameters could be detected by detailed X-ray diffraction (XRD) studies [48]. It was
also shown by XRD that BaF1.1Br0.9 is a single phase compound with at most 1% of
other phases present. Attempts to vary the non-stoichiometry by firing different mix-
tures of BaF2 and NH4Br failed to generate a single phase material.
3.2 Non-stoichiometric BaFBr
3
9
No oxygen luminescence [49, 50] could be excited in BaF1.1Br0.9 [51, 52]. Thus, this
non-stoichiometric material is considered to be oxygen-free. X-irradiation at RT gener-
ates F(Br−) and F(F−) centres. The presence of F(Br−) centres could be measured by
powder-EPR, while the concentration of F(F−) centres was not sufficient to be seen in
the powder-EPR spectrum. The presence of F centres, however, can also be tested by
exciting their infrared (IR) luminescence. Surprisingly, the IR luminescence peaking at
1.14 eV (1088 nm) of F(F−) could be detected, while that of F(Br−) centres at 0.92 eV
(1348 nm) could not be detected [51]. This latter result points to a possible nearby pres-
ence of another defect which may cause the excited state of the F(Br−) centre to decay
non-radiatively.
The question arises by what mechanism the F(Br−) centres are generated. Since there is
no OF
2− in this material, either Br− vacancies must be present or a process such as the
exciton decay resulting in F-H-centre pairs as is known from the alkali halides, may be
operative [30]. It could be argued that the F− excess in BaF1.1Br0.9 could cause the in-
corporation of Br− vacancies. If 18% Br− vacancies and 9% Ba2+ vacancies were pres-
ent (the latter for charge compensation, the concentrations of Br− and Ba2+ vacancies
would explain the formula BaF1.1Br0.9), then there should be a significant change of
more than 10% in the density which is caused by a decrease of the molecular mass
while assuming no change of volume. This was, however, not measured within experi-
mental error of ±2% [22].
Consequently, speculations arose that perhaps the F− excess was incorporated as F− on
Br− sites, i.e. as F− antisites. The density change would then be as small as found within
the experimental error of ±2%. The presence of F− antisites could be established by
NMR experiments. In order to obtain NMR lines narrow enough to resolve a possibly
small chemical shift between 19F on a regular lattice site and an F− on a Br− lattice site,
magic angle spinning nuclear magnetic resonance (MAS-NMR) spectroscopy [13] was
applied. It was shown (see section 3.2.2) that indeed F− antisites are present in non-
stoichiometric BaF1.1Br0.9 [52].
The question still remained whether the F− antisites provide a mechanism whereby
F(Br−) centres can be created or whether, in addition to the F− antisites, Br− vacancies
are present in the material. This question is answered in the following section on EPR
experiments after X-irradiation at low temperature and at RT. Both assumptions are
correct: There are Br− vacancies in the material, and there is a mechanism generating
F(Br−) centres which originates in the F− antisites.
The X-ray storage phosphor BaFBr:Eu2+
40
3.2.1 Photostimulated luminescence
The PSL excitation spectra of stoichiometric BaFBr:Eu and of non-stoichiometric
BaF1.1Br0.9:Eu powder yielded that the peak of the stoichiometric sample is at 540 nm
(figure 3.6a) while that of the non-stoichiometric one is at 580 nm, i.e. at lower photon
energy, and there is an additional shoulder at 510 nm (figure 3.6b). The intensity ratio
between the 580 nm peak and the high-energy shoulder at 510 nm is approximately 2:1.
In both materials the luminescence is due to the stimulation of F(Br−) and F(F−) centres.
The absorption peak of F(Br−) centres is at 580 nm for the electrical light vector per-
pendicular to the crystal c-axis, while it is at 510 nm for E || c. For F(F−) centres the ab-
sorption peaks are at 470 nm for E ⊥ c and at 520 nm for E || c [34]. It is seen qualita-
tively in figure 3.6, that fluorine excess leads to an enhanced concentration of F(Br−)
centres [51]. The absorption peaks of F(Br−) centres agree well with the peak and
shoulder of figure 3.6b. The intensity ratio of 2:1 is caused by the statistical distribution
of the parallel and perpendicular crystallite orientations with respect to the electrical
light vector. Apparently in the non-stoichiometric BaFBr very few F(F−) centres are
generated in comparison to F(Br−) centres.
Figure 3.6 PSL excitation spectra measured at RT of (a) stoichiometric BaFBr:Eu2+
powder and (b) non-stoichiometric BaF1.1Br0.9:Eu2+ powder after X-
irradiation at RT [52].
3.2 Non-stoichiometric BaFBr
4
1
3.2.2 Identification of fluorine antisites with MAS-NMR
Figure 3.7 shows the MAS-NMR spectra of 19F of a pulverised stoichiometric single
crystal of BaFBr and of non-stoichiometric BaF1.1Br0.9 powder. The peaks due to the
lattice 19F nuclei of both spectra coincide within experimental error at 150.9 ppm. A
new line appears in BaF1.1Br0.9 powder at lower frequency, at 145.3 ppm, which also
shows up in the spinning side bands (marked with asterisks in figure 3.7). The intensity
of the new line as measured by the area in comparison to that of the lattice 19F is 7.3%.
The addition of the contributions of the spinning side bands yields in total 8.6% inten-
sity. Thus, the new line shifted by almost 6 ppm to lower frequencies is due to approxi-
mately 9% of 19F nuclei with a different site compared to the lattice nuclei. This is close
to the 10% excess of F in the lattice determined by chemical analysis [48]. Since for
Figure 3.7 19F MAS-NMR spectra of (a) a crushed stoichiometric BaFBr single crys-
tal and (b) non-stoichiometric BaF1.1Br0.9 powder. The spinning side
bands are marked with asterisks [52].
The X-ray storage phosphor BaFBr:Eu2+
42
electrostatic reasons it is very unlikely that the new site is an interstitial site, the new
line was assigned to F− on Br− vacant sites, i.e. to F− antisites [52].
Figure 3.8 shows the high and low frequency edge singularities of the 137Ba static pow-
der spectrum having second order quadrupole interaction [53]. The edge singularities of
Ba as well as those of Br (which are not shown here) are identical for both the stoichi-
ometric and the non-stoichiometric BaFBr. Thus, the F− excess does not change the
crystal structure, i.e. the geometry must be practically identical for both crystals.
The MAS-NMR spectra confirm an earlier suggestion that the F− to Br− ratio of about
1.1 to 0.9 is achieved by 10% enhanced fluorine incorporation and a simultaneous 10%
bromine reduction [51]. In this case a charge compensation on the cationic sublattice is
not necessary. The change in density is approximately 2% which could not be observed
experimentally within experimental error of ±2% [22].
Figure 3.8 Edge singularities for 137Ba of (a) a crushed stoichiometric BaFBr single
crystal and (b) non-stoichiometric BaF Br powder [52].
3.2 Non-stoichiometric BaFBr
4
3
The two chemical shifts found for 19F for the regular F lattice site and the antisite 19F
are not very different: 150.9 ppm for the regular sites, 145.3 ppm for the antisites. The
difference of 5.6 ppm is smaller than that found for 19F in the divalent fluorides CaF2,
SrF2 and BaF2 (see e.g. [54, 55]) which vary between 58 ppm and 152 ppm, respec-
tively. It is interesting to note that the Ba2+-F− distance in BaF2 (2.68 Å) is almost iden-
tical to that in BaFBr (2.66 Å). In [54] it was argued that the chemical shift is deter-
mined by the metal-F− distance. The 150.9 ppm found for the regular 19F lattice nuclei
tie in well with this rule. Not so the 19F antisites having a distance of 3.36 Å to the near-
est Ba2+ neighbour (along the c-axis) and 3.42 Å to the four next nearest Ba2+ neigh-
bours, respectively. Perhaps the larger screening is the result of a larger site for F− and
the more expanded electron core.
3.2.3 Identification of electron and hole trap centres with EPR
After X-irradiation at RT the EPR spectrum of non-stoichiometric BaF1.1Br0.9 powder
shows the lines of F(Br−) centres and a powder-EPR line the g factor of which indicates
that it is caused by a hole trap centre (figure 3.9b). The spectrum can be simulated well
by assuming an axial centre with the g values of g⊥ = 2.02 and g|| = 2.002 [52]. In fig-
ure 3.9a the powder-EPR spectrum of stoichiometric BaFBr is shown which contains
oxygen and where the resonances of OF
− centres as well as of F(Br−) centres are seen.
The positive shift of the g values of the new centre indicates that it is indeed a hole trap
centre.
Its nature as a hole trap centre is further supported by the observation that its intensity
increases proportional to the X-ray dose as does that of the F(Br−) centres. The simulta-
neous growth of the hole trap centre and that of the F(Br−) centre suggests that the hole
trap centre is the “anticentre” of the F(Br−) centre, i.e. electron-hole separation results in
the F(Br−)-hole trap centre pairs. When exciting the F(Br−) band with light, a simulta-
neous destruction of F(Br−) and the hole trap centres is observed. All F centres disap-
pear, while 30% of the hole trap centres remain. These observations support the view
that F centres and hole trap centres are generated as pairs. The fact that not all hole trap
centres are recombined with the electrons of the F centres can be explained by the for-
mation of RT-stable F aggregate centres which occurs simultaneously when exciting in
the F band at RT.
The X-ray storage phosphor BaFBr:Eu2+
44
After X-irradiation of stoichiometric BaFBr powder at temperatures below 77 K intense
resonances lines of the V (Br
K2
−) centres and a weak EPR signal of the F(Br−) centres
were observed [52]. After annealing up to 300 K, the V (Br
K2
−) centre lines disappear
whereas the OF
− centre line appears. The V (Br
K2
−) centre decays at about 120 K [32].
The moving hole is trapped by an OF
2− impurity to form the OF
− centre [56].
After X-irradiation at temperatures below 77 K the EPR spectrum of the non-stoichio-
metric BaFBr powder showed again intense resonance lines of the V (Br
K2
−) centres and
a weak signal of the F(Br−) centres as well as the powder EPR line of the new hole trap
centre described above. When using a low X-ray dose a rather strong signal of the VK
centres appears relative to that of the new hole trap centre. Upon increasing the X-ray
dose the signal of the new hole trap centre increases rapidly while that of the VK centre
grows at a slower rate. The experiments were not carried as far as reaching a saturation
of the VK centre signal, but it appears that such a saturation can be reached while no
sign of an incipient saturation was observed for the new hole trap centre [52].
After annealing up to 300 K the V (Br
K2
−) centre EPR lines have disappeared. This
thermal decay of the V (Br
K2
−) centre did not cause a change in the EPR line intensity
Figure 3.9 Powder EPR spectra of (a) stoichiometric and (b) non-stoichiometric
BaFBr powder after X-irradiation at RT, recorded at 10 K applying a mi-
crowave frequency of 9.335 GHz [52].
3.2 Non-stoichiometric BaFBr
4
5
of the new hole trap centre, but part of the F centre signal was destroyed. Unfortunately,
the S N ratio of the F centre was not good enough, in order to check whether as many
F centres were destroyed as VK centres have disappeared.
3.2.4 Generation of electron and hole trap centres
The generation of F centres in non-stoichiometric BaF1.1Br0.9 powders seems to occur
via two mechanisms. At 77 K two kinds of hole trap centres were observed: V Br
K()
2
−
centres and the new hole centres, which are considered to be H centres in the F− sublat-
tice (see below). The V Br
K()
2
− centre generation at 77 K seems to indicate that in spite
of no oxygen contamination the non-stoichiometric BaFBr contains also Br− vacancies.
Upon annealing to RT the V Br
K()
2
− centres and part of the F(Br−) centres disappear:
Mobile V Br
K()
2
− centres recombine with F(Br−) centres. However, some F(Br−) centres
remain and so does the EPR signal of the new hole trap centre, which is not changed at
all upon the availability of mobile V Br
K()
2
− centres.
Since F(Br−) centres and the new hole trap centre are electron and hole trap centres cre-
ated simultaneously and proportional to each other, it is suggested that F− antisites
(F
Br
−) are the origin of the F(Br−) centres and the new hole trap centre according to the
reaction
FBr
− X-ray
→ F(Br−) + F F2,
−(3.2)
i.e. an F(Br−) and an F2
− molecular centre in the F− sublattice (F F2,
−) are created in an F-
H-process where the electron trap centre (F centre) is formed in the Br− sublattice and
the H centre is formed in the F− sublattice. To support this suggestion, the powder EPR
spectra of the new H centre was analysed in more detail. It is known that F2
− molecular
centres on F− sites can be produced in alkali earth fluorides by X-irradiation below 77 K
[57]. There, the two fluorine nuclei of the H-type centre are not equivalent, i.e. one fluo-
rine nucleus is placed on an interstitial site (“fluorine interstitial”) whereas the second
one is on a regular lattice site (“fluorine substitutional”). The hyperfine (hf) interactions
of the fluorine interstitial and the fluorine substitutional are different.
