Optically Stimulated Luminescence Dosimetry: Stephen W.S. Mckeever

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Optically stimulated luminescence dosimetry

Stephen W.S. McKeever


*
Department of Physics, Oklahoma State University, 145 Physical Sciences Building, Stillwater, OK 74078-3072, USA
Received 17 January 2001; received in revised form 16 March 2001
Abstract
Models and the conceptual framework necessary for an understanding of optically stimulated luminescence (OSL)
are described. Examples of various OSL readout schemes are described, along with examples of the use of OSL in
radiation dosimetry. 2001 Elsevier Science B.V. All rights reserved.
1. Introduction
In recent years optically stimulated lumines-
cence (OSL) has become a popular procedure for
the determination of environmental radiation do-
ses absorbed by archaeological and geological
materials in eorts to date those materials. In this
procedure the target samples (usually natural
grains of quartz and/or feldspar) are exposed in
the laboratory to a steady source of light of ap-
propriate wavelength and intensity, and the lumi-
nescence stimulated from the mineral during this
procedure is monitored as a function of the stim-
ulation time. The integral of the luminescence
emitted during the stimulation period is a measure
of the dose of radiation absorbed by the mineral
since it was last exposed to light. Through cali-
bration of the signals against known doses of ra-
diation, the absorbed dose can be obtained and
through a separate determination of the environ-
mental dose rate the age of the sample can be
determined. Huntley et al. [1] rst used the meth-
od, now known as ``CW-OSL'', for this purpose
and the latest developments in this eld have been
described in the triennial conferences on lumines-
cence and ESR dating [25].
The use of OSL as a personal dosimetry tech-
nique, however, is not yet so widespread, despite
the fact that its use in this eld has a much longer
genesis. It was rst suggested for this application
several decades ago by Antonov-Romanovskii et
al. [6] and was later used in this context by Braun-
lich et al. [7] and Sanborn and Beard [8]. Since these
early developments, however, the use of OSL in
radiation dosimetry has not been extensively re-
ported, mainly because of the lack of a good lu-
minescent material which was both highly sensitive
to radiation, and had a high optical stimulation
eciency, a low eective atomic number, and good
fading characteristics (i.e. a stable luminescence
signal at room temperature). MgS, CaS, SrS and
SrSe doped with dierent rare earth elements such
Nuclear Instruments and Methods in Physics Research B 184 (2001) 2954
www.elsevier.com/locate/nimb
*
Tel.: +1-405-744-5796; fax: +1-405-744-6811.
E-mail address: u1759aa@okstate.edu (S.W.S. McKeever).
0168-583X/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 5 8 3 X ( 0 1 ) 0 0 5 8 8 - 2
as Ce, Sm and Eu were among the rst phosphors
suggested for OSL dosimetry applications [79].
They possess a high sensitivity to radiation and a
high eciency under infrared stimulation at a
wavelength around 1 lm, but they suer from sig-
nicant fading of the luminescence at room tem-
perature. These phosphors also have a very high
eective atomic number and, as a result, they ex-
hibit strong photon energy dependence, which is
unacceptable for use in personal dosimetry.
Several research groups have tried to use optical
stimulation as a dosimetric tool by using light to
transfer trapped charge carriers from deep traps to
shallow traps and then monitoring the phospho-
rescence at room temperature as the charge leaked
away from the shallow traps. This approach was
suggested for fast neutron dosimetry for which one
can mix the phosphor with polyethylene to mea-
sure the absorbed dose from recoil protons and
perform the luminescence measurements at room
temperature. Several phosphors such as BeO
[10,11], CaF
2
:Mn [12,13] and CaSO
4
:Dy [14,15]
were used in this mode but they each exhibited
relatively lowsensitivity. This OSL readout mode is
often described as ``delayed'' OSL (DOSL) [16].
A similar technique is sometimes referred to as
cooled optically stimulated luminescence (COSL)
[17] but, to be in keeping with past tradition, this
should be dened more correctly as phototrans-
ferred thermoluminescence (PTTL). It consists of
cooling an irradiated phosphor to a temperature
below room temperature and performing an illu-
mination of the sample at the low temperature in
order to transfer charge from deep traps to shallow
traps. The sample is then heated to record the low
temperature thermoluminescence (TL) caused by
the charge transfer. In this sense it is similar to
DOSL, but the shallow traps involved in this
process are stable at the temperature of illumina-
tion such that the sample needs to be heated before
the luminescence can be subsequently stimulated.
In yet another OSL mode, irradiation of the
detector material induces stable radiation-induced
defects, the concentration of which is proportional
to the absorbed dose. Subsequent illumination of
the sample with light results in the excitation of
electrons from the defect ground state to the defect
excited state. Relaxation from the excited back to
the ground state yields luminescence, the intensity
of which is in proportion to the absorbed dose.
This approach is signicantly dierent from those
discussed so far in that the stimulation with light
does not result in the ionization of the defect, but
only in its excitation. Thus, the dose can be read
multiple times without destroying the signal. Dis-
advantages are that the signal cannot be reduced
to zero by this approach, and the sensitivity of the
technique is relatively low because it requires a
high concentration of radiation-induced defects
(i.e. a high level of absorbed dose). Examples of
this approach are given for F
2
-centers (or M-cen-
ters) in alkali halides [18,19] and radiophotolumi-
nescence (RPL) from glasses [20,21].
For those OSL methods which rely upon the
ionization of the radiation-induced defect, a new
modication, called pulsed optically stimulated
luminescence (POSL), was recently reported using
crystalline Al
2
O
3
:C as the luminescent material
[2224]. Here one exposes irradiated Al
2
O
3
:C to a
pulsed light source and synchronously detects the
emitted luminescence between pulses, but not
during the pulse. This synchronous arrangement
allows one to use less optical ltration than with
CW-OSL, which is used in the latter method to
discriminate between the stimulation light and the
luminescence. At the same time the POSL method
allows one to bias against the slow phosphores-
cence processes, which make up the main signal in
DOSL measurements. These features grant the
POSL technique both a high sensitivity and
weaker temperature dependence compared with
the DOSL method (and the PTTL/COSL method),
and does not require the use of cryogenics as in
PTTL/COSL. The high sensitivity and rapid
readout features also allow use of the method for
imaging the distribution of the dose over large area
detectors [25].
In general, OSL techniques have advantages
over conventional TL techniques for a number of
reasons. The most obvious advantage lies in the
fact that the readout method is all optical, requir-
ing no heating of the samples (although some ad-
ditional advantages may be gained by performing
the optical stimulation at slightly elevated tem-
peratures, as will be discussed later in this paper).