The X-ray storage phosphor BaFBr:Eu2+
46
Figure 3.10 shows two calculated powder EPR spectra of an F2
− centre with two
equivalent fluorine nuclei. Hereby typical fluorine hf interaction values [57] of A⊥ =
200 MHz and A|| = 2000 MHz or A⊥ = 20 MHz and A|| = 2000 MHz, respectively, were
assumed. In both cases the powder EPR lines indicating the hf interaction for an orien-
tation parallel to the molecular axis are very weak. In figure 3.10 the lines are scaled up
by a factor of 50. The powder EPR lines indicating the hf interaction perpendicular to
the molecular axis are clearly visible in figure 3.10a. Assuming a small value for A⊥ (=
20 MHz), the hf lines are superimposed by the central lines (figure 3.10b). Thus, by
comparison with figure 3.9b, it was proposed that the hole trap centre has A⊥ ≤ 20 MHz
[52].
Figure 3.11 shows a model of the F2
− centre in BaFBr with two equivalent fluorine nu-
clei. The molecular axis of the F2
− centre with the two equivalent fluorine nuclei is par-
allel to the a-axis (b-axis) of the crystal as calculated in [58]. Since a powder EPR
spectrum is a summation over all possible orientations of the magnetic field vector, the
information about the orientation of the interaction tensors to the crystal axes is lost.
Figure 3.10 Calculated powder EPR spectra of a F2
− centre with two equivalent fluo-
rine nuclei. The g tensors with g⊥ = 2.02 and g|| = 2.002 are axial as well as
the 19F hyperfine tensors. The bars indicate the corresponding hyperfine
splittings. The microwave frequency is 9.335 GHz [52].
3.2 Non-stoichiometric BaFBr
4
7
Therefore, the measured powder EPR spectrum of the new hole trap centre does not
allow to decide on the position of the molecular axis.
A qualitative view of the F-H process would be that the valence electron of the F− an-
tisite is excited upon an exciton decay at the FBr
− antisite into a diffuse excited state and
that the F0 becomes mobile and moves to the F− sublattice, where it associates itself
with a lattice F− to form the H (F
2
−) centre on a fluorine site (figure 3.12). It was calcu-
lated by Baetzold [58] that such an H centre is stable and also that the formation of FBr
−
antisites is exothermic in BaFBr [59]. It was not possible to say whether the F(Br−) and
H centres are nearest neighbours or further apart. Judging from the results obtained in
the alkali halides, they will be further apart, otherwise they would probably recombine
[60].
Figure 3.11 Model of an H-type F2
− centre in BaFBr with two equivalent fluorine nu-
clei after [58].
The X-ray storage phosphor BaFBr:Eu2+
48
To be stable at RT it is important for the hole trap centre to avoid recombination with
the F centre. It seems reasonable to argue that the hole can move away more easily from
the F centre in the F− sublattice than in the Br− sublattice. This is because the F−-F−
distance is much shorter (3.18 Å) than the Br−-Br− distance, either in plane (4.50 Å) or
out of plane (3.72 Å). Also there are linear [110] or [100] chains of equivalent F− ions,
whereas the hole motion within the Br− double layer would need to proceed in a zigzag
motion to separate from the F centre. The former pathway resembles more the situation
Figure 3.12 Schematic presentation for the F-H centre generation process in non-
stoichiometric BaFBr [5].
3.3 Red-shift of the PSL excitation upon Ca2+ or Sr2+ doping
4
9
in the alkali halides, where the hole motion is known to proceed along the [110] halogen
chain.
The experiments have shown that the VK centre production seems to saturate upon in-
creasing the X-ray dose, while that of the H centre production showed no sign of satu-
ration [52]. The maximum number of VK centres depends on the number of Br− vacan-
cies present, which cannot be excessively large, as otherwise a significant change of the
density would have been observed. Although it was not possible to determine the num-
ber of VK centres quantitatively from the powder spectra, it will be of the order of
1016 cm−3 and that should be the order of magnitude of the Br− vacancies present. On
the other hand, F− antisite defects are “abundant” (10%) in comparison. Thus, no satu-
ration in H centre production is expected in line with this model. The decay of the
VBr
K()
2
− centres did not influence the EPR line intensity of the new hole trap centre.
Had the moving holes of the decaying V Br
K()
2
− centres been trapped by F− interstitials,
the EPR signal of H (F
2
−) hole trap centres would have been enhanced. This was not the
case. Therefore, it was suggested that the F(Br−) centre production in non-stoichiometric
BaFBr and the simultaneous generation of the F2
− hole trap centre is caused by F− an-
tisites and not by F− interstitials.
3.3 Red-shift of the PSL excitation upon Ca2+ or Sr2+ doping
For practical use, apart from a high sensitivity, i.e. a high conversion efficiency of X-
rays into photostimulable defects, it is desirable that the phosphors can be stimulated
with low laser light intensity, particularly in the infrared spectral region. It was shown
that the PSL excitation of BaFBr doped with Eu2+ can be made sensitive to stimulation
further into the infrared by additional Ca2+ or Sr2+ doping [51, 61].
The X-ray storage phosphor BaFBr:Eu2+
50
After X-irradiation at RT the MCDA of BaFBr, recorded for B || c, shows two deriva-
tive-like structured bands due to the two possible F centres (F(Br−) and F(F−) centre)
(figure 3.13a) [34]. The centres of these bands, where the MCDA changes sign, are at
2.15 eV and 2.65 eV, respectively. In Ca2+ doped BaFBr these two F centres are also
observed after X-irradiation at RT (figure 3.13b). The shape and spectral position of the
MCDA band of the F(F−) centre remain the same as in undoped BaFBr, whereas in the
case of the F(Br−) centre the spectrum shows a superposition of at least two bands. The
minimum of the perturbed F(Br−) centre band keeps its position while the maximum
shifts to lower energies. The perturbed F(Br−) centre is supposed to be a F centre where
a Ba2+ ion in a nearest-neighbour position is replaced by a Ca2+ ion, in analogy to what
was observed for F centres in alkali halides [62]. The MCDA bands observed in Ca2+
doped BaFBr are thus a superposition of the MCDA band of the unperturbed F(Br−)
centre and of that of the perturbed FA(Br−, Ca2+) centre.
Figure 3.13 MCDA spectra of F(Br−), F(F−), and FA(Br−, Ca2+) centres in (a) undoped
BaFBr and (b) Ba0.98Ca0.02FBr after X-irradiation at RT, recorded for B || c
at 1.5 K [61].
3.3 Red-shift of the PSL excitation upon Ca2+ or Sr2+ doping
5
1
MCDA-detected ENDOR measurements performed on the perturbed F(Br−) centre band
yielded that there are resonances due to the unperturbed F(Br−) centre as well as those
of the FA(Br−, Ca2+) centre (figure 3.14). The two ENDOR lines marked with 19F at
about 52 MHz are due to the nearest-neighbouring fluorine nuclei of the unperturbed
F(Br−) centre in BaFBr [34]. The two peaks marked with 19F* at about 58 MHz cannot
be explained with the shf interactions of neighbouring nuclei of the unperturbed F(Br−)
centre. Thus, they must be due to the FA(Br−, Ca2+) centre. A field shift experiment
where the shift in the frequency of the ENDOR lines due to the nuclear Zeeman inter-
action was measured with respect to the variation of the external magnetic field, showed
that the lines at 58 MHz originate from fluorine nuclei (for details of the field shift
method see e.g. [10]). In analogy with the ENDOR lines of the unperturbed F(Br−) cen-
tre these signals were assigned to the nearest fluorine neighbours of the FA(Br−, Ca2+)
centre. This assignment was confirmed by the ENDOR analysis of the conventional
ENDOR measurements. The splitting of the 19F and the 19F* lines for the magnetic field
parallel to the c-axis is caused by a slight misorientation of the crystal. The other four
ENDOR lines are caused by the nearest-neighbouring bromine nuclei of the FA(Br−,
Figure 3.14 MCDA-detected ENDOR spectra of F(Br−) and FA(Br−, Ca2+) centres in
Ca2+ doped BaFBr after X-irradiation at RT, B || c, B = 871 mT, recorded
at a photon energy of 1.984 eV and at 1.5 K applying a microwave fre-
quency of 24 GHz. The lines marked by asterisks are from FA(Br−, Ca2+)
centres [61].
The X-ray storage phosphor BaFBr:Eu2+
52
Ca2+) centre. For the magnetic field exactly parallel to the c-axis the 19F ENDOR lines
of the F(Br−) centre coincide, so do the 19F* lines for the FA(Br−, Ca2+) centre which
shows that the four ions of the first fluorine shell are magnetically equivalent. There-
fore, the perturbing Ca2+ ion must be located on the fourfold axis of the FA(Br−, Ca2+)
centre (figure 3.15). A more detailed investigation of the symmetry of the Ca2+ per-
turbed F(Br−) centre was made with conventional ENDOR measurements [61].
A rough estimate of the relative concentrations of the FA(Br−, Ca2+) and the F(Br−)
centre created with X-irradiation at RT can be made, assuming that the ENDOR effect
of the nearest fluorine lines is not much different for both defects. The ratio of the
FA(Br−, Ca2+) centre to F(Br−) centre is approximately 3:1, i.e. 75% of F centres are FA
centres. It seems that the exciton decay mechanism producing F centres upon X-
irradiation occurs preferentially at the Ca2+ perturbed sites. Otherwise it would not have
been possible to observe such a large ratio between FA and F centres which deviates
from pure statistical probability according to the doping by almost two orders of mag-
nitude.
The observed red-shift of the FA centre absorption band can be explained by a shallower
Madelung potential well binding the electron of the F centre. The smaller Ca2+ ion
(ionic radius 0.99 Å) replacing the larger Ba2+ ion (ionic radius 1.34 Å) along the c-axis
Figure 3.15 Defect model of the FA(Br−, Ca2+) centre in Ca2+ doped BaFBr. One Ca2+
replaces either the upper or the lower Ba2+ along the c-axis [61].
3.4 Surrounding of the activator Eu2+
5
3
may be relaxed away from the Br− vacancy leaving a larger space for the F centre elec-
tron. According to the Mollwo-Ivey relation the peak energy of the F centre absorption
band is proportional to 12
d where d is the lattice constant [62]. The lattice constant
will be increased along the c-axis resulting in a red-shift of the optical absorption band.
However, it is not possible to decide which of the two Ba2+ ions along the c-axis is re-
placed by Ca2+. It seems highly improbable that both are substituted by Ca2+.
3.4 Surrounding of the activator Eu2+
It is reasonable to assume that Eu2+ substitutes for Ba2+ in BaFBr. An EPR investigation
[63] yielded an axial centre with the z-axes of the fine structure and hf tensors of Eu2+
along the c-axis as expected for such a site [63]. What could not be inferred from EPR
was whether or not there are lattice relaxations about the Eu2+. The ionic radius of the
replaced Ba2+ is 0.134 nm, that of Eu2+ 0.109 nm, i.e. about 20% smaller. Therefore, a
substantial lattice relaxation about Eu2+, or a non-central position of Eu2+, can be an-
ticipated. This is of interest since a lattice relaxation could in principle be an explana-
tion for the observation that upon X-irradiation, electron and hole trap centres are cre-
ated with a spatial correlation to the activator Eu2+, an important feature for the func-
tioning of BaFBr:Eu2+ as a storage phosphor. Information about this question could be
obtained from an ENDOR investigation, in which the shf interaction tensors with the
lattice neighbours are determined [64].
3.4.1 EPR and ENDOR of crystalline BaFBr:Eu2+
The EPR spectrum of Eu2+ in BaFBr consists of seven fine-structure groups each hav-
ing a hf splitting of 12 lines. Eu2+ has an S=72 ground state and two stable isotopes:
151Eu with I=5 2 and 47.82% abundance and 153Eu with I=5 2 and 52.18% abun-
dance. Each Eu isotope thus has a hf structure of 6 lines; since the two nuclear g factors
differ significantly so do the hf interactions (gn(151Eu) = 1.389, gn(153Eu) = 0.6134), and
consequently the two sextets are clearly resolved [63].
Figure 3.16 shows a ENDOR spectrum of 151Eu2+ measured in the mS transition −12-
↔+12 for a orientation of the magnetic field parallel to the crystal c-axis. Besides the
two lines of 151Eu at 4.1 and 20.3 MHz there are several intense lines due to 19F.
Around the Larmor frequency of 19F at 13.2 MHz there are lines at 11.8 MHz and
The X-ray storage phosphor BaFBr:Eu2+
54
15.0 MHz from the nearest and at 13.0 MHz and 13.4 MHz from the next-nearest fluo-
rine nuclei as was determined from their angular dependence. The low intensity lines at
low frequency are probably due to 79Br and 81Br. They were not analysed. Ba lines were
not seen, most likely due to their low abundance of magnetic isotopes 135Ba (6.6%) and
137Ba (11.3%) [64].
In figure 3.17 the fluorine nuclei the shf interaction of which with the Eu2+ activator
were determined are shaded grey. The analysis of these fluorine shf interactions yielded
that no significant shift of Eu2+ away from the regular Ba2+ position occurs [64].
3.4.2 EPR and ENDOR of powdered BaFBr:Eu2+
EPR on powdered BaFBr doped with Eu2+ yielded resonances in a magnetic field range
from 200-500 mT in X-band (9.235 GHz) [64]. The strongest lines are seen between
300 and 370 mT in agreement with the EPR angular dependence of a single crystal,
where the intense line groups of the mS transitions −32↔ −12, −12↔ +12 and
+12↔ +3 2 occur. The spectrum is asymmetric with respect to its transition through
zero at 340 mT. This is a consequence of the rather high fine structure interaction.