Apart from removing the need to provide a reliable
30 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
and reproducible heating scheme, this also means
that problems due to thermal quenching of the
luminescence eciency are also removed. This is
particularly true for Al
2
O
3
:C dosimeters for which
adoption as a TL material has been handicapped
by a heating-rate dependence of the TL sensitivity
caused by thermal quenching of the F-center lu-
minescence eciency [26]. By using optical stimu-
lation, however, the readout of the luminescence is
performed at temperatures lower than those for
which thermal quenching occurs, and thus a sig-
nicant increase in sensitivity is achieved. The all-
optical nature of the readout process also allows
the use of ``plastic'' dosimeters namely, lumines-
cence phosphors impregnated into a plastic matrix
(e.g. polytetrauoroethylene, PTFE). In this way,
robust dosimeters may be manufactured and ad-
vantages may be gained for neutron dosimetry
through the interaction of neutrons with hydrogen
atoms producing knock-on protons, which then
yield luminescence from the phosphor through
ionization processes. The high sensitivity of OSL
also leads to advantages related to multiple read-
ings since it is sometimes not necessary to stimulate
all of the trapped charge in order to read a su-
cient luminescence signal. In this way, a residual
signal remains which can be stimulated at a later
time if second, or third, etc., readout of the signal is
necessary for dose verication purposes. Finally,
the readout process can be made very fast by ad-
justing the stimulating light intensity (power)
leading to advantages associated with the rapid
analysis of large numbers of dosimeters.
In this review we shall examine some of the
fundamental processes taking place during optical
stimulation, and discuss the relationship of the
OSL signals with TL and with PTTL. Some basic
models for describing the OSL kinetics will be
discussed, and examples of the use of OSL in
dosimetry will be examined.
2. Optical phenomena
2.1. Bleaching, phototransfer and trap lling
OSL dosimetry relies upon the optically in-
duced transitions of electronic charge between ra-
diation-induced defect centers, leading to the
release of energy via luminescence. This is in
contrast to TL dosimetry, which relies on ther-
mally stimulated transitions to yield the same end
eect. However, it is not clear that the same defect
centers are involved in both processes in any one
material. To begin the discussion of the possible
optical processes that are relevant to the produc-
tion of OSL, therefore, we will start with an ex-
amination of how optical transitions can aect TL
processes, with examples extracted from several
well-known dosimetry materials.
Optical processes can be important in TL in
three ways:
1. Optical bleaching of a radiation-induced TL sig-
nal. If, after irradiation, the sample is exposed
to visible light, one often nds that the TL sig-
nal level is reduced from that which would be
expected if the material had not been exposed
to light. The eect is due to optically stimulated
transfer of the charge out of the trap. The long-
er the light exposure, the weaker the subsequent
TL signal. This eect is also known as ``light-in-
duced fading''.
2. Phototransferred TL arises when an irradiated
sample is exposed to light and charge is opti-
cally stimulated out of deep traps and trans-
ferred to shallower traps. If the shallow traps
are empty at the start of the illumination a TL
signal from these traps can be induced due to
the phototransfer. If the shallow traps are par-
tially lled at the start of the illumination, a lar-
ger TL peak can result than would have been
obtained without the illumination. To observe
PTTL experimentally the sample is either illu-
minated at a lower temperature than that at
which it is irradiated, or the sample is preheated
after the irradiation to empty the shallow traps
before the illumination.
3. Optically induced TL arises when an unirradi-
ated sample is exposed to light. Optically stimu-
lated transitions of electronic charge from
pre-existing defects can result in free charges
that subsequently become trapped at empty
trapping levels, thereby inducing a TL signal.
Whenever a sample is illuminated each of the
above eects can be taking place. The strength of
each eect depends upon the irradiation and
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 31
pre-heating conditions before the illumination.
Each may be exploited in dosimetry in varying
ways, and have been used in dierent materials for
a variety of dosimetry purposes. In the sections to
follow we will discuss the processes in detail and,
in doing so, comment on their relationship to
OSL. First, however, we discuss some basic fea-
tures of optically stimulated transitions in dosi-
metry materials.
2.2. General optical phenomena
2.2.1. Optical absorption transitions
The absorption of light of incident intensity
I
0
(k) at wavelength k as a function of distance x in
a solid is described by the LambertBeer Law,
namely
I(k; x) = I
0
(k) expa(k)x; (1)
where a(k) is the absorption coecient and I(k; x)
is the intensity at position x. The value of a de-
pends upon the optical absorption process occur-
ring at that wavelength. A schematic view of the
various optical absorption transitions possible and
important for dosimetry is illustrated in Fig. 1.
Free carrier transitions to higher states are not
considered. Band-to-band transitions (transition
(1)) are not important for wide band gap insula-
tors that may be used in dosimetry (e.g.
Al
2
O
3
; SiO
2
), but are important for narrow band
semiconductors (e.g. ZnSe). Exciton formation
(transition (2)) can lead to TL, but such transitions
generally occur in the vacuum ultraviolet and are
not usually important in dosimetry. However, all
other transitions in Fig. 1 can be important. Ex-
amples of defect ionization (transitions of type 3)
include ionization of the F-center in Al
2
O
3
:C in
which absorption of a photon at approximately
6.05 eV induces an electron transition from the 1A
ground state to the 2P excited state of the F-center.
Since the 2P state is accessible to the conduction
band, ionization can occur, leading to trapping
and, subsequently, TL [27]. This is an example of
process (c) from the previous section.
Intraband transitions of type 5 are important in
certain optically stimulated processes involved in
the dosimetry of natural materials. Specically,
absorption on an IR photon (at approximately 840
nm) induces a ground-state-to-excited-state tran-
sition in a radiation-induced defect in feldspar
minerals. If the temperature is suciently high
thermal ionization of the excited state electron can
then occur, leading to luminescence. This phe-
nomenon is exploited in the dosimetry of natural
radiation absorbed by feldspars in luminescence
dating applications [28,29].
Transitions of type 4 yield OSL. Such transi-
tions result from initial localization of charge by
traps during irradiation, and the optically stimu-
lated release of those trapped charges from the
traps through the absorption of light. Subsequent
recombination of the charge results in OSL emis-
sion. Transitions of this type also give rise to
phototransfer eects (e.g. PTTL) and optical
bleaching of TL. In each case the subsequent lu-
minescence signal (OSL, PTTL or TL) is a func-
tion of the initial dose of radiation absorbed, and
the intensity, wavelength and duration of the op-
tical stimulation.
It is worth noting that the OSL eects discussed
here are to be distinguished from photolumines-
cence, which is the luminescence emitted following
absorption of light by an internal transition, such
as that depicted in transition (5). Since these do
not involve transport of charge from one defect
site to another such transitions do not aect the
TL signal, unless the excited state is thermally
unstable, as discussed above for feldspars.
Fig. 1. Schematic representation of the possible optical ab-
sorption transitions in an insulator: (1) ionization (excitation
across the band gap); (2) exciton formation; (3) defect ioniza-
tion; (4) trap ionization; (5) internal intra-center transition.
32 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
However, they can yield useful information about
the available luminescence sites, for example, the
site symmetries of the rare earth emitters in rare-
earth doped alkaline earth uorides [30]. Fur-
thermore, photoluminescence (PL) can also be
used as a dosimetry method if the defect in which
the internal transition is occurring is itself a
radiation-induced defect. Examples here include
radiation-induced photoluminescence absorption
in glasses [20,21] and LiF [18,19]. The latter uses
the photoluminescence signal from F
2
-centers
formed by the reaction between two radiation-
induced F-centers.
2.2.2. Photoionization
Ionization of radiation-induced trapped charge
(type-4 transitions, Fig. 1) requires the absorption
photons of energy greater than or equal to the
defect ionization energy (i.e. the optical trap
depth). The optical trap depth E
0
can be substan-
tially larger than the thermal trap depth E
t
by an
amount equal to S = hx
p
where S is the Huang
Rhys factor and hx
p
is the lattice phonon vibra-
tional energy. Estimates of the ratio of the optical
to thermal trap depths yield an approximation
E
0
=E
t
= e
0
=e
s
, where e
0
and e
s
are the high fre-
quency and static dielectric constants, respectively
[31]. The absorption coecient for ionization of a
defect from a trap with optical stimulation of
wavelength k is
a(k) = Nr
0
(k); (2)
where N is the defect concentration and r
0
(k) is
the photoionization cross-section of the defect at
that wavelength. For deep levels and parabolic
bands [32] this is expressed by
r
0
(hm) = C

E
0
p
(hm E
0
)
3=2
hm(hm gE
0
)
2
for hm PE
0
(3a)
and
r
0
(k) = 0 for hm 6E
0
: (3b)
Here hm is the energy of the stimulating photon of
wavelength k. In the above expression C is a
constant and g is related to the electron eective
mass m
+
by
g = 1
m
0
m
+
(4)
where m
0
is the free carrier rest mass.
The photoionization cross-section of a trap is
the most important property in dictating the
wavelength or wavelength range that is most
appropriate to use for optical stimulation of the
trapped charge. It can be determined experi-
mentally by a number of techniques. Consider
illumination of an irradiated sample with n
trapped electrons at N traps, each with an op-
tical ionization energy of E
0
. Electrons are
stimulated from the traps into the conduction
band (viz. transition 4a in Fig. 1) from where we
allow n
c
free electrons to either be retrapped or
recombine with trapped holes to produce OSL.
If the illumination ux at wavelength k is U(k),
then under steady-state conditions (i.e.
dn
c
=dt = 0), we have
U(k)r
0
(k)n = A(N n)n
c
A
m
n
c
m; (5)
where A is the probability (in m
3
s
1
) of capture of
free electrons, and A
m
is the probability of re-
combination of free electrons with trapped holes
of concentration m. Under conditions of weak
stimulation n and m remain approximately con-
stant during the stimulation period (i.e. the num-
ber of charge carriers removed is much less than
the initial number, or Dn n). Thus, we have
from Eq. (5)
r
0
(k) =
K
U(k)

n
c
; (6)
from which we see that the photoionization cross-
section is directly proportional to the free carrier
density, n
c
. Here K is a constant equal to
(A(N n) A
m
m)=n, with n and m approximately
constant under the stated conditions.
Using Eq. (6) there are three ways to determine
the change in the photoionization cross-section
with incident photon energy. The rst, proposed
by Grimmeiss and Ledebo [32,33], consists of il-
luminating a sample with light and monitoring the
photoconductivity, which is proportional to the
free carrier concentration, n
c
. By varying the in-
tensity of the light as the wavelength changed (i.e.
varying U(k)) a constant photoconductivity (i.e.
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 33
constant n
c
) may be maintained. The wavelength
dependence of the photoionization cross-section is
then inversely proportional to required photon
ux. The second and third methods in contrast
both maintain a constant photon ux (i.e. U(k)
constant) and follow the variation in n
c
as a
function of wavelength. This can be done by
monitoring either photoconductivity (proportional
to n
c
) or optically stimulated luminescence (pro-
portional to n
c
=s = n
c
A
r
m, where s is the lumi-
nescence lifetime). When using OSL it is necessary
that s remains constant as the wavelength changes.
Under conditions of weak stimulation (Dn n)
this condition holds true.
A comparison of the three methods, and an
indication of the shape of a typical r(k) curve
(actually a r(hm) curve), is shown in Fig. 2. Here
we show the normalized photoconductivity and
OSL stimulation spectra for irradiated (300 Gy)
Al
2
O
3
:C single crystals. The photoconductivity
data were taken either with a constant ux (dotted
line) or the constant photoconductivity (dashed
line) method. The OSL spectrum (full line) was
obtained for a xed emission wavelength of 420
nm [34]. The stimulation range is wide, indicating
a multiple array of trapping sites giving rise to the
photoconductivity and OSL signal. Thus, the data
must be interpreted as a convolution of the
weighted sum of multiple photoionization cross-
sections, with the weights being determined by the
trapped charge populations.
Note that these methods do not yield absolute
values for the photoionization cross-sections, but
only relative values. Absolute values for r, for a
xed stimulation wavelength, can be obtained
from linear-modulation OSL (LM-OSL), to be
described in a later section. Alternatively, one can
use the analysis of Ditlefsen and Huntley [35] who
monitor the decay rate of the OSL as a function of
illumination time and determine r from the ratio
of the OSL intensity to the slope of the OSL decay
curve.
The shape of the stimulation curves shown in
Fig. 2 show an edge-like behavior as a function
of wavelength due to defect-to-band ionization.
The absorption line shape contains contributions
from all relevant continuum states into which the
electron is being excited. The absorption band
will not exhibit a resonance under these circum-
stances [36,37]. Following the transition, the
charge on the defect will change by one electronic
charge and signicant lattice relaxation may oc-
cur. This is also true during the inverse process
[38].
Another example of this type of excitation
spectrum is shown in Fig. 3(a) for OSL from
natural quartz, while Fig. 3(b) illustrates an en-
tirely dierent excitation spectrum shape, from
amelia albite. Resonances are observed in the lat-
ter spectrum, at approximately 1.4 and 2.5 eV. The
resonance at 1.4 eV is consistent with the model of
H utt et al. [28,29] for OSL in these materials in
which the optical absorption at 1.4 eV corresponds
to an internal transition from a ground state to an
excited state, from where thermal stimulation to
the conduction band leads eventually to the pro-
duction of the OSL signal. The calculated thermal
activation energies for the thermal transition are
also indicated in the gure, as a function of stim-
ulation wavelength [39].
Fig. 2. Normalized photoconductivity and OSL stimulation
spectra for irradiated (300 Gy) Al
2
O
3
:C single crystals. The
photoconductivity data were taken either with a constant ux
(dotted line) or the constant photoconductivity (dashed line)
method. The OSL spectrum (full line) was obtained for a xed
emission wavelength of 420 nm. The data were normalized to
unity at a stimulation energy of 4.0 eV (from [34]).