Figure 3.16 ENDOR spectrum of Eu2+ in crystalline BaFBr for B || c, B = 331.1 mT
(mS transition −12↔ +12
)
, T = 20 K and a microwave frequency of
9.345 GHz [64].
3.4 Surrounding of the activator Eu2+
5
5
The powder-ENDOR spectrum measured at 331 mT is shown in figure 3.18. In order to
identify the lines, magnetic field shift measurements were performed (see e.g., [10]).
The powder spectrum is a superposition of many single spectra due to the different ori-
entations of the crystalline axes with respect to the magnetic field. If the magnitude of
the magnetic field is changed, the ENDOR lines are measured in single crystal EPR
lines which belong to different magnetic field orientations, i.e. upon shifting the mag-
netic field the field orientation is changed indirectly. Here, because of the possibility of
comparison with the single crystal spectra and their angular dependencies, the assign-
ment of other lines can be made.
The broad intense line between 12.5 and 14 MHz is centred around the Larmor fre-
quency of 19F. The assignment of the ENDOR lines at 10.4 and 16.5 MHz, which be-
long to the nearest fluorine neighbours, was achieved by magnetic field shift measure-
ments in the field range between 320 and 350 mT. Between 325 and 345 mT the lines
are intense. Choosing magnetic field values outside this range, e.g. 320 or 350 mT, the
lines decrease to about a third of the former intensity. The angular dependence of the
single crystal EPR lines yielded that the EPR mS transition −12↔ +1 2 is within the
range of 325 to 345 mT for all orientations of the magnetic field. The angular depend-
Figure 3.17 Centre model of the Eu2+ centre in BaFBr. The defect and its nearest fluo-
rine neighbours the shf interactions of which were determined are shaded
grey.
The X-ray storage phosphor BaFBr:Eu2+
56
ence of the ENDOR lines of the nearest fluorine neighbours belonging to this EPR tran-
sition have many lines per angle in the range of 10.4-16.5 MHz. Therefore, choosing the
magnetic field within the range of 325 to 345 mT one induces the EPR mS transition
−12↔ +12 and thus the corresponding fluorine ENDOR lines. If the magnetic field
is 320 mT or 350 mT, this EPR transition is not saturated, the corresponding fluorine
lines do not contribute to the ENDOR spectrum. One can only see ENDOR lines be-
longing to the weaker contributions of the neighbouring fine-structure groups −32↔-
−12 and +12↔ +32. In this way the assignment of the ENDOR lines presented in
figure 3.18 was made. Even a small change in the local structure of Eu2+, i.e. a change
in the shf interaction parameters by only 10%, would have been seen in the corre-
sponding powder ENDOR spectra by a significant shift of the powder ENDOR lines
[64].
3.4.3 Influence of the production process
Another question is whether the local environment of Eu2+ is changed, depending on
how the phosphor is produced. Stoichiometric or non-stoichiometric BaFBr powders
Figure 3.18 Powder ENDOR spectrum of Eu2+ in BaFBr for B = 331 mT, T = 20 K and
a microwave frequency of 9.233 GHz. The 19F lines of the nearest and the
next-nearest fluorine neighbours as well as their mS quantum numbers are
identified. The frequency dependent background is subtracted [64].
3.4 Surrounding of the activator Eu2+
5
7
could, in principle, differ from the single crystal environment around Eu2+ due to the
different production processes [52]. Since, in particular, it was interesting to study de-
tails about the local structure of Eu2+ in non-stoichiometric BaFBr, which can only be
made as a powder, the above described investigations were performed: First, EPR and
ENDOR of Eu2+ in crystalline BaFBr were measured and analysed. The single crystal
was afterwards crushed to a powder, and the powder EPR as well as the powder-
ENDOR spectrum were recorded and analysed using the single crystal data. Then, the
powder-ENDOR spectra of Eu2+ in stoichiometric and in non-stoichiometric BaFBr
powders, generated by means of a solid state reaction firing appropriate mixtures of
fluorides and bromides [52], were measured. It turned out that no significant shift of
Eu2+ away from the regular Ba2+ position occurs, and that the local properties of Eu2+
are identical within experimental error in all investigated samples.
The X-ray storage phosphor BaFBr:Eu2+
58
59
Chapter 4
Alkali halides and elpasolites
One of the disadvantages of present X-ray storage phosphors is still the unsatisfactory
spatial resolution of the X-ray images. The light scattering of the scanning laser beam
during the read-out process is certainly one of the reasons for that. The BaFBr crystal-
lites in the X-ray storage phosphor image plates are birefringent, a consequence of the
matlockite structure. With respect to the light scattering it would be advantageous to
replace the optically anisotropic BaFBr crystallites by optically isotropic crystals, e.g.
cubic crystals. It is, therefore, a challenge to find cubic X-ray storage phosphors sys-
tems which have a similarly high performance as a standard BaFBr:Eu2+ phosphor
Storage phosphor stimulation
(nm)
CE
(pJ / mm2/ mR)
SE
(µJ / mm2)
CE / SE
(arb. units)
BaFBr:Eu2+ 6331) 20.4 15.7 1300
6802) 14.4 28 510
RbBr:In+6802) 1.9 25.0 77
RbBr:Ga+6802) 5.6 3.9 1470
CsBr:In+6802) 3.0 23.0 140
CsBr:Ga+6802) 5.4 4.3 1370
Table 4.1 Performance of some In+ and Ga+ doped alkali halide X-ray storage phos-
phors in comparison with a standard BaFBr:Eu2+ X-ray storage phosphor
screen after [8]. 1) HeNe-laser, 2) laser diode.
Alkali halides and elpasolites
60
screen. For special applications, in particular for non-destructive testing, systems can
also be used where e.g. a longer radiative life time of the activator luminescence is not a
hindrance, because the read-out time can be longer or the screen is smaller, i.e. less pix-
els have to be read out.
There are two cubic systems which have proved to have storage properties and which
have been investigated very intensely (mostly by luminescence spectroscopy): Alkali
halides doped with Ga+, In+ or Tl+ (e.g. [65, 66]) and Ce3+ or Pr3+ doped elpasolites
(e.g. [9]). In table 4.1 some of the best alternative systems to BaFBr:Eu2+ are listed for
which the conversion efficiency (CE) and the stimulation energy (SE) as well as the
quantity CE / SE are given in comparison to a standard BaFBr:Eu2+ screen. It is seen
that the two alkali halides RbBr:Ga+ and CsBr:Ga+ have excellent figures of merit. In
the following a short characterisation is presented for the X-ray storage phosphors
KBr:In+ and RbI:Tl+, whereas RbBr:Ga+ and CsBr:Ga+ are described in more detail
because of their importance. After that first results on Eu2+-doped RbBr and CsBr are
presented. Finally, the elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+ are briefly char-
acterised.
4.1 KBr:In+
KBr:In+ has been shown to have storage and PSL properties [6]. Although from the
practical point of view it will not be a very important system, since K is not heavy
enough to guarantee a sufficient stopping power. It was, however, interesting to study
Figure 4.1 Models of F, V and H centres in cubic alkali halides.
4.1 KBr:In+
6
1
the mechanism in order to see whether in this case the activator plays the role of the
hole trap centre.
It was shown with MCDA and MCDA-EPR that the radiation damage centres formed
after X-irradiation at RT are In2+, In0(1) (In0 next to an anion vacancy [67]) and F cen-
tres [68]. Upon photostimulation into the F band at 2.06 eV (602 nm), two luminescence
bands of In+ are observed around 2.9 eV (428 nm) and 2.4 eV (517 nm) [6, 69, 70, 71].
In [68] it was shown, that there is, as in BaFBr:Eu2+, a replenishment effect, but only
for a low (100 ppm) In+ doping level. For a high (1000 ppm) In+ doping level, no re-
plenishment effect was observed [8]. The PSL active centres in KBr:In+ are F centres
and In2+ hole trap centres. The role of In0(1) centres is not clear.
The PSL efficiency of KBr:In+ depends clearly on the activator concentration. It was
largest for an In+ concentration between 8·1016 and 2·1017 cm−3 [72]. Investigations of
the stability of the PSL active centres showed that after X-irradiation at RT 20%-30% of
the initially generated F centres decay within 20-30 min [68].
Figure 4.2 Schematic representation of the spatial correlation between F and H centre
pairs in KBr as estimated from cross relaxation spectroscopy. The orienta-
tion of the H centre cannot be inferred from the experiments [60].
Alkali halides and elpasolites
62
4.1.1 Generation of electron and hole trap centres
After X-irradiation of KBr:In+ at 4.2 K the MCDA spectra of F, VK and H centres have
been detected [68]. For a high (1000 ppm) In+ doping level also In2+ hole trap centres
were found [8]. H centres are found as result of the decay of self-trapped excitons,
whereby F centres are formed simultaneously [30]. For the creation of VK centres elec-
tron traps have to be present in the crystal. However, VK centres could also be generated
in particularly pure KBr [60], i.e. there are most probably Br− vacancies present which
capture the electrons. A schematic representation of these centres in cubic alkali halides
is shown in figure 4.1. No In0(1) centres could be observed when X-irradiating at 4.2 K
in contrast to what was found after X-irradiation at RT. It was shown by cross-
Figure 4.3 (a) Absorption of F, VK and H centres in KBr after X-irradiation below
20 K. (b) MCDA spectrum of KBr after X-irradiation at 4.2 K, recorded at
1.5 K [60].
4.1 KBr:In+
6
3
relaxation (CR) spectroscopy in undoped KBr that after X-irradiation at 4.2 K all F and
H centres are spatially correlated. A quantitative analysis of the CR effects yielded a F-
H pair separation of four lattice spacings along the [110] directions as shown in fig-
ure 4.2 [60].
The dominant band in the optical absorption spectrum (figure 4.3a) as well as in the
corresponding MCDA spectrum (figure 4.3b) is that of the F centre at 2.06 eV (602 nm)
[60, 62, 68]. The three extrema in the MCDA at 3.2 eV, 1.65 eV and 1.4 eV correlate
with the peaks of the VK absorption bands [73]. The extremum at 2.8 eV cannot be cor-
related with any known absorption band. It is caused by forbidden transitions of the VK
centre, which become dipole allowed due to the loss of the inversion symmetry by slow
oscillations [74]. Since the known H centre bands at 3.26 eV and 2.50 eV [75] are su-
perimposed by the intense bands of the F and VK centre, respectively, no MCDA band
can be clearly assigned to the H centre. Due to the strong overlap of the H centre bands
with those of the F and the VK centre, respectively, the identification of H centres by
MCDA-EPR turned out to be difficult. This superposition is clearly seen in the MCDA-
EPR spectrum of figure 4.4a which shows not only the EPR line of H centres with their
axes perpendicular to the magnetic field, but also that of the F centre. The fact that only
those H centres can be observed was explained in [31]. Figure 4.4b shows a calculated
Figure 4.4 (a) MCDA-EPR spectrum of KBr X-irradiated at 4.2 K, recorded at 1.5 K
for B || [100]. (b) EPR simulation of H centres in KBr with their axes per-
pendicular to B using g⊥ = 2.074, A⊥ = 5 mT and a halfwidth of 5 mT for
the hyperfine lines [60].
Alkali halides and elpasolites
64
EPR spectrum of H centres with their axes perpendicular to the magnetic field. Upon
annealing to 70 K, the H centres decay thermally.
The failure to detect In2+ centres after X-irradiation at 4.2 K demonstrates that F-In2+
pairs are not produced as primary radiation defects. The doped In+ impurities do not act
as primary hole traps. However, the fact that In2+ are observable at RT shows that the
formation of In2+ centres is thermally activated. From mobile hole trap centres the In+
must capture the hole to form In2+. The analysis of the MCDA-EPR measurements of
In2+ yielded that there is no other defect such as a nearest-neighbour vacancy associated
with the In2+ centre. It was proposed that VK centres decay at an In+ site to form In2+
[68].
4.2 RbI:Tl+
The X-ray induced defects in RbI:X (X = Tl+, In+, Eu2+, Pb2+) and their role in the PSL
process were investigated in detail in [7]. The following section is focussed on RbI:Tl+
which has, in principle, very good storage and PSL properties. However, the stored im-
age fades away within minutes at RT. Therefore, unless a very fast read-out process is
initiated immediately after taking the image, the system is less adequate for practical
use. As to the mechanism, RbI:Tl+ is well understood. After X-irradiation the generated
electron trap centre (F centre) can be photostimulated. The F centre electron recombines
with the complementarily generated Tl2+ hole trap centre leading to the Tl+ emission at
2.86 eV (433 nm). The maximum of the PSL excitation spectrum is at 1.69 eV
(735 nm).