34 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
2.3. Kinetics of bleaching
2.3.1. Bleaching of TL signals
Removal of an electron from a trap reduces the
TL signal that is associated with that trap. The
simplest model that one can envision to describe
this phenomenon is a one-trap/one-recombination-
center model for which we can describe the rate of
detrapping from the trap as
dn
dt
= nf n(N n)A; (7)
with
dn
c
dt
= nf n
c
(N n)A n
c
mA
m
: (8)
Here f is the wavelength-dependent optical exci-
tation rate given by f = Ur. (Under steady-state
conditions, Eq. (8) reduces to Eq. (5).) For the
simplest case of negligible retrapping
(nf n
c
(N n)A) we arrive at
n = n
0
expft; (9)
and one expects an exponential decay of the
trapped charge from its initial level n
0
at t = 0,
with a decay constant of 1=f . For the simple
model described, the TL also decays in a similar
fashion. Eventually, the TL signal must
asymptotically approach zero for long enough il-
lumination. This is true even if retrapping events
become important at long stimulation times, al-
though the decay is no longer exponential. With
retrapping
dn
dt
=
nfmA
m
(N n)A mA
m
; (10)
and with N n; R = A=A
m
; n = m, and (N n)A
mA
m
, we get
n
0
n
= 1 n
0
ft
NR
: (11)
As t , so n 0. An example is shown in Fig.
4, for which we see the initial exponential decay of
the
60
Co-induced 450 K TL peak in Al
2
O
3
:C, with
the decay becoming non-rst-order at longer
bleaching times [40].
Sometimes, however, it is observed that a TL
signal is reduced by prolonged illumination, but
does not decay to zero. A simple interpretation of
this is that the TL signal is in fact made up of
several overlapping signals, some of which are
bleached eciently by the light, and some of which
Fig. 4. The decay of a gamma-induced TL peak at 450 K in
Al
2
O
3
:C for illumination at 540 nm at room temperature (from
[40]).
Fig. 3. OSL stimulation spectra for (a) irradiated natural quartz and (b) irradiated amelia albite. Also shown in (b) are the thermal
activation energies required for OSL production following absorption of photons of the wavelength shown (from [39]).
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 35
are unaected by the light (due to the optical
ionization energy being higher than the photon
energy). However, a more interesting possibility is
that the TL peak is from a single trap level, the
population of which is reduced initially but not to
zero no matter how long the illumination is ap-
plied. An apparent example of this type of be-
havior is shown in Fig. 5 in which the 245C TL
peak in
60
Co-irradiated natural quartz is reduced
to an apparent constant level even after prolonged
illumination at 460 nm [41]. The data obtained can
be described by a rst-order decay plus an un-
bleachable residual.
In order to explain this kind of behavior it is
necessary to consider a model which is consider-
ably more complex that the simple one-trap/one-
center model described so far. Chen et al. [42]
introduced a model in which they allowed for re-
trapping into the electron trap, but also included
the optical excitation of electrons from the re-
combination level into the conduction band during
illumination. The signicance of this latter transi-
tion is that the recombination center lies below the
Fermi level, and thus the concentration of holes at
this center actually increases during the illumina-
tion phase. As a result, it is impossible for the
subsequent TL signal to decay to zero as a result of
the illumination. An important prediction is that
the nal residual level is independent of the origi-
nal absorbed dose, or the intensity of illumination.
In fact the residual level at long illumination times
is the same whether the traps are initially empty or
initially full. Thus, with this model, the illumina-
tion is seen to generate a TL signal in a previously
unirradiated specimen. With respect to the results
for quartz insucient data presently exist to de-
termine if the residual level observed in Fig. 5 is
dose dependent or independent.
An alternative model for the explanation of a
non-zero residual level after bleaching is illustrated
in Fig. 6, after McKeever [43]. Here the TL signal
of interest is due to the release of electrons from
level 2, but the illumination does not bleach this
Fig. 5. The decay of the 240C TL peak from gamma-irradiated quartz with time of illumination with 460 nm light at room tem-
perature (from [41]).
Fig. 6. A model to explain optical bleaching of TL signals in
which a non-zero residual level is reached after long bleaching
times. Three electron traps (1, 2 and 3) and two holes traps
(4 and 5) are involved. The TL signal is due to electrons trapped
in level 2 being thermally released to recombine with holes in
level 4. Level 5 is a non-radiative recombination center. The
optical stimulation only releases charge from level 3 (from [43]).
36 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
level; instead it empties charge from a deep level,
termed level 3. Also included in the model is a
shallow trap (level 1) and two recombination
centers (4 and 5), only one of which (level 4) is
radiative. The important features of the model are
that the charge is not bleached from the level of
interest (level 2) during illumination and thus this
trapped charge concentration does not go to zero
during illumination. In fact, because of photo-
transfer from level 3, it may actually increase
during the illumination process. The reduction in
the TL signal is due to the fact that there are two
recombination centers competing for the charge
released from level 3 during bleaching. Further-
more, during heating, the model of Fig. 5 repre-
sents a classic competition-during-heating
situation in which several behaviors are possible,
depending upon the relative concentrations of the
trapped charges in the various levels. The bleach-
ing process stops when the charge from level 3 is
depleted. At this stage, a residual ``hard-to-bleach''
component remains, such that n
1
n
2
= m
4

m
5
, where n
1
; n
2
; m
4
and m
5
are the con-
centrations of electrons and holes in levels 1, 2, 4
and 5 remaining when the concentration of elec-
trons in level 3 has been depleted. If n
3
is large
enough at the start of the bleaching procedure,
m
4
may be ~0, and thus the nal residual TL
level will also be zero. However, if n
3
is such that
m
4
,= 0, a non-zero residual TL level will be
reached.
2.3.2. Thermal assistance
Often transitions from a defect ground state to
a delocalized band do not take place directly, but
instead occur via a localized excited state. As an
example consider illumination of LiF containing
F-centers. Illumination at a wavelength of 250 nm
is able to transfer electrons from the F-centers to
stable electron trapping states. The transfer is
observed to be temperature dependent, with an
activation energy of 0.12 eV [44]. The F-center in
LiF is believed to have an excited state 0.16 eV
below the conduction band and the measured ac-
tivation energy for phototransfer is suggested to
correspond to thermal activation from this excited
state into the conduction band. Similar mecha-
nisms have been suggested by H utt and colleagues
[28,29] who monitored the OSL that results from
illumination of irradiated feldspar samples with
infrared light (825950 nm). The production of
OSL in this manner was observed to have an ac-
tivation energy of ~0.10 eV (see Fig. 3). Since TL
traps emptied by this illumination are deep the
infrared illumination is believed to excite trapped
electrons from a localized ground state to a lo-
calized excited state, from where thermal excita-
tion into the conduction band (activation energy
0.10 eV) leads to bleaching of the TL traps and
production of OSL.
Poolton et al. [45,46] examined this proposal
further for a variety of feldspars. Using the Bohr
model for the hydrogen atom one can calculate the
energy E
n
and radius R
n
of the nth excited state.
Assuming a hydrogenic model for the defects re-
sponsible for IR-stimulated OSL, the transition
energies to the rst excited state are calculated to
be 1.441, 1.422 and 1.225 eV for orthoclase, albite
and anorthite, respectively. This compares very
well with the measured IR absorption energies
required for production of OSL in these materials
namely, 1:440 0:003 eV; 1:422 0:003 eV and
1:275 0:004 eV, respectively. This gives broad
support to the model proposed by H utt and col-
leagues for thermal assistance of the OSL mecha-
nism. Poolton et al. [45] go on to propose that the
measured thermal activation energy is the hopping
energy from the excited state of the ``donor'' defect
to an ``acceptor'' defect, which acts as a recombi-
nation center. This donoracceptor recombination
model predicts a zero value for the thermal acti-
vation energy if the overlap between the excited
state wavefunctions is high enough, and Poolton
et al. suggest that this explains why the measured
activation energy decreases as one goes from or-
thoclase to anorthite.
Spooner [47] also observed that the OSL from
quartz when illuminated with visible light is de-
pendent upon temperature. In this material, the
measured thermal activation energy for OSL pro-
duction depends smoothly upon the wavelength of
the light used. This is inconsistent with a H utt-type
mechanism for which one would expect that either
the thermal activation energy would be indepen-
dent of the wavelength, or it would vary discretely
as higher excited states are populated. Because of
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 37
this Spooner suggests a model based on the ther-
mal excitation of charge within an array of ground
state levels, from where optical excitation into the
conduction band then occurs.
2.4. Phototransfer
2.4.1. Phototransferred TL
PTTL results when charge is optically stimu-
lated out of deeper traps and transferred to shal-
lower traps, thereby increasing or (if the shallow
traps were initially empty) creating a TL signal due
to the shallow traps. The size of the PTTL signal is
dependent upon the initial absorbed dose (which
was used to populate the deep traps) and the pa-
rameters of the optical stimulation (intensity,
wavelength, duration and temperature). Several
applications of this phenomenon have been sug-
gested in dosimetry, including a second reading of
the initial absorbed dose. Heating a TLD material
immediately after irradiation leads to a TL signal
which is proportional to the dose, but this signal
cannot be re-read since the readout mechanism is
``destructive''. However, by optically transferring
stable charge from deep traps into the now-empty
shallower traps, one can induce a second TL signal
that, for xed stimulation conditions, is propor-
tional to the initial dose. Thus, one has the ability
to ``re-read'' TL signals after the rst TL reading.
Example applications can be found in [48,49].
A second application uses a xed initial radia-
tion dose and uses the PTTL signal to provide an
estimate of how much optical exposure there has
been. This approach has been used successfully for
UVB dosimetry [50] in which the stimulation light
used to activate the phototransfer is centered in the
UVB wavelength range.
2.4.2. PTTL models
In general, PTTL is useful for examining the
optically stimulated charge transfer processes that
occur between trapping centers in OSL materials
during illumination. As such, the models to de-
scribe PTTL are informative for understanding
OSL processes. The simplest model necessary to
produce PTTL is that of one deep trap, one shal-
low trap, and one recombination center. (These
are the minimum centers required for the eect to
occur.) If n
1
and n
2
are the shallow and deep
electron trap populations, respectively, and m is
the concentration of holes in recombination cen-
ters, then, at some point after irradiation and im-
mediately at the start of the illumination period,
one might have n
10
= 0 and n
20
= m
0
. The illumi-
nation is then assumed to excite electrons from the
deep trap (the ``donor trap'') at a rate f into the
shallow trap (the ``acceptor traps''), such that we
may write
dn
2
dt
= n
2
f n
c
(N
2
n
2
)A
2
; (12a)
dn
1
dt
= n
c
(N
1
n
1
)A
1
(12b)
and
dm
dt
= n
c
mA
m
=
m
s
; (12c)
where s = (n
c
A
m
)
1
is the recombination lifetime
and all other terms have their usual meaning. At
quasiequilibrium (dn
c
=dt dn
i
=dt; dm=dt; i = 1
or 2) and no retrapping into the deep trap during
illumination the solution for the growth of n
1
as a
function of illumination time t is
n
1
(t) = N
1
[1 expBt[; (13a)
(where B = n
c
A
1
), and both n
2
and m decay expo-
nentially, thus
n
2
(t) = n
20
exptf (13b)
and
m(t) = m
0
expt=s: (13c)
When considering the warming of the sample after
an illumination period t
+
one has to take into ac-
count the competition for the released charges to
either retrap in empty deep traps (concentration
N
2
n
2
(t
+
)), or recombine with trapped holes
(concentration m(t
+
)). Chen and McKeever [51]
demonstrate that under these conditions the PTTL
signal is given by
S
PTTL
(t
+
) =
C expt
+
=sN
1
[1 expBt
+
[
[N
2
=n
20
expt
+
f [
;
(14)
38 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
where C is a constant. Eq. (14) describes a
monotonically increasing function i.e. the PTTL
signal grows with illumination time until either the
donor traps are depleted or the acceptor traps
become full. The maximum PTTL signal (at
t = ) is then S
PTTL
() = Cn
2
1
=N
2
. An example
of this type of behavior is given in Fig. 7(a).
However, one often nds cases in which the
PTTL intensity increases to a maximum, and then
decreases for prolonged illumination times. An
example of this type of behavior is shown in
Fig. 7(b) and results from the simultaneous lling
and emptying of the shallow trap during illumi-
nation. For the example shown in this gure
(Al
2
O
3
:C) the wavelengths used for the PTTL (307
nm) are known to not only empty charges from the
deep, donor traps, but also to empty them from
the shallower, acceptor traps. As a result of the
competition between trap lling and trap empty-
ing, the PTTL signal reaches a maximum and then
decays (eventually decaying to zero).
A more unusual behavior is illustrated in Fig. 8
(from [52]) in which we show the PTTL curves
from a shallow trap in quartz. Here the PTTL
increases, then decreases, but does not obviously
decay to zero at long illumination times. In an
earlier section we discussed a model in which one
could get bleaching of a TL signal but only to a
Fig. 7. Examples of PTTL versus illumination time. (a) PTTL from MgO:Cu, with the PTTL glow curve shown in the inset. (b) PTTL
from Al
2
O
3
:C, with the PTTL glow curve shown in the inset.
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 39
particular level, below which no further bleaching
takes place. A similar model can be used to explain
the data of Fig. 8. Referring to the model of Fig. 6,
level 4 represents a radiative center and recombi-
nation of electrons with holes at this center pro-
duces OSL during the illumination period.
However, level 5 is a competing non-radiative
center and electrons recombining here are lost
from both PTTL and OSL. The PTTL signal is a
result of electrons in the acceptor trap (level 1)
recombining with holes in the radiative center
(level 4). It is important to realize, therefore, that
the PTTL signal will follow the smaller of these
two concentrations, i.e. S
PTTL
= min(n
1
; m
4
). Thus,
even though the electrons in level 1 may be in-
creasing during illumination, the number of
available holes in level 4 is decreasing and, de-
pending on the initial conditions and the relative
concentrations in the various levels, one can arrive
at the situation where the electron concentration in
the acceptor trap (n
1
) is greater than the hole
concentration in the radiative center (m
4
). Thus,
after an initial increase, the PTTL due to level 1
will start to follow m
4
and will decrease with illu-
mination time. At very long times, one can deplete
the charge in the donor traps such that the nal
PTTL curve will be characterized by an increase,
followed by a decrease, followed by a stable (non-
zero) level.