4.2.1 Generation of electron and hole trap centres
X-irradiation of RbI:Tl+ at liquid nitrogen temperature (LNT) generates mainly F and
VK centres, only very few Tl2+ centres are directly formed (figure 4.5a). This was
shown by MCDA and MCDA-EPR experiments [76]. Additionally Tl0 centres can be
detected by their optical absorption bands at 0.87 eV, 1.32 eV and 2.27 eV [7]. Upon
annealing to 150 K the VK centre decays and the MCDA bands of the Tl2+ centre in-
crease (figure 4.5b). Note, that the VK centres become mobile at 125 K [77]. This is in
agreement with the thermoluminescence glow curve of RbI:Tl+ after X-irradiation at
4.2 K which shows two characteristic peaks, at 125 K and at 175 K [7]. No Tl+ dis-
4.2 RbI:Tl+
6
5
turbed VK centres [76] could be detected. At 180 K the Tl0 centres become mobile and
recombine with Tl2+ centres. Consequently, after an additional annealing step to 220 K
almost 95% of the Tl2+ have disappeared (figure 4.5c).
Tl2+ is paramagnetic and has seven absorption bands which could be identified by
“tagged” MCDA experiments [78]. Theoretical calculations [79] showed that the un-
paired 6s-electron of Tl2+ is mainly located at the I− ligands of a [TlI6]4− complex. The
two UV absorption bands (4.07 eV and 4.82 eV) belong to the spin-orbit split 212
P and
232
P components of the a1g
* → t
1u
* transition, whereas the five additional transitions
(1.57 eV, 2.03 eV, 2.35 eV, 2.7 eV and 3.25 eV) can be explained by the components of
the t1u(
π
) → a1g
*, t1u(
σ
) → a1g
*, and t2u(
π
) → a1g
* charge transfer transitions. Bleaching
into any of the seven identified Tl2+ absorption bands at LNT destroys the Tl2+ centres
and leads to the Tl+ emission as well as to the formation of VK centres (figure 4.6, from
left to right). Upon annealing above the stability temperature of the VK centres, 95% of
the Tl2+ centres are restored (figure 4.6, from right to left). Thus, the VK centres gener-
ated in this way are spatially correlated to Tl+.
Figure 4.5 MCDA spectra of RbI:Tl+, recorded at 4.2 K (a) immediately after X-
irradiation at 80 K, (b) after subsequent annealing to 150 K and (c) after
subsequent annealing to 220 K [76].
Alkali halides and elpasolites
66
Photostimulation into the F centre absorption band at temperatures above 150 K leads to
a recombination between Tl2+ and F centres giving rise to the Tl+ luminescence, i.e. the
PSL effect.
4.3 RbBr:Ga+ and CsBr:Ga+
Before X-irradiation the excitation of RbBr:Ga+ or CsBr:Ga+ with UV light leads to a
single luminescence band peaking at 2.25 eV (550 nm) and at 2.41 eV (515 nm), re-
spectively [80, 81]. The corresponding excitation spectra revealed a broad band peaking
at 4.8 eV (260 nm, RbBr:Ga+) and at 5.0 eV (250 nm, CsBr:Ga+), respectively. No lu-
minescence can be detected by exciting RbBr:Ga+ or CsBr:Ga+ with light in the range
between 600 nm and 800 nm. After X-irradiation at RT the Ga+ luminescence can be
excited in the F centre absorption band at 1.85 eV (670 nm, RbBr:Ga+) and at 1.94 eV
(640 nm, CsBr:Ga+), respectively. The PSL decay time is 26.4 µs for RbBr:Ga+ and
18.6 µs for CsBr:Ga+, respectively [82].
4.3.1 Sample preparation
RbBr and CsBr were doped with 200-10 000 ppm Ga+ in the melt and single crystals
were grown by the Bridgman method under inert gas. By using appropriate amounts of
GaBr3 and elementary gallium it was attempted to avoid the incorporation of trivalent
gallium [80, 81]. Due to the crystal growth method the single crystal part grown first
contains only a small Ga+ concentration. The Ga+ concentration in the melt increases
with the crystal growth. Therefore, the crystal’s end part is doped with a much larger
Figure 4.6 Model for the charge transfer character of Tl2+ in RbI:Tl+.
4.3 RbBr:Ga+ and CsBr:Ga+
6
7
amount of Ga+. In addition, the end part of the single crystal also contains more other
unavoidable impurities than the first part. Samples from the first part and from the end
part have been investigated. Henceforth, the samples are referred to as “L” (low Ga+
concentration) and “H” (high Ga+ concentration). The absolute Ga+ concentration in-
corporated into the crystal could not be determined. The maximum concentration is
probably one order of magnitude less than the doping level. From optical absorption
measurements the Ga+ concentration ratio between the L and H samples was found to
be about 5:1.
4.3.2 Generation of electron and hole trap centres
After X-irradiation of RbBr:Ga+ at 4.2 K F, VK, H and I (interstitial halide) centres are
formed. The latter two are not stable above 40 K [80, 83]. The F centre remains stable
up to RT, whereas the VK centre becomes mobile at temperatures above 180 K [80].
Note, that the VK centre in undoped RbBr is stable only to about 170 K [77]. In undo-
ped CsBr the VK centres has two critical temperatures for migration, namely 106 K for
0° jumps and 130 K for 90° jumps, respectively [84]. The maxima at 110 K and 145 K
in the thermoluminescence glow curve of CsBr:Ga+ were thus assigned to the VK decay
[81].
After X-irradiation at RT the MCDA spectra of RbBr:Ga+ and CsBr:Ga+, respectively,
show a derivative-like structured band at 1.85 eV (670 nm, RbBr:Ga+) and at 1.94 eV
(640 nm, CsBr:Ga+), respectively, which belong to the corresponding F centres. Besides
the F centre bands, several MCDA bands were detected in the UV spectral range.
MCDA-EPR experiments showed that the UV bands belong to two different Ga2+ hole
trap centres which are henceforth labelled with (Ga2+)I and (Ga2+)II (figure 4.7). The
EPR lines of each Ga2+ centre are split by a hyperfine (hf) interaction between the un-
paired 4s-electron and the two magnetic isotopes 69Ga (60.4% natural abundance) and
71Ga (39.6% natural abundance), both having a nuclear spin of I=3 2 . The hf interac-
tion leads to four “allowed” ∆mI = 0 transitions marked by bars in figure 4.7. The tran-
sitions labelled with asterisks are due to “forbidden” ∆mI = ±1, ±2 transitions. The for-
bidden transitions are nearly as intense as the allowed ones. This can occur in the
MCDA detection scheme when the allowed transitions are strongly saturated due to
long spin-lattice relaxation times [85]. The allowed quartet lines have the same inten-
sity, but since they are superimposed to forbidden transitions, the intensity pattern is
Alkali halides and elpasolites
68
different from that expected for allowed transitions only. Unfortunately, it was not pos-
sible to resolve structural differences between these two centres by EPR.
The assignment of the UV bands to their corresponding Ga2+ hole trap centre was done
by “tagged” MCDA measurements (figure 4.8), where an energy of 3.35 eV (370 nm) is
very suitable for measuring the MCDA-detected EPR of the (Ga2+)I centres and 4.29 eV
(289 nm) for the (Ga2+)II centres [80, 81].
4.3.3 Generation of (Ga2+)I and (Ga2+)II centres in RbBr:Ga+
When X-irradiating at 4.2 K only F and VK centres are observed in MCDA experiments.
No Ga2+ centres could be found. After annealing to about 180 K, where the VK centres
have begun to disappear, Ga2+ centres start to appear. For RbBr:Ga+ (200 ppm, sample
L), the MCDA band of (Ga2+)I centres appears at 180 K and reaches its maximum value
at 220 K. A weak MCDA band of the (Ga2+)II centres also appears at 180 K, reaches its
maximum value when annealing further to 250 K and then remains unchanged to RT
(figure 4.9). Corresponding to the increase of the MCDA band of the (Ga2+)II centres
the MCDA band of the (Ga2+)I centres decreases between 220 K and 250 K. At 250 K
about half of (Ga2+)I centres are destroyed. For sample H, the MCDA band of (Ga2+)I
Figure 4.7 MCDA-detected EPR of (Ga2+)I and (Ga2+)II centres in RbBr:Ga+
(200 ppm, sample L) after X-irradiation at RT, recorded at 1.5 K applying
a microwave frequency of 23.9 GHz. The detection wavelength was
3.35 eV (370 nm) and 4.28 eV (289 nm), respectively [80].
4.3 RbBr:Ga+ and CsBr:Ga+
6
9
centres appears at 180 K with a further increase to its maximum value at about 220 K.
No MCDA of (Ga2+)II centres is observed, even after annealing up to RT (figure 4.9). In
general, the MCDA of (Ga2+)I centres reaches its maximum value at about 220 K, i.e.
by an annealing step above the decay temperature of the VK centres, while the MCDA
of (Ga2+)II centres reaches its maximum at 250 K.
It is obvious that the (Ga2+)I centres are formed from a mobile VK centre upon hole
capture by Ga+, while the formation of (Ga2+)II centres needs as well the mobility of
another species, thought to be cation vacancies (vc). The (Ga2+)I centre is probably an
isolated Ga2+, whereas the (Ga2+)II centre is proposed to be a Ga2+-cation vacancy com-
plex (figure 4.10), similarly to what was proposed in [86] to occur upon X-irradiation in
KCl and NaCl doped with Ga+. The observation that the VK centre does not recombine
with the F centre electron, but that it is captured by Ga+ could be explained by the fact
that the ionic radius of the Ga+ activator (0.81 Å) is almost twice smaller than that of
Rb+ (1.47 Å). Thus, a relaxation of the bromine neighbours towards the Ga+ could occur
which causes an attractive potential for the VK centres.
The formation process of the (Ga2+)II centres is more complex than that proposed for
(Ga2+)I centres. In KCl and RbCl cation vacancies become mobile at about 220 K [87,
Figure 4.8 “Tagged” MCDA of (Ga2+)I (200 ppm, sample H) and (Ga2+)II centres
(200 ppm, sample L) in RbBr:Ga+, after X-irradiation at RT, recorded at
1.5 K [80].
Alkali halides and elpasolites
70
88, 89]. A similar mobility temperature is also expected for RbBr since the migration
energy of cation vacancies in RbBr (0.81 eV) is very close to that of KCl (0.84 eV) and
RbCl (0.80 eV) [90]. The mobile cation vacancies can be captured by a Ga2+, which
attracts the negative cation vacancies due to its positive charge. The resulting complex
is electrically neutral and stable. Apparently, the investigated RbBr samples with low
Ga+ concentration contain cation vacancies in a considerable concentration, i.e. in the
same order of magnitude as the concentration of Ga+ in those samples which is probably
due to the doping method [80]. When following the formation of Ga2+ centres above
220 K, the initial concentration of (Ga2+)I centres is found to decrease from 220 K to
250 K at the expense of the formation of more (Ga2+)II centres. Thus, cation vacancies
are attracted by the positive (Ga2+)I centres and form (Ga2+)II centres.
After X-irradiation at RT in RbBr with low Ga+ concentration the MCDA shows a sig-
nificant band of (Ga2+)II centres, but only a small band of (Ga2+)I centres. This is in
agreement with the X-irradiation at 4.2 K and subsequent annealing procedure (fig-
ure 4.9). In RbBr with high Ga+ concentration the intensities of both MCDA bands are
of the same order of magnitude with a smaller MCDA band of (Ga2+)II centres. This is
contrary to the fact that after X-irradiation at 4.2 K and subsequent annealing to RT
very few (Ga2+)II centres were found (figure 4.9). A very high mobility of the cation
vacancies during the RT X-irradiation seems to favour the creation of the (Ga2+)II cen-
tres, while after low temperature X-irradiation and annealing to 220 K the VK centres
Figure 4.9 Generation of (Ga2+)I and (Ga2+)II centres in RbBr:Ga+ (200 ppm, sample
L and sample H) after X-irradiation at 4.2 K and subsequent annealing
steps to RT [92].
4.3 RbBr:Ga+ and CsBr:Ga+
7
1
are preferentially trapped at Ga+ to form (Ga2+)I centres and are not converted to
(Ga2+)II centres. It is assumed that RbBr with high Ga+ concentration contains less ca-
tion vacancies than RbBr with low Ga+ concentration. This together with an enhanced
vacancy mobility under RT X-irradiation may be the reason for the different ratio be-
tween (Ga2+)I and (Ga2+)II centres when produced by low temperature X-irradiation and
annealing or by RT X-irradiation.
4.3.4 PSL experiments with RbBr:Ga+ and CsBr:Ga+
Stimulation of the F centre absorption band of X-irradiated RbBr:Ga+ to excite the Ga+
luminescence could not destroy the (Ga2+)II MCDA signal, neither at RT nor at 4.2 K,
but it decreases the (Ga2+)I MCDA band. Thus, only the (Ga2+)I centres are participat-
ing in the read-out process upon photostimulation of the F centre electron. From the
proposed centre models for the two Ga2+ centres this is understandable, since (Ga2+)I
centres are positively charged and attract mobile electrons, while the neutral (Ga2+)II
centres do not. For the use of RbBr:Ga+ as a storage phosphor, the generation of
(Ga2+)II centres must be avoided, since they may compete for primary holes.