This simple ``hand-waving'' argument, how-
ever, is too simplistic even for a relatively simple
model such as that shown in Fig. 6. Alexander and
McKeever [53] analyzed this and several other
models in detail and solved the simultaneous dif-
ferential equations numerically without introduc-
ing assumptions (such as quasiequilibrium, or no
retrapping). This analysis indicated that a variety
of PTTL-versus-time curve shapes are possible,
including shapes such as those illustrated in Figs. 7
and 8. The key parameters, for a particular model
and illumination conditions, are the initial con-
centrations in the various traps at the start of the
illumination period, and the photoionization
cross-sections at the wavelengths used. The im-
portant points illustrated by these analyses are
that:
1. A decrease in the PTTL signal at long illumina-
tion times does not necessarily mean that there
is simultaneous optical excitation out of the ac-
ceptor trap.
2. In all cases a steady-state level will eventually be
reached at long illumination times. The steady-
state value may be zero, or non-zero, depending
upon the relative sizes of the donor trapped
electron concentration and the radiative center
concentration at the start of the illumination pe-
riod.
2.5. Wavelength dependence
In the above discussions of PTTL and optical
bleaching of TL an important factor is the exci-
tation rate from the trap, f. This parameter is re-
lated to the wavelength k and intensity U of the
illuminating light by f = Ur, where r is the al-
ready-dened photoionization cross-section,
which is related to the wavelength of the excitation
light through Eq. (3a). Not all TL peaks are
bleached during illumination with a given wave-
length. Similarly, the shape of the PTTL-versus-
illumination time curve is also dependent on the
wavelength of the light used.
Fig. 9 illustrates the glow curve from a sample
of natural quartz without (NTL) and with
Fig. 8. PTTL versus 325 nm illumination time at room tem-
perature and at 82 K for the 386 K PTTL peak in quartz. The
inset shows the behavior of the 270 K peak at 100 K for 24 h
(from [53]).
40 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
(NTL+bleach) bleaching using a solar simulator.
(NTL is the natural thermoluminescence signal.)
The dierence curve reveals the TL peak that is
removed from the glow; the remainder is little af-
fected by the bleach. A small PTTL peak appears
at ~180C. Clearly, with the optical spectrum used
in these studies, not all traps are equally stimulated
by the optical wavelengths employed. One can
expect a result like this by considering the
photoionization cross-section and its dependence
on wavelength. A typical result has already been
described in Fig. 2. For a given trap, with a par-
ticular optical threshold energy, the value of r
increases with wavelength. For a given wave-
length, the photoionization cross-section value will
be dependent upon the optical threshold energy,
such that deeper traps will have a smaller r. In-
deed, for certain wavelength ranges r may be 0 for
the deep traps and >0 for the shallower traps.
Hence, we can expect optical bleaching of lower
temperature TL peaks but not (or less so) from
deeper traps.
This is observed more easily in Fig. 10 in which
we show the bleaching spectrum for the shallower
``325C'' trap and the deeper ``370C'' and ``480C
traps'' from natural quartz (where the trap labels
are dened according to the temperatures of the
corresponding TL peaks at a heating rate of 5C/
s). We observe the energies required to bleach the
325C TL peak completely, and to bleach the
370C and 480C peaks to 80% of their initial
values. Also shown is the energy required to pro-
duce 5%, 50% and 90% of the corresponding OSL
signal. The similarity of the latter with the
bleaching curve for the 325C peak led Spooner
et al. [54] to propose that the OSL signal emerges
from the stimulation of charge from the 325C
trap, and not from the deeper traps, over this
wavelength range. These assertions were conrmed
in later studies [55].
A bleaching spectrum for TL from Al
2
O
3
:C is
shown in Fig. 11, expressed as the percentage of
TL lost after bleaching an irradiated sample with a
constant photon uence at the wavelengths shown.
Three data points are shown for comparison.
These are extracted from the rate of the depletion
of the TL signal, at three dierent illumination
wavelengths i.e. they are the normalized slopes of
the decay curves of the type shown in Fig. 4.
However, one has to be careful when interpreting
Fig. 10. The bleaching spectrum for TL from natural quartz,
expressed as the energy required to deplete the TL peak to a
certain level. The curves are compared with the energy required
to produce OSL signals form the same material (from [54]).
Fig. 9. Glow curves from natural and bleached quartz measured
at a heating rate of 10 K/s. The initial TL peak (``NTL'') is
modied after bleaching with a solar simulator
(``NTL+bleach''). The dierence in the two glow curves yields
the single peak (the ``325C peak'') that was removed by the
bleach. The remainder of the glow curve is unaected by the
bleach. A small PTTL peak is observed at low temperatures
(~ 180C) after bleaching (from [47]).
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 41
data of the type shown in Fig. 11 as being a re-
ection of the wavelength dependence of the
photoionization cross-section r(k). This is because
simultaneously with bleaching of the charge out of
the traps, one can also have phototransfer of the
charge into the traps, from deeper traps. That is,
the percentage TL lost is actually given by an ex-
pression of the type: %TL
lost
(k) = TL
bleached
(k)
TL
transferred
(k). Clearly, the value of % TL
lost
is then
not only a function of the wavelength, but is also a
function of the extent to which the deep traps are
lled. This is illustrated in Fig. 12 in which we
show the %TL
lost
(k) curves as a function of the
pre-dose given to the sample. For the larger doses
and shorter wavelengths the percentage ``lost'' is
actually negative i.e. the TL peak actually in-
creased in size during ``bleaching''. This illustrates
again that multiple transfer processes occur during
optical stimulation of an irradiated sample and the
behavior depends on the details of the trapped
charge distribution at the start of the illumination
period.
Experimental PTTL-versus-time curves for
PTTL from quartz obtained at three dierent
wavelengths are shown in Fig. 13. Excitation with
shorter wavelength light leads to the maximum of
the PTTL peak occurring at a shorter wavelength,
and a faster decay of the PTTL signal at longer
times. By numerically solving the rate equations
for various models for PTTL, Alexander and col-
leagues [52,53] demonstrated that this was a con-
sequence of an increase in the excitation rate f,
through a decrease in the wavelength k. The
wavelength dependence of the phototransfer pro-
cess can be extracted from these data by measuring
the initial slope of the curves in Fig. 13 as a
function of wavelength. Such an analysis is shown
Fig. 12. Percentage TL ``lost'' from the Al
2
O
3
:C glow curve as a
function of illumination wavelength for three dierent values of
pre-dose given to the sample. For the larger pre-doses charge
transfer from deeper traps predominates at short wavelengths
and a growth in the TL signal (i.e. % TL
lost
< 0) is observed
(from [40]).
Fig. 13. The 358 K PTTL peak in quartz as a function of
illumination time for three dierent wavelengths (from [52]).
Fig. 11. TL bleaching spectrum for Al
2
O
3
:C as a function of
wavelength (from [40]).