After stimulation at 4.2 K it was possible to restore part of the (Ga2+)I centres by subse-
quent annealing to RT in RbBr with low Ga+ concentration, but not in RbBr with high
Ga+ concentration. This “replenishment” effect [32, 33] of the (Ga2+)I centres starts to
appear after annealing the sample to about 200 K and reaches its maximum after an-
nealing to RT (figure 4.11). For a tentative explanation of this observation it is sug-
gested [80] that the Ga+ concentration does not suffice to capture all holes from the
Figure 4.10 Models for the (Ga2+)I and the (Ga2+)II centre in RbBr:Ga+.
Alkali halides and elpasolites
72
generated VK centres. Some holes are trapped elsewhere. Above 200 K they become
mobile and can be trapped by Ga+ centres having become available again after the read-
out process, i.e. after (Ga2+)I centres have recombined with the photostimulated elec-
trons of the F centres. When the dose is lower, the replenishment effect decreases, since
relative to the number of VK centres there are more Ga+ centres available to form
(Ga2+)I centres. In RbBr with high Ga+ concentration practically all VK hole trap centres
are trapped at Ga+ and no replenishment effect could be observed.
In CsBr:Ga the PSL active (Ga2+)I centres clearly dominate over the PSL inactive
(Ga2+)II species for all investigated Ga+ doping levels [81]. The structure of these two
Ga2+ hole trap centres is assumed to be the same as in RbBr:Ga+ [80]. However, in
contrast to RbBr:Ga+, there is no clear replenishment observable in CsBr:Ga+. It has
therefore less complications compared to RbBr:Ga+ with the X-ray induced formation
of PSL inactive (Ga2+)II centres depending on the doping level (and possibly the way of
Ga+ doping).
Figure 4.11 Temperature dependence of the replenishment effect in RbBr:Ga+
(200 ppm, sample L). The PSL decay curves were detected at 2.21 eV
(560 nm) under continuous excitation with 1.85 eV (670 nm) light. The
sample was X-irradiated at RT, read out at 80 K, annealed up to the re-
spective temperature and afterwards again read out at 80 K [92].
4.3 RbBr:Ga+ and CsBr:Ga+
7
3
4.3.5 Optimal activator concentration and Ga+ aggregation
The Ga+ concentration plays an important role in the hole trap centre formation, par-
ticularly in RbBr, and decides whether (Ga2+)I alone or additionally competing (Ga2+)II
centres are formed. In KCl:Ga+ a clear tendency of Ga+ to form aggregates has been
observed [91]. Thus, for the practical use of the systems RbBr:Ga+ and CsBr:Ga+ it is
important to know whether the doped activator forms aggregates or not. The possible
aggregate formation can be detected by measuring the UV excited Ga+ luminescence as
a function of the doping level. The luminescence was detected before and after an an-
nealing / quenching procedure in order to see whether or not aggregates may have been
formed which can be destroyed by the annealing / quenching step [92].
It turned out that in RbBr:Ga+ very high Ga+ doping levels do not seem advantageous,
since a Ga+ aggregation occurs which could not be countered by the annealing and
quenching treatment used. A somewhat surprising result came from the investigation of
the PSL effect as a function of Ga+ doping level. The highest PSL signal is obtained for
a low Ga+ concentration. This means, that apparently the formation of Ga+ aggregates is
Figure 4.12 Ga+ luminescence of CsBr:Ga+ (3000 ppm, sample L) before (curve 1) and
after an annealing / quenching procedure (curves 2 and 3). The annealing
temperature was 340 °C (curve 2) and 520 °C (curve 3), respectively [81].
Alkali halides and elpasolites
74
negative for the PSL effect, while the formation of (Ga2+)II centres seems not to affect
the PSL efficiency. For practical purposes it seems that even a low concentration of
(Ga2+)I centres is sufficient, while too many aggregates seem negative. It remains to be
investigated how the formation of (Ga2+)II centres can be minimised compared to
(Ga2+)I centres. If RbBr:Ga+ is to be used in practical screens, an investigation of the
effect of annealing should be carried out with a variation of annealing temperatures.
In CsBr:Ga+ a significant tendency of Ga+ to aggregate was also found. Annealing (and
subsequent quenching) destroys the aggregates. However, the annealing temperature has
to be sufficiently high. Figure 4.12 shows that an annealing temperature of 340 °C was
not high enough to break all the aggregates, whereas annealing up to 520 °C and subse-
quent quenching yielded an even larger increase in the Ga+ luminescence. Note, that the
melting temperature of CsBr is at 632 °C. Re-aggregation under light takes place within
about a day at RT. When the sample is kept in the dark, experiments on the re-
aggregation showed no significant difference between the signal measured directly after
the annealing and subsequent quenching treatment within a day. Thus the Ga+ re-
aggregation in this case lasts at least some days or weeks.
The PSL is also influenced by the Ga+ aggregation, in particular for higher doping lev-
els. If one wants to use CsBr:Ga+ as X-ray storage phosphor, it is certainly necessary to
consider the Ga+ aggregation. The cooling process has to be adjusted accordingly.
4.3 RbBr:Ga+ and CsBr:Ga+
7
5
4.3.6 PSL fading
The fading of the PSL, i.e. the electron and hole recombination with time in the dark
and thus the partial loss of the stored information, is an important feature for the appli-
cation of RbBr:Ga+ or CsBr:Ga+ as X-ray storage phosphors. Figure 4.13 shows the
fading of the stored information as a function of time measured for RbBr:Ga+
(200 ppm) which revealed the best PSL performance. The sample was X-irradiated for
15 min (60 kV, 15 mA). The decay of the PSL with time was followed by stimulating
the sample with F light (670 nm) and detecting the Ga+ emission at 550 nm several
times from directly after the X-ray exposure up to two days. The stimulating 670 nm
light was very weak, such that the PSL signal measured twice within a few seconds
showed no change in its magnitude. The measured PSL fading with time could be fitted
by two exponential functions with
τ
1 = 5 h and
τ
2 = 125 h. The initial PSL intensity is
halved after approximately 14 h [92].
For RbBr:Ga+ the fading of the PSL was followed in the dark detecting the PSL inten-
sity with weak stimulation light. In case of CsBr:Ga+ the investigation of the F centre
fading by detecting its MCDA showed that after 20 hours one third of the initial F cen-
tre concentration is lost.
Figure 4.13 Fading of the PSL effect of RbBr:Ga+ (200 ppm) in the dark at RT, excita-
tion at 670 nm, detection at 550 nm [92].
Alkali halides and elpasolites
76
4.3.7 Red-shift of the PSL excitation
It is known from previous work on alkali halides that doping with lighter alkali cations
leads to the formation of perturbed F centres, so-called FA centres [62]. In these centres
one of the nearest neighbour cations of the F centre is replaced by an alkali ion of
smaller size, so that an F centre with reduced local symmetry is formed. Figure 4.14
shows the FA centre in RbBr (fcc lattice structure) doped with lithium and the corre-
sponding absorption bands. In this configuration the FA centre absorption band splits
into two bands one of which is red-shifted (FA1), the other one only slightly different
from the normal F band (FA2). If the incident light is polarised along the F-Li+ direction,
the so-called FA1 absorption band is measured which represents the more red-shifted
one. Choosing the polarisation perpendicular to the F-Li+ direction the FA2 band is ob-
served. In table 4.2 the literature data of FA centres in RbBr and CsBr for several co-
dopings are summarised. The FA1 absorption band is further red-shifted the lighter the
doped alkali cation (note, that this simple rule generally applies to the alkali halides
with fcc structure, not to those with bcc structure).
crystal dopant FA2 FA1
1.86 eV (670 nm)
Li 1.78 eV (697 nm) 1.57 eV (790 nm)
Na −−
RbBr
K 1.85 eV (671 nm) 1.67 eV (742 nm)
1.81 eV / 685 nm
Li 1.78 eV (697 nm) 1.60 eV (775 nm)
Na 1.83 eV (678 nm) 1.65 eV (751 nm)
K 1.77 eV (701 nm) 1.58 eV (785 nm)
CsBr
Rb 1.75 eV (709 nm) 1.66 eV (747 nm)
Table 4.2 Optical absorption bands of several FA centres in RbBr and CsBr [62, 93].
4.3 RbBr:Ga+ and CsBr:Ga+
7
7
What is known from the literature is the formation of FA centres using doping levels of
the order of a few percent and additive coloration. Only a statistically small fraction of
F centres is formed as the FA species. Upon bleaching into the F band at RT, a larger
fraction (usually about 50% of F centres) is transformed into FA centres.
It is not clear a priori whether FA centres in alkali halide compounds are formed upon
X-irradiation at all, whether they are possibly preferentially formed compared to normal
F centres and whether they contribute to the read-out process. Upon X-irradiation of
BaFBr:Eu co-doped with Ca2+ or Sr2+, respectively, FA(Ca2+) or FA(Sr2+) centres were
formed which generated the desired red-shift of the PSL excitation spectrum [51, 61].
According to table 4.2 RbBr doped with Li+ should yield the largest red-shift of the FA
centre absorption bands. Figure 4.15 shows the MCDA spectrum of RbBr:(0.1% Ga+,
1% Li+) after X-irradiation at RT. Besides the strong F centre band an additional band
with its maximum peaking at 1.57 eV (790 nm) appears. Corresponding to the data of
table 4.2 this band is assigned to the FA1 transition of perturbed F centres due to the Li+-
doping. After bleaching into the F band at RT using a HeNe laser, a fraction of the F
centres could be converted into FA centres. PSL excitation spectra of RbBr:(Ga+, Li+)
proved that the FA centres are photostimulable in all their absorption bands and, there-
fore, permit read-out of stored X-ray information in the near infrared region [92].
Figure 4.14 Schematic representation of the FA centres in RbBr (fcc structure) doped
with Li+ and the corresponding absorption spectra in comparison with the
unperturbed F centre band [62].
Alkali halides and elpasolites
78
The dose dependence of the amount of the X-ray induced F and FA centres was ana-
lysed by measuring the MCDA of both centres for X-ray doses from 40 mR to 1400 mR
at 1.5 K in 1% Li+-doped RbBr:Ga+. The results are presented in figure 4.16: The mag-
nitudes of the MCDA maxima of the F and FA centres (for the latter the FA1 maximum)
are determined for each dose and plotted versus the X-ray dose. In the high dose regime
the F and FA bands increase approximately to the same degree, whereas in the low dose
regime the F band increases faster than the FA1 band with increasing dose. For doses
lower than 40 mR the signal to noise ratio was too poor to clearly separate the FA1
MCDA band from the background.
Figure 4.17 presents the percentage of FA1 / F versus the applied X-ray dose, calculated
from the data of figure 4.16. It is seen more clearly in this diagram that the FA formation
is preferred when X-irradiating with low doses, comparable to doses applied in medical
diagnostics. The relative fraction of FA1 / F increases to about 10% at 40 mR for the in-
vestigated sample which exceeds the 6% statistical chance (6 next nearest neighbours)
that an F centre is formed adjacent to a Li+ cation. A disadvantage is that the FA(Li+)
centres have a faster fading component compared to the F centres in RbBr:Ga+. Espe-
Figure 4.15 MCDA spectra of RbBr:(0.1% Ga+, 1% Li+), recorded at 1.5 K (a) imme-
diately after X-irradiation at RT and (b) after 60 sec bleaching (HeNe la-
ser) at RT [92].
4.3 RbBr:Ga+ and CsBr:Ga+
7
9
cially a high Li+ doping level of 10% seems disadvantageous, since the half time of the
FA centres is only 2 hours compared to the 14 hours of the F centres [92].
It was not possible to produce FA(K+) centres in RbBr:Ga+ by X-irradiation. However,
Na+-doping leads to an FA1(Na+) band peaking at about 750 nm, less red-shifted than
that of FA(Li+). FA(Na+) has not been followed further, since it seems less attractive.
Figure 4.16 X-ray dose dependence of the intensity of the F and FA1 MCDA bands in
1% Li+-doped RbBr:Ga+. The spectra were recorded at 1.5 K after X-
irradiation at RT [92].
Figure 4.17 Relative amount of FA over F as a function of the X-ray dose for 1% Li+-
doped RbBr:Ga+ [92].
Alkali halides and elpasolites
80
Co-doping CsBr:Ga+ with lighter alkali ions to achieve a red-shift of the PSL excitation
seems less attractive. However, a red-shift of the PSL excitation is feasible even though
it is not as much pronounced as in RbBr:(Ga+, Li+). Unfortunately, the FA1 band re-
sponsible for the red-shift could not be enhanced by a bleaching technique. Moreover,
the creation of a red-shift of the PSL excitation in CsBr:Ga+ by co-doping with light
alkali ions is disadvantageous regarding the PSL performance. Only a Rb+ co-doping
seems promising [92].
4.4 RbBr:Eu2+ and CsBr:Eu2+
CsBr doped with Eu2+ has a figure of merit as high as BaFBr:Eu2+ [22]. Although no
UV excited photoluminescence of Eu2+ could be observed, there was a significant PSL
effect after X-irradiation of CsBr:Eu2+ at RT. It is known that Eu2+ doped into CsBr
forms dipoles of Eu2+ and charge compensating cation vacancies (vc) which agglomer-
ate very quickly to different types of aggregates [94]. The investigated CsBr:Eu2+ sam-
ples did not show any photoluminescence, but an intense X-ray luminescence (XL). The
XL spectra comprised several luminescence bands which are probably due to different
kinds of Eu2+ aggregates. After X-irradiation at RT the PSL spectra stimulated in the F
centre absorption band reproduce the XL spectra quite well. It seems that all Eu2+ ag-
gregates observed in XL participate in the PSL process. The radiative life time of these
Eu2+ aggregates is about 1 ms [22] and thus much longer to the one of isolated Eu2+
(≈ 1 µs).
After X-irradiation at RT there was also a PSL effect in RbBr:Eu2+. Here, it was even
possible to observe UV excited photoluminescence due to isolated Eu2+-vc dipoles and
different Eu2+ aggregates [95]. MCDA investigations yielded several intense bands in
the spectral range between 220 and 420 nm. MCDA-detected EPR showed not only
EPR lines from the [110] oriented Eu2+-vc dipole [96], but also from further Eu2+ ag-
gregates.
4.5 The elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+
8
1
4.5 The elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+
Elpasolites of type Cs2NaYF6 have a cubic crystal structure [97] (figure 4.18). The Na+
and Y3+ ions build a sublattice with NaCl structure. The Cs+ ions occupy the centre of
the unit cell of the Na+-Y3+ sublattice whereas the F− ions reside on the axis connecting
Na+ and Y3+. A trivalent activator, such as Ce3+ or Pr3+, substitutes for Y3+ in sites with
octahedral symmetry.
Under X-ray or UV excitation Ce-doped Cs2NaYF6 shows a strong emission band at
3.44 eV (360 nm) arising from the allowed 5d-4f transition of Ce3+. Cs is a heavy ion,
so the expectation is that if electron and hole trap centres can be formed, storage and
PSL properties will be found. Indeed, it has been found that upon X-irradiation the lu-
minescence of the activator Ce3+ at 3.44 eV (360 nm) can be photostimulated by excita-
tion at about 2.25 eV (550 nm) at 300 K [2, 9, 98]. The PSL decay time of Ce3+ is fast,
namely 42 ns at RT, and is, within experimental error, identical with the Ce3+ radiative
life time upon direct UV excitation. Thus, the bottleneck for the PSL decay time is the
Figure 4.18 Crystal structure of the cubic elpasolite Cs NaYF .
Alkali halides and elpasolites
82
radiative life time of the activator Ce3+. One can assume that the electron-hole recombi-
nation process occurs via tunnelling and not via the conduction band. PSL has also been
observed for Pr3+ as activator [2, 9, 98]. However, the radiative life time of the PSL is
much longer, i.e. 4 ms at RT.
The maximum of the PSL stimulation of Cs2NaYF6:Ce3+ is at about 530 nm, i.e. a fre-
quency-doubled Nd-YAG (533 nm) would be ideal for photostimulation. For a stimula-
tion at 633 nm (HeNe laser) the PSL efficiency is already reduced to 40%. In compari-
son to BaFBr:Eu2+ the PSL stimulation is shifted to shorter wavelength which makes
the use of cheap semiconductor light emitting diodes (LED) as light sources for photo-
stimulation impossible. Investigations on the stability of the photostimulable centres
showed that the PSL efficiency is reduced to half its value after storing X-irradiated
Cs2NaYF6:Ce3+ in the dark at RT for 80 min. This value is still acceptable for an X-ray
storage phosphor. After 30 h the PSL efficiency decreases to a constant value of 15%.
The conversion efficiency (CE) of Cs2NaYF6:Ce3+ was determined to 15 pJ / mm2/ mR,
i.e. comparable to BaFBr:Eu2+, whereas the stimulation energy was 90 µJ / mm2, i.e.
somewhat higher, but still in the same order of magnitude as BaFBr:Eu2+ [99].
MCDA and MCDA-EPR spectroscopy showed that the photostimulable electron trap
centres are F centres. ENDOR measurements have shown that the F centres in this
crystal are of lower symmetry than, for example, in the alkali halides or in BaFBr [100,
101]. The hole trap centre is the activator, i.e. Ce3+ or Pr3+. The proof for this was es-
tablished when examining diamagnetic Pr3+ as an activator, since after X-irradiation this
turned into paramagnetic Pr4+, which could be detected by EPR. Ce3+ ([Xe]4 1
f) is par-
amagnetic due to its unpaired f-electron, whereas Pr3+ ([Xe] 4 2
f) has no unpaired
electron and is thus diamagnetic. After X-irradiation Ce3+ captures a hole and is con-
verted to diamagnetic Ce4+ ([Xe]), but Pr3+ to paramagnetic Pr4+ ([Xe] 4 1
f). To check
whether Ce4+ or Pr4+ is really the corresponding hole trap centre in the PSL process, it
was easier to observe a rise of the Pr4+ EPR than a decrease of the very intense Ce3+
signal intensity.
X-irradiation of undoped Cs2NaYF6 below 100 K creates mainly F and VK centres
which are present in a comparable number. The g tensor and the hyperfine tensor of the
VK centre are very similar to those of the VK centre found in NaF [77]. The VK centre
has two optical absorption bands, at 1.43 eV and at 3.69 eV, which correspond to the
Σu-Πg and the Σu-Σg transition of the X2− molecule, respectively [77]. This assignment
was confirmed by MCDA-EPR. The VK centre decays thermally at temperatures above
4.5 The elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+
8
3
150 K whereby new hole trap centres of X2− type appear which are not exactly aligned
along [110] what is expected for an unperturbed alignment of these centres. It is as-
sumed that they are K+- or Rb+-perturbed H centres (HA centres) [100, 101]. The optical
absorption bands of the HA centre are shifted to higher energies compared to the VK
centre, i.e. to 1.66 eV and 3.86 eV, respectively. The F centre concentration is almost
unaffected by this annealing step to 150 K. The decay of the HA centre at about 230 K is
accompanied by a significant decrease in the F centre concentration.
The F and hole trap centre concentration of Cs2NaYF6:Pr3+ after X-irradiation at 10 K
and subsequent stepwise annealing to RT is shown in figure 4.19. X-irradiation at 10 K
generates not only F and VK centres, but also a small number of Pr4+ centres. Note, that
in figure 4.19 the Pr4+ concentration is scaled up by a factor of 10. The holes are liber-
ated by the VK centre decay at 150 K and captured by interstitial fluorines forming HA
centres. Some of the holes are captured by Pr3+ causing an increase of the Pr4+ concen-
tration by a factor of 2 whereby the PSL efficiency is increased by the same amount.
Annealing to 230 K leads to the HA centre decay and simultaneously to a partial decay
Figure 4.19 F and hole trap centre concentration in Cs2NaYF6:Pr3+ after X-irradiation
at 10 K and subsequent annealing steps up to RT. Gd3+ was present as an
unintended impurity [100].
Alkali halides and elpasolites
84
of the F centres. The F centre concentration is reduced further by annealing to RT,
which is probably due to a recombination with unidentified hole trap centres, whereas
the Pr4+ concentration and the PSL efficiency remains unchanged.
Stimulation experiments have shown that about 50% of the PSL active F and Pr4+ cen-
tres are spatially correlated. This result was obtained by detecting the EPR signal of the
paramagnetic Pr4+ after X-irradiation at RT. Stimulating at 10 K into the F centre ab-
sorption band reduces the Pr4+ concentration considerably in the first minutes, but
reaches a constant value of about 50% upon continued stimulation. The initial exponen-
tial decrease indicates a tunnelling recombination of pairs of electron (F centre) and
hole trap centres (Pr4+) which are separated by definite distances having only a small
variation. In BaFBr doped with the activator Eu2+ the PSL process occurs also via a
tunnelling recombination between the created electron and hole trap centres [23]. A
subsequent stimulation at RT destroys all the remaining Pr4+ centres. This observation
can be explained by the assumption that 50% of the PSL-active F and Pr4+ centres, gen-
erated by X-irradiation at RT, are spatially correlated. It is possible to read them out
even at 10 K by inducing a tunnelling recombination. The non-correlated centres are not
PSL-active at low temperatures. Their recombination requires thermal activation. The
optically excited electron of the F centre may then escape thermally from its relaxed
excited state into the conduction band. Figure 4.20 shows the two proposed recombina-
Figure 4.20 Scheme of the PSL mechanism and the thermal conversion of the gener-
ated hole trap centres in X-irradiated Cs NaYF :Ce3+, Pr3+ [98].
4.5 The elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+
8
5
tion paths and the thermal conversion of the created defects. In Cs2NaYF6:Ce3+ the
mechanism is considered to be analogue to the one in the Pr3+-doped elpasolite, i.e.
Ce4+ plays here the role of the PSL-active hole trap centre [98].
Alkali halides and elpasolites
86
87
Chapter 5
Glasses and glass ceramics
A continuing difficulty in the dimensional X-ray imaging is that in X-ray storage phos-
phor screens considerable scattering of the read-out light occurs, limiting the resolution
to around 5 line pairs / mm. One possible improvement to this scattering problem would
be a phosphor based on a cubic system as discussed in section 5 or a phosphor based on
a glass. There have been reports of the observation of a weak PSL in Eu-doped fluoro-
aluminate glass [102], and in Ce- or Eu-doped borate based glasses [103, 104]. Unfor-
tunately, the magnitude of the weak PSL effect was not stated. A glass phosphor would
have the additional advantages of arbitrary shapes useful for applications in medicine
and non-destructive testing.
In contrast to the literature it was reported [105] that in Eu- or Ce-doped fluoroalumi-
nate glass, or in a Eu-doped borate glass no PSL was observed at RT. However, a sig-
nificant PSL was seen in a modified Eu2+ doped fluorozirconate glass, where a substan-
tial fraction (5%) of the fluoride ions were replaced by bromide ions in the melt result-
ing in a glass ceramic containing small crystallites of a high-pressure phase of BaBr2.
5.1 Production of fluorozirconate glasses and glass ceramics
The well-known ZBLAN20 composition for fluorozirconate glasses was modified by
replacing most of the NaF with NaBr so that the total number of Br− ions was 5% of the
total number of anions, and by substituting some of the LaF3 with YF3 to improve the
glass quality. The composition of the modified glass was then 53% ZrF4, 20% BaF2, 6%
NaF, 14% NaBr, 1.5% LaF3, 3% AlF3, 1% EuF2 or CeF3 and 1.5% YF3. The glass raw
materials were first mixed and then melted in a glassy carbon crucible at about 750 °C,
Glasses and glass ceramics
88
before being poured out into a brass mould at 220 °C and cooled to RT over 12 hours.
The whole glass manufacture was done in an inert nitrogen atmosphere. There was evi-
dence of crystallisation in all of the doped glasses, ranging from a near-transparent glass
which was yellow in transmitted light but blue in scattered light, to one which was
opaque and milky-white. These differences presumably reflect variations in crystallite
size and concentration [105].
5.2 X-ray diffraction of fluorozirconate glasses and glass ce-
ramics
X-ray diffraction of the base glass with no bromine doping showed no sharp peaks, just
broad maxima at 26 and 47 degrees (for Cu Kα radiation) typical for glasses close to the
ZBLAN formulation. However, the bromine-doped glass (milky-white) showed a pat-
tern of relatively sharp diffraction lines from an included crystalline phase superim-
posed on the broad glass background (figure 5.1). Almost all of the lines in the pattern
could be attributed to the so-called high pressure form of BaBr2 which has the anti-Fe2P
structure [107, 108].
Figure 5.1 X-ray diffraction spectrum for Cu Kα radiation of bromine-doped fluoro-
zirconate glass, recorded at RT [106].
5.3 Photoluminescence and PSL of fluorozirconate glasses and glass ceramics
8
9
5.3 Photoluminescence and PSL of fluorozirconate glasses
and glass ceramics
The photoluminescence spectrum of the base glass, without bromine doping, showed no
fluorescence from Eu2+ or Ce3+ ions, although it was clear from EPR and Mössbauer
studies that Eu2+ was certainly present in the Eu2+ doped glass [109]. This quenching of
the fluorescence has been tentatively ascribed to competing charge transfer processes
involving the Zr4+ ions [109]. In contrast, the glass with bromine doping and either Eu2+
or Ce3+ co-doping each show two PL bands, at 413 nm and a broader band centred at
485 nm for Eu2+, and one band at 320 nm and a broader band at 425 nm for Ce3+, as
shown in figure 5.2. Since Eu2+ and Ce3+ ions in the glass itself do not fluoresce, the
observed emissions probably come from Eu2+ or Ce3+ in the crystallites. The bands
were tentatively assigned to 5d-4f transitions of Eu2+ or Ce3+ ions at the two Ba2+ sites
in the BaBr2 crystallites [105]. The PL spectrum which was observed in the Eu2+ glass
ceramic is quite different from the single band observed in the low pressure form of
BaBr2 doped with Eu2+ [110]. However, the emissions could also come from Eu2+ or
Ce3+ containing Br− aggregates in the glass. At present this is an open question.
Of the two luminescent peaks for Eu2+ only that at 413 nm shows a measurable PSL
effect. For the Ce3+-doped glass ceramic, only the peak at 425 nm showed PSL. The
Figure 5.2 Photoluminescence spectra recorded at RT for (a) the Eu2+ doped glass
ceramic and (b) the Ce3+ doped glass ceramic, excited at 270 nm [105].