42 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
in Fig. 14 where we show the data extracted for the
270 K PTTL peak. The experimental data have
been tted to a summation of three curves of the
type given by Eq. (3a), namely
S(k) =
X
3
i=1
K
i
r
i
(k)
=
X
3
i=1
C
i

E
0i
p
(hm E
0i
)
3=2
hm(hm gE
0i
)
2
; (15)
where K
i
and C
i
are constants, and S represents the
normalized initial slope of the data of Fig. 13. In
this case, the analysis suggests that three donor
traps contribute to the PTTL signal, with optical
threshold energies (E
0i
; i = 1; 2; 3) of 2.37, 2.84
and 3.77 eV, with g = 0:559 for quartz [52].
3. Optically stimulated luminescence
3.1. Stimulation modes
The basis of OSL measurement is to stimulate
an irradiated sample with light of a selected
wavelength and to monitor the emission from the
sample at a dierent wavelength. A number of
modes of stimulation are available, known as
continuous wave-OSL (CW-OSL), linear-modula-
tion OSL (LM-OSL) and pulsed OSL (POSL).
1. CW-OSL: The traditional way is simply to illu-
minate the sample with a constant intensity
source and to monitor simultaneously the lumi-
nescence emission during stimulation. Termed
CW-OSL, one uses a laser or a broadband
source and monochromator (or lter) to select
a particular stimulation wavelength. The lumi-
nescence emission is monitored continuously
while the stimulation beam is on and narrow
band lters are used to discriminate between
the excitation light and the emission light, and
to prevent scattered stimulation light from en-
tering the detector. The OSL is monitored from
the instant that the stimulation source is
switched on and is usually of the form of an ex-
ponential-like decay until all the traps are emp-
tied and the luminescence ceases. The integrated
emission (i.e. the area under the decay curve,
less background) is recorded and is used to de-
termine the dose of absorbed radiation.
In several cases the decay curve may be exactly
exponential, or may be the sum of several expo-
nentials. In other instances the decay is clearly
non-exponential and, indeed, in some cases the
OSL signal is observed to increase initially before
decaying. This wide variety of OSL decay curve
shapes suggests a multiplicity of possible recom-
bination pathways and processes for the produc-
tion of OSL, some of which are discussed below.
2. LM-OSL: Instead of illuminating the sample
with a constant intensity light source, one ar-
ranges for the intensity of the stimulation light
to increase linearly with time. One then pro-
duces what is known as linear-modulation
OSL, or LM-OSL. The OSL output is observed
to increase, initially linearly as the stimulation
power increases until the traps become depleted,
after which the OSL intensity decreases non-lin-
early to zero. Thus, the OSL signal is in the
shape of a peak, the position of which (in time)
depends upon the rate of linear increase in in-
tensity of the stimulating light and the photoi-
onization cross-section of the trap being
emptied. For a given ramp rate and stimulation
wavelength, traps with dierent photoionization
cross-section values thus appear at dierent
Fig. 14. Excitation spectra for the 270 K PTTL peak in natural
quartz. The data are extracted from the initial slopes of the data
shown in Fig. 13 and have been tted to a summation of three
photoionization cross-sections (Eq. (3a)) with threshold ener-
gies of 2.37, 2.84 and 3.77 eV (from [52]).
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 43
times, and the method is thus able to distinguish
between OSL originating from dierent traps.
3. POSL: Pulsed OSL results when the stimulation
source is pulsed at a particular modulation fre-
quency and with a particular pulse width appro-
priate to the lifetime of the luminescence being
observed. In this mode of excitation one mea-
sures the OSL emission only between pulses,
rather than during the pulses. In this way, dis-
crimination between the excitation light and
the emission light is achieved by time resolution,
rather than wavelength resolution. Less optical
ltration is necessary with this method than
with the CW-OSL or LM-OSL methods and,
for the right timing parameters, a high OSL col-
lection eciency may be achieved.
3.2. OSL models
3.2.1. CW-OSL
In general, the shape of a CW-OSL decay curve
is not an exponential and does not lend itself to
simple description. Overall, the CW-OSL decay
curve shape is dependent upon the sample, the
absorbed dose, the wavelength of stimulation, the
stimulation intensity and the sample temperature.
Example decay curves are shown in Fig. 15 for a
variety of materials and under a variety of condi-
tions. Attempts to t the decay curve to a single
exponential usually prove inadequate. Several au-
thors have described the decay curve as the sum of
multiple exponentials (e.g. OSL in quartz [56]).
Others have described the decays with more gen-
eralized, empirical functions. For example, Baili
and Poolton [57] described the OSL decay from
several feldspars by a A(1 Bt)
P
law, where A; B
and P are constants. For microcline, sanadine and
albite, A is temperature dependent and P = 1; for
labradorite A = 1 and P = 2. Often a long, non-
zero (i.e. above background) tail is observed. This
tail can be caused by either (or both) shallow traps
(in which the stimulated charge becomes tempo-
rarily trapped and slowly released over the time
scale of the experiment) or deep traps (for which
the photoionization cross-section is small at the
wavelengths used in the measurement and trap
emptying is slow).
Fig. 15. Example CW-OSL decay curves from a variety of
materials and under dierent conditions. (a)(c) OSL from
dierent quartz samples (natural and synthetic) (from [63]).
(d) OSL from Al
2
O
3
:C at dierent temperatures (from
[58]).
44 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
In many cases there is no model-based analyt-
ical function that can describe the overall shape of
the CW-OSL decay curve. Nevertheless, the vari-
ety of curve shapes and behaviors exemplied in
Fig. 15 can often be predicted through numerical
solutions of dierential equations used to describe
the charge transfer processes occurring on the
basis of simple models. Through an examination
of these models the contributions to the process of
both shallow traps, deep traps and non-radiative
centers can be elucidated and understood. Most of
the models assume the transport of the optically
stimulated charge via the conduction band and
some of the more simple ones are illustrated in Fig.
16, starting with the simplest of all Fig. 16(a).
With the usual denitions of terms we write that at
the start of the illumination period (after irradia-
tion) n
0
= m
0
. The rate equations describing the
ow of charge during CW-OSL production are
then
dn
c
dt
=
dn
dt

dm
dt
; (16a)
I
OSL
=
dm
dt
=
dn
dt
= nf ; (16b)
where we assume quasiequilibrium (dn
c
=dt dn=
dt; dm=dt; n
c
n; m). Charge neutrality during
illumination dictates that n
c
n = m.
Under these assumptions the solution is
I
OSL
= n
0
f exptf = I
0
expt=s; (17)
where I
0
is the initial CW-OSL intensity at t = 0
and s = 1=f is the decay constant. One observes a
straightforward exponential decay in which the
initial intensity I
0
is directly proportional to the
stimulation intensity U and the photoionization
cross-section r (where f = Ur); and the decay
constant is proportional to 1=Ur.