Glasses and glass ceramics
90
relative magnitudes of the effects were determined by performing identical spectro-
scopic measurements on a piece of crystalline BaFBr:Eu2+ (1000 ppm europium) and
the glass ceramic (milky-white), except for a longer exposure time in the latter case.
This comparison showed that the PSL effect in the glass ceramic was approximately
4000 times weaker in the case of Eu2+ and 2000 times weaker in the case of Ce3+ com-
pared to BaFBr:Eu2+, after the longer exposure time was compensated for. The excita-
tion spectrum for the PSL, being the luminescence intensity at 413 nm for Eu2+ and at
425 nm for Ce3+ as a function of the photostimulation wavelength, comprises a broad
peak centred at about 570 nm, with a similar spectral shape in both cases.
5.4 Electron and hole trap centres
It is known from other X-ray storage phosphors such as elpasolites (e.g. [9]) that Ce3+
acts as a very efficient hole trap centre being converted to Ce4+. Assuming that the dia-
magnetic Ce4+ is the hole trap centre, magnetic resonance methods allow the investiga-
tion of the complementary electron trap centres.
The MCDA of a Ce3+-doped fluorozirconate glass showed an intense paramagnetic
band peaking at 285 nm, which can be assigned to the Ce3+ absorption [111], and a sec-
ond one between 300 and 550 nm with its maximum at 310 nm having opposite sign.
Figure 5.3 MCDA-detected EPR of Ce-doped fluorozirconate glass, detected at (a)
570 nm and (b) 350 nm, recorded at 1.5 K after X-irradiation at RT ap-
plying 24 GHz microwave frequency [114].
5.4 Electron and hole trap centres
9
1
MCDA-detected EPR on the band peaking at 285 nm yielded a single line which can be
simulated using the theoretical g tensor g⊥ = 18 7 and g|| = 6 7 for the J = 5 2 ground
state of the 4 1
f electron of Ce3+ [112]. Note, that in a powder or glass sample all an-
gular orientations are represented with equal probability. The effect of this angular av-
eraging can be mathematically simulated, leading to the calculation of “powder pat-
tern”. In general, paramagnetic centres in a glass will be characterised by a statistical
distribution in the spin Hamiltonian parameters as a consequence of vitreous disorder.
This contrasts with powdered crystalline materials where discrete sets of parameters
pertain [113].
After X-irradiation at RT the MCDA spectrum showed a new band at approximately
570 nm in addition to an increased paramagnetic signal in the range from 300 to 700 nm
which corresponds to the maximum of the stimulation of the PSL [105]. EPR detected
in the MCDA at 570 nm (figure 5.3a) yielded a double-structured line with peaks at
g = 1.98±0.01 and at g = 1.91±0.01, whereas MCDA-detected EPR at 350 nm (fig-
ure 5.3b) showed only a single line at g = 1.90±0.01. The g values of these resonances,
being smaller than ge for the free electron, indicate that electron trap centres have been
formed (see e.g., [10]). These lines did not appear before X-irradiation. “Tagged”
Figure 5.4 “Tagged” MCDA spectra of Ce-doped fluorozirconate glass, detected at
(a) 880 mT and (b) 920 mT, recorded after X-irradiation at 1.5 K applying
24 GHz microwave frequency. For the sake of clarity the spectra are verti-
cally displaced [114].
Glasses and glass ceramics
92
MCDA measurements showed that the low field line at g = 1.98 (detected at 880 mT)
belongs to the new paramagnetic band at 570 nm (figure 5.4a) whereas the high field
lines having g = 1.90-1.91 (detected at 920 mT) belong to the band between 300 and
700 nm (figure 5.4b). Probably X-irradiation creates two defect centres having different
MCDA bands. One centre has a g value of g = 1.98 whereas the second one has
g = 1.90. The double-structured line measured in the MCDA band at 550 nm is thus a
superposition of two EPR lines since both show MCDA there.
93
Summary
The investigation of the information storage and read-out processes in the X-ray storage
phosphor BaFBr:Eu2+ showed that it is necessary to distinguish between stoichiometric
and non-stoichiometric BaFBr. It turned out that the stoichiometric material is always
contaminated with oxygen whereas no oxygen contamination was found in the non-
stoichiometric one.
For stoichiometric as well as for non-stoichiometric BaFBr, the PSL active electron trap
centres are F(Br−) and F(F−) centres. In stoichiometric BaFBr, “impurity” sites (e.g.
oxygen or the activator) are preferred sites for the radiation-induced electron-hole centre
generation. The F centre (electron trap) generation is significantly influenced by the
incorporated oxygen impurities which provide the necessary anion vacancies for the F
centres. The electron-hole pairs are obviously not created via a F-H process. In spite of
thorough research by optically-detected magnetic resonance techniques no H centres
(( )halogen 2
− molecules occupying single halide sites) have been observed. The F centre
generation seems thus to be impurity-limited. The hole centre taking part in the pho-
tostimulated luminescence process is still unknown.
It was shown by magic angle spinning nuclear magnetic resonance that in non-
stoichiometric BaFBr, 10% fluorine “antisites” (fluorines on bromine sites) are present.
The F centre generation occurs via a F-H process between the F− and Br− sublattices
whereas no F-H process within one sublattice was observed. This is in agreement with
the observations in stoichiometric BaFBr where no H centres have been found either.
The existence of fluorine antisites enables the F centre generation. In non-stoichiometric
BaFBr it is thus not impurity limited as in stoichiometric BaFBr. This feature is very
promising for the use of non-stoichiometric BaFBr as X-ray storage phosphors for very
high doses.
The spatial resolution of X-ray storage phosphor screens based on the optically aniso-
tropic BaFBr is limited by the scattering of the read-out light. The scattering can proba-
Summary
94
bly be reduced by using optically isotropic crystal systems instead. The search for such
X-ray storage phosphors, which have also comparable figures of merit as Eu2+-doped
BaFBr, yielded the alkali halides RbBr:Ga+ and CsBr:Ga+ as best candidates, as well as
CsBr:Eu2+. The elpasolite Cs2NaYF6:Ce3+ is also very promising whereas KBr:In+ and
RbI:Tl+ are not so interesting for the practical applications.
The information storage and read-out process in RbBr:Ga+ and CsBr:Ga+ can be ex-
plained by a simple pair mechanism: Upon X-irradiation F centres are generated as
electron trap centres and Ga2+ centres as complementary hole trap centres. Two differ-
ent types of Ga2+ hole trap centres were observed: (Ga2+)I and (Ga2+)II centres. Type I is
an isolated Ga2+ on a Rb+ or Cs+ site, whereas type II is a complex between Ga2+ on
Rb+ or Cs+ site and a nearest neighbouring cation vacancy. Only the type I Ga2+ centres
are taking part in the PSL process. In RbBr:Ga+ a sufficient high Ga+ concentration
suppresses the (Ga2+)II generation. In CsBr:Ga+ the (Ga2+)I centres clearly dominate
over the PSL-inactive (Ga2+)II centres for all Ga+ doping levels.
For practical use, it is advantageous to red-shift the PSL-excitation to the near-infrared
in order to use low cost, but high intensity laser diodes as stimulation light sources in
the read-out process. BaFBr:Eu2+ as well as the two alkali halides RbBr:Ga+ and
CsBr:Ga+ can be co-doped with smaller cations, leading to the formation of perturbed
electron trap centres (FA centres) the absorption bands of which are clearly red-shifted.
It turned out that these sites are preferred sites for the electron-hole centre generation.
Especially for low X-ray doses the ratio between the perturbed FA and unperturbed F
centres is clearly in favour of the perturbed one. This is in analogy to the red-shift of the
PSL excitation in BaFBr:Eu2+: The PSL excitation can be made sensitive to stimulation
further into the infrared by additional Ca2+ or Sr2+ doping. Upon X-irradiation FA(Br−,
Ca2+ or Sr2+) centres are formed preferentially as photostimulable centres electron traps.
The energy storage and read-out processes in the alkali halides KBr:In+ and RbI:Tl+ and
in the elpasolites Cs2NaYF6:Ce3+ and Cs2NaYF6:Pr3+ are completely understood as-
suming a simple pair model of an F centre as electron trap centre and the activator itself
as complementary hole trap centre.
The observation that upon X-irradiation the electron and hole trap centres are always
created with a spatial correlation to the activator is of particular interest for the func-
tioning of the corresponding materials as storage phosphors. The reason for this spatial
correlation is still an open question. A possible explanation for this phenomenon could
be a lattice distortion around the activator due to an ionic radius misfit. However, an
Summary
9
5
electron nuclear double resonance investigation of Eu2+ in BaFBr yielded that such a
lattice distortion would only be very small (below 2%).
Another approach to reduce the scattering of the read-out would be a phosphor based on
a glass. It was shown that a fluorozirconate glass ceramic doped with Eu2+ or Ce3+
shows a significant PSL effect after X-irradiation at RT. The application of such glass
ceramics as X-ray storage phosphors will depend upon optimising the crystallite size
and dopant concentration so as to maximise efficiency and minimise scattering. The
development of such systems is still at its beginning.
Summary
96
97
References
[1] M. Sonoda, M. Takano, J. Miyahara, and H. Kato, Radiology 148, 833 (1983)
[2] J.-M. Spaeth, Th. Hangleiter, F.-K. Koschnick, and Th. Pawlik, Radiat. Eff. De-
fects Solids 135, 499 (1995)
[3] A. R. Lakshmanan, phys. stat. sol. (a) 153, 3 (1996)
[4] H. von Seggern, Brazilian Journal of Physics 29, 254 (1999)
[5] J.-M. Spaeth, Proceedings ICDIM2000 (Johannesburg-Midrand, South Africa,
2000), to be published in Radiation Effects and Defects in Solids
[6] A. Kalnins, I. Plavina, L. Trinkler, and M. Trinkler, Proceedings LUMDETR’91
(University of Latvia, Riga, 1991), p. E18
[7] M. Thoms, H. von Seggern, A. Winnacker, J. Appl. Phys. 76 (3), 1800 (1994)
[8] Th. Hangleiter, U. Rogulis, C. Dietze, J.-M. Spaeth, P. Willems, L. Struye,
P. J. R. Leblans, Proceedings SCINT’95 (Delft University Press, The Nether-
lands)
[9] Th. Pawlik and J.-M. Spaeth, Proceedings SCINT’95 (Delft University Press,
The Netherlands, 1995), p. 392
[10] J.-M. Spaeth, J. R. Niklas, and B. H. Bartram, Structural Analysis of Point De-
fects in Solids, Springer Series in Solid State Sciences 43 (Springer, Berlin,
1992)
[11] L. F. Mollenauer and S. Pan, Phys. Rev. B 6 (3), 772 (1972)
[12] F. Lüty and J. Mort, Phys. Rev. Letters 12, 45 (1964)
[13] C. P. Slichter, Principles of Magnetic Resonance, Springer Series in Solid State
Sciences 1, 3rd Edition (Springer, Berlin, 1989)
References
98
[14] N. Bloembergen, S. Shapiro, P. S. Pershan, and J. O. Artman, Phys. Rev. 114,
445 (1959)
[15] F.-K. Koschnick, Doctoral Thesis, Paderborn (1991)
[16] H. Seidel, Z. Phys. 165, 239 (1961)
[17] K. W. H. Stevens, Proc. Phys. Soc. 65, 209 (1952)
[18] S. Schweizer and J.-M. Spaeth, J. Phys. Chem. Solids 58 (6), 859 (1997)
[19] A. Meijerink, G. Blasse, and L. Struye, Mater. Chem. Phys. 21, 261 (1989)
[20] R. C. Gonzalez and P. Wintz, Digital Image Processing (Addison Wesley, Mas-
sachusetts, 1977)
[21] P. P. Dendy and B. Heaton, Physics for Radiologists (Blackwell Scientific Pub-
lications, Oxford, 1987)
[22] P. Willems, Private Communication (Agfa-Gevaert, Mortsel, Belgium)
[23] H. von Seggern, T. Voigt, W. Knüpfer, and G. Lange, J. Appl. Phys. 64, 1405
(1988)
[24] M. Thoms, H. von Seggern, A. Winnacker, Phys. Rev. B 44, 9240 (1991)
[25] G. Blasse, Journal of Alloys and Compounds 192, 17-21 (1993)
[26] B. W. Liebich and D. Nicollin, Acta Cryst. B33, 2790 (1977)
[27] H. P. Beck, Z. anorg. allg. Chem. 451, 73 (1979)
[28] K. Takahashi, J. Miyahara and Y. Shibahara, J. Electrochem. Soc. 132, 1492
(1985)
[29] H. H. Rüter, H. von Seggern, R. Reiniger and V. Saile, Phys. Rev. Letters 65,
2438 (1990)
[30] K. S. Song and R. T. Williams, Self-trapped Excitons, Springer Series in Solid
State Sciences 105, 2nd Edition (Springer, Berlin, 1993)
[31] J.-M. Spaeth, W. Meise, and K. S. Song, J. Phys.: Condens. Matter 6, 1801-
1814 (1994)
[32] Th. Hangleiter, F.-K. Koschnick, J.-M. Spaeth, R. H. D. Nuttall, and R. S.
Eachus, J. Phys.: Condens. Matter 2, 6837 (1990)
References
9
9
[33] Th. Hangleiter, F.-K. Koschnick, J.-M. Spaeth, and R. S. Eachus, Radiation Ef-
fects and Defects in Solids 119-121, 615 (1991)
[34] F.-K. Koschnick, Th. Hangleiter, J.-M. Spaeth, and R. S. Eachus, J. Phys.: Con-
dens. Matter 4, 3001 (1992)
[35] R. S. Eachus, F.-K. Koschnick, J.-M. Spaeth, R. H. D. Nuttall and W. G.
McDugle, Proceedings Intern. Conf. on Defects in Insulating Materials, Nord-
kirchen, Germany, ed. O. Kanert and J.-M. Spaeth, World Scientific (Singapore)
p. 267
[36] R. S. Eachus, R. H. D. Nuttall, W. G. McDugle, F.-K. Koschnick and J.-M.
Spaeth, Proceedings Intern. Conf. on Defects in Insulating Materials, Nord-
kirchen, Germany, ed. O. Kanert and J.-M. Spaeth, World Scientific (Singapore)
p. 1172
[37] R. S. Eachus, W. G. McDugle, R. H. D. Nuttall, M. T. Olm, F.-K. Koschnick,
Th. Hangleiter, and J.-M. Spaeth, J. Phys.: Condens. Matter 3, 9327 and 9339
(1991)
[38] R. S. Eachus, R. H. D. Nuttall, M. T. Olm, W. G. McDugle, F.-K. Koschnick,
Th. Hangleiter, and J.-M. Spaeth, Phys. Rev. B 52, 3941 (1995)
[39] S. Schweizer and J.-M. Spaeth, J. Phys.: Condens. Matter 11, 1723 (1999)
[40] F.-K. Koschnick, J.-M. Spaeth, R. S. Eachus, W. G. McDugle and R. H. D.
Nuttall, Phys. Rev. Letters 67, 3571 (1991)
[41] R. U. Bauer, J. R. Niklas and J.-M. Spaeth, phys. stat. sol. (b) 118, 557 (1983)
[42] K. Takahashi, K. Kohda, J. Miyahara, Y. Kanemitsu, K. Amitani, S. Shionoya,
J. Luminescence 31 & 32, 266 (1984)
[43] Y. Iwabuchi, C. Umemoto, K. Takahashi, and S. Shionoya, Journal of Lumines-
cence 48 & 49, 481 (1991)
[44] A. Meijerink and G. Blasse, J. Phys. D 24, 626 (1991)
[45] D. M. De Leeuw, T. Kovats and S. P. Herko, J. Electrochem. Soc. 134, 491
(1987)
[46] F.-K. Koschnick, J.-M. Spaeth, and R. S. Eachus, J. Phys.: Condens. Matter 4,
8919 (1992)
References
100
[47] Y. Kondo, T. Tezuka, and Y. Iwabuchi, Proceedings ICDIM2000 (Johannes-
burg-Midrand, South Africa, 2000), to be published in Radiation Effects and
Defects in Solids
[48] R. J. Klee, J. Phys. D: Appl. Phys. 28, 2529 (1995)
[49] F. K. Koschnick, T. Hangleiter, K. S. Song, and J.-M. Spaeth, J. Phys.: Condens.
Matter 7, 6925 (1995)
[50] E. Radzhabov and V. Otroshok, J. Phys. Chem. Solids 56, 1 (1995)
[51] C. Dietze, Th. Hangleiter, P. Willems, R. J. R. Leblans, L. Struye, and J.-M.
Spaeth, J. Appl. Phys. 80, 1074 (1996)
[52] S. Schweizer, J.-M. Spaeth and T. J. Bastow, J. Phys.: Condens. Matter 10, 9111
(1998)
[53] M. H. Cohen and F. Reif, Solid State Physics 5, 321 (1957)
[54] N. Boden, P. K. Kahal, A. Mee, M. Mortimer, and G. N. Peterson, Journal of
Magnetic Resonance 54, 419 (1983)
[55] R. W. Vaughan, D. D. Elleman, W.-K. Rhim, and L. M. Stacey, Journal of
Chem. Phys. 57, 5383 (1972)
[56] F.-K. Koschnick, J.-M. Spaeth, and R. S. Eachus, J. Phys.: Condens. Matter 4,
3015 (1992)
[57] W. Hayes, Crystals with the fluoride structure (Clarendon, Oxford, 1974)
[58] R. C. Baetzold, J. Chem. Phys. Solids 50, 915 (1989)
[59] R. C. Baetzold, Phys. Rev. B 36, 9182 (1987)
[60] W. Meise, U. Rogulis, F.-K. Koschnick, K. S. Song, and J.-M. Spaeth, J. Phys.:
Condens. Matter 6, 1815 (1994)
[61] S. Schweizer, P. Willems, P. J. R. Leblans, L. Struye, J.-M. Spaeth, J. Appl.
Phys. 79, 4157 (1996)
[62] Physics of Color Centres, edited by W. B. Fowler (Academic Press, New York
and London, 1968)
[63] D. Nicollin and H. Bill, J. Phys. C 11, 4803 (1978)
[64] S. Assmann, S. Schweizer, and J.-M. Spaeth, phys. stat. sol. (b) 212, 325 (1999)
References
1
01
[65] H. von Seggern, A. Meijerink, T. Voigt, and A. Winnacker, J. Appl. Phys. 66,
4418 (1989)
[66] P. F. Braslavets, A. Kalnins, I. Plavina, A. I. Popov, B. I. Rapoport, A. Tale,
phys. stat. sol. (b) 170, 395 (1992)
[67] F. J. Ahlers, F. Lohse, Th. Hangleiter, J.-M. Spaeth, and R. H. Bartram, J. Phys.
C: Solid State Phys. 17, 4877 (1984)
[68] U. Rogulis, I. Tale, Th. Hangleiter, and J.-M. Spaeth, J. Phys.: Condens. Matter
7, 3129 (1995)
[69] A. Fukuda, Phys. Rev. B 1, 4161 (1970)
[70] I. Plavina, I. Tale, and A. Tale, Proceedings LUMDETR’91 (University of Lat-
via, Riga, 1991), p. D4
[71] L. E. Trinkler, M. F. Trinkler and A. I. Popov, phys. stat. sol. (b) 180, K31
(1993)
[72] A. Kalnins, I. Plavina, and A. Tale, Nuclear Instruments and Methods in Physics
Research B 84, 95-101 (1994)
[73] C. J. Delbecq, W. Hayes, and P. H. Yuster, Phys. Rev. B 121, 1043 (1961)
[74] U. Rogulis, K. S. Song, and J.-M. Spaeth, J. Phys.: Condens. Matter 7, 7699
(1995)
[75] B. J. Faraday and W. D. Compton, Phys. Rev. A 138, 893 (1965)
[76] U. Rogulis, C. Dietze, Th. Pawlik, Th. Hangleiter, and J.-M. Spaeth, J. Appl.
Phys. 80, 2430 (1996)
[77] D. Schoemaker, Phys. Rev. B 7, 786 (1973)
[78] U. Rogulis, J.-M. Spaeth, I. Cabria, M. Moreno, I. A. Aramburu, and M. T. Bar-
riuso, J. Phys.: Conden. Matter 10, 6473 (1998)
[79] I. Cabria, M. Moreno, I. A. Aramburu, M. T. Barriuso, U. Rogulis, and J.-M.
Spaeth, J. Phys.: Conden. Matter 10, 6481 (1998)
[80] U. Rogulis, S. Schweizer, S. Assmann, and J.-M. Spaeth, J. Appl. Phys. 84, 4537
(1998)
[81] U. Rogulis, S. Schweizer, S. Assmann, and J.-M. Spaeth, J. Appl. Phys. 87, 207
(2000)
References
102
[82] C. Dietze, Diploma Thesis, Paderborn (1995)
[83] Yu. V. Kolk and A. Ch. Lushchik, Sov. Phys. Solid State 28, 805 (1986)
[84] J. J. Pilloud and C. Jaccard, phys. stat. sol. (b) 92, 233 (1979)
[85] F.-K. Koschnick, K.-H. Wietzke, and J.-M. Spaeth, Phys. Rev. B 58 (12), 7707
(1998)
[86] P. G. Baranov and V. A. Khramtsov, Sov. Phys. Solid State 20, 1080 (1978)
[87] C. J. Delbecq, R. Hartford, D. Schoemaker, and P. H. Yuster, Phys. Rev. B 31,
3631 (1976)
[88] F. van Steen and D. Schoemaker, Phys. Rev. B 19, 55 (1979)
[89] E. Goovaerts, J. Andriessen, S. V. Nistor, and D. Schoemaker, Phys. Rev. B 24,
29 (1981)
[90] D. K. Rowell and M. J. L. Sangster, J. Phys. C: Solid State Phys. 14, 2909-2921
(1981)
[91] F. J. Ahlers, Doctoral Thesis, Paderborn (1985)
[92] S. Assmann, Doctoral Thesis, Paderborn (2000)
[93] S. Radhakrishna, N. N. Nigam and V. S. Sivasankar, Phys. Rev. B 15, 1187
(1977)
[94] V. P. Savelev, V. P. Avdonin, L. D. Dugarova, A. P. Nadashkovskii, and B. T.
Plachenov, Soviet Physics Solid State 16, 700 (1974)
[95] C. P. Medrano, H. S. Murrieta, and J. O. Rubio, Journal of Luminescence 29,
223 (1984)
[96] E. P. Munoz, J. O. Rubio, H. S. Murrieta, G. S. Aguilar, and J. O. Boldu, J.
Chem. Phys. 62, 3416 (1975)
[97] D. Babel, Mater. Res. Bull. 8, 1371 (1973)
[98] Th. Pawlik and J.-M. Spaeth, J. Appl. Phys. 82 (9), 4236 (1997)
[99] Th. Pawlik, Doctoral Thesis, Paderborn (1996)
[100] Th. Pawlik and J.-M. Spaeth, Material Science Forum 239-241, 287 (1997)
[101] Th. Pawlik and J.-M. Spaeth, phys. stat. sol. (b) 203, 43 (1997)
References
1
03
[102] J. Qiu, Y. Shimizugawa, Y. Iwabuchi, and K. Hirao, Appl. Phys. Lett. 71, 759
(1997)
[103] J. Qiu, Y. Shimizugawa, Y. Iwabuchi, and K. Hirao, Appl. Phys. 71, 43 (1997)
[104] J. Qiu, Y. Shimizugawa, N Sugimoto, and K. Hirao, J. Non-Cryst. Solids 222,
290-295 (1997)
[105] A. Edgar, J.-M. Spaeth, S. Schweizer, S. Assmann, P. J. Newman, and D. R.
MacFarlane, Appl. Phys. Lett. 75, 2386 (1999)
[106] S. Schweizer, S. Assmann, A. Edgar, and J.-M. Spaeth, Nuclear Instruments and
Methods in Physics Research B 166-167, 508-510 (2000)
[107] Joint Committee on Powder Diffraction Standards, (Swarthmore, USA, 1996),
Powder Diffraction File entry #45-1314
[108] G. Liu and H. A. Eick, Journal of Less Common Metals 149, 47 (1989)
[109] D. R. MacFarlane, P. J. Newman, J. D. Cashion, and A. Edgar, J. Non-Cryst.
Solids 257, 53-58 (1999)
[110] N. Iwase, S. Tadaki, S. Hidaka and N. Koshino, Journal of Luminescence 60 &
61, 618 (1994)
[111] G. M. Williams, T.-E. Tsai, C. I. Merzbacher, and E. J. Friebele, Journal of
Lightwave Technology 15, 1357 (1997)
[112] A. Abragam and B. Bleaney, Electron Paramagnetic Resonance of Transition
Ions (Dover Publications, New York, 1986)
[113] D. L. Griscom, J. Non-Cryst. Solids 40, 211-272 (1980)
[114] S. Schweizer, S. Assmann, A. Edgar, and J.-M. Spaeth, Nuclear Instruments and
Methods in Physics Research B 166-167, 505-507 (2000)
References
104
1
05
Mein Dank geht an erster Stelle an Herrn Prof. Dr. J.-M. Spaeth, der mich in allen Pha-
sen meiner wissenschaftlichen Laufbahn stets unterstützt und gefördert hat. Er hat mir
mit zahlreichen Diskussionen und persönlichen Gesprächen ganz wesentlich bei der
Anfertigung dieser Arbeit geholfen.
Im weiteren möchte ich mich bei Herrn Dr. S. Assmann und Herrn Dr. U. Rogulis für
die ausgezeichnete Zusammenarbeit bedanken.
Für die problemlose Versorgung mit flüssigem Helium danke ich Herrn Dr. F. Lohse
und Herrn J. Pauli. Weiterhin möchte ich mich bei den Kollegen des Kristallzuchtlabors
Herrn Dr. Th. Hangleiter, Herrn D. Niggemeier und Herrn R. Winterberg bedanken.
Allen namentlich nicht erwähnten Mitgliedern der Arbeitsgruppe danke ich für die an-
genehme Arbeitsatmosphäre.