For p optically active traps (concentrations n
i
,
with i = 1; . . . ; p) the solution is the sum of mul-
tiple exponentials, thus,
I
OSL
=
X
p
i=1
I
i0
expt=s
i
: (18)
Alternatively, if we use the model of Fig. 16(b) in
which the additional traps act as competing traps
only we have
I
OSL
= n
10
f exptf
dn
2
dt
; (19)
where n
01
is the initial concentration of electrons
trapped in the trap from which the charge is being
stimulated and n
2
is the concentration of electrons
in the competing trap. If the latter do not empty
either optically (at the wavelength used) or ther-
mally (i.e. deep traps), then
dn
2
dt
= n
c
(N
2
n
2
)A
2
: (20)
With the additional constraint that N
2
n
2
, then
dn
2
=dt ~ C, where C is a constant and thus the
CW-OSL intensity is reduced by the extent of the
retrapping into the deep, competing traps. As
t , n
c
0 and so C 0. Thus, C is in fact a
very slowly varying function of time and gives rise
to a long, weakly temperature-dependent tail in
the CW-OSL curve.
If the competing traps are shallow traps (Fig.
16(c)) which are thermally unstable at the tem-
perature (T) of the CW-OSL measurement, then
Eq. (20) becomes
Fig. 16. Simple models for OSL. (a) Simplest model; involving
one trap and one radiative recombination center. (b) Model
containing an additional deep, competing trap. (c) Model
containing a shallow, competing trap. (d) Model containing a
competing non-radiative recombination center.
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 45
dn
2
dt
= n
c
(N
2
n
2
)A
2
n
2
p; (21)
where p is the thermal excitation rate out of the
shallow trap, given by p = s expE=kT. Here E
and s are the usual thermal trap depth and fre-
quency factor, respectively, and k is Boltzmann's
constant. The CW-OSL is now
I
OSL
= n
10
f exptf n
c
(N
2
n
2
)A
2
n
2
p;
(22)
and the last two terms give rise again to a long-
lived, but strongly temperature-dependent tail in
the OSL curve. Furthermore, these two terms may
give rise to an initial increase in the CW-OSL
signal, depending on the relative size of these terms
compared with the rst term. An example of such
a shallow-trap-related increase was shown in Fig.
15(d). The relative sizes of these terms also depend
on the excitation rate f, such that at low f (i.e. low
values of U and/or r) the temperature-dependent
term may be signicant.
The nal simple model, illustrated in Fig. 16(d),
deals with one trap but two recombination centers
(of concentrations m
1
and m
2
), one of which is
non-radiative. The CW-OSL signal is now
I
OSL
= n
0
f exptf
dm
2
dt
: (23)
Again, the OSL intensity is reduced due to the
presence of a non-radiative pathway. The relative
size of the two recombination centers is time de-
pendent, thus
m
1
m
2
~
m
10
m
20
exptn
c
[A
m1
A
m2
[: (24)
However, if A
m1
~ A
m2
then the ratio remains ap-
proximately constant. In this case the CW-OSL
decay curve remains approximately exponential
according to
I
OSL
=
1
K
n
0
f exptf ; (25)
where K is a constant given by K = (m
1
m
2
)=m
1
.
In each of the above analyses retrapping into
the optical sensitive trap is not allowed. If we allow
for retrapping, then even the simple one-trap/one-
center case becomes non-exponential, according to
I
OSL
= I
0
1


n
0
ft
N(A=A
m
)

2
: (26)
A more general view may be obtained by numer-
ically solving the rate equations describing the
charge ow. Thus, McKeever et al. [58] examined
various solutions to the dierential equations for a
model which contained one optically sensitive
trap, a shallow trap, a deep disconnected trap, one
radiative recombination center and one non-radi-
ative recombination center. Example solutions to
the numerical equations are illustrated in Fig. 17
for various values of temperature, excitation in-
tensity and initial dose.
A real material may have combinations of the
above simple conditions prevailing. It can be seen,
therefore, that, in general, one should not expect a
simple exponential decay, which should be viewed
as an exception, rather than the rule.
3.2.2. LM-OSL
Bulur [59] introduced a new readout method for
OSL in which he linearly increased the intensity of
the stimulation light, producing a series of OSL
peaks as a function of readout time. Each peak
corresponded to a dierent trap, with traps with
the largest photoionization cross-section emptying
rst. If the excitation rate is increased according to
U(t) = ct (or, f (t) = rU(t) = rct where c is the
rate of increase in the stimulation intensity, i.e. the
``ramp rate''), then at quasiequilibrium and with
no retrapping, we have
I
OSL
=
dn
dt
= n
0
rct exp

rct
2
2

; (27)
from which we see that the OSL curve shape starts
from zero at t = 0 and exhibits a maximum at
t = t
m
=

1=rc
p
. Thus, from knowledge of the
absolute ramp rate c, one can determine the
photoionization cross-section r at the wavelength
used in the experiment. Note that if one uses
broadband stimulation light there will be a distri-
bution of r values and a corresponding spreading
(widening) of the LM-OSL peak for any given
trap. Fig. 18 shows a series of theoretical LM-OSL
46 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
curves for various ramp rates using Eq. (27). The
LM-OSL technique has been applied to a variety
of materials, including semiconductors [60], and
quartz [6163] and sapphire [64]. We note once
again that the position of the LM-OSL peak is
dependent upon the photoionization cross-section.
Thus, the order of the LM-OSL peaks does not
necessarily match that of the corresponding TL
peaks, which are ordered in order of increasing
thermal stability. This is illustrated in data for
quartz by Bulur et al. [62], shown in Fig. 19, in
which are presented the LM-OSL curves for irra-
diated quartz after post-irradiation annealing at
the temperatures indicated. It is clear that the
second LM-OSL peak has been removed after
heating to 125C. During the heating the so-called
110C TL peak of quartz was produced. The rst
LM-OSL peak, however, corresponds to the 325C
TL peak of quartz. Thus, the thermally unstable
``110C trap'' has a smaller photoionization cross-
Fig. 19. LM-OSL curves for irradiated quartz following post-
irradiation annealing at the temperatures indicated (from [62]).
Fig. 17. (a) Simulated CW-OSL curves using the model de-
scribed in the text, at a variety of temperatures. (b) Behavior of
the predicted decay curves for variations on the stimulation
intensity. (c) Behavior as a function of absorbed dose. Mc-
Keever et al. [57] describe the model in more detail, including
the parameter values used in the calculations.
Fig. 18. Calculated LM-OSL curves for various values of the
ramp rate (denoted here as g) using Eq. (27).
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 47
section than the deeper ``325C trap'' at the
wavelength used in this experiment.
It is interesting to compare CW-OSL curves
and LM-OSL curves from the same sample. Fig.
20 shows the ts of a summation of three expo-
nentials to CW-OSL from a quartz sample, and
the corresponding ts of the LM-OSL data to
three peaks, assuming no retrapping. Both data
sets t well. However, we note here that if one
arranges the LM-OSL experiment such that U(t)
reaches the value U used in the CW-OSL experi-
ment at time t
f
(i.e. U
LM
(t
f
) = U
CW
) then we have
s = 1=rct
f
, and s = t
2
m
=t
f
. From the obtained CW-
OSL and LM-OSL data Kuhns et al. [63] show
that the derived parameters are not consistent with
this simple relationship (i.e. s ,= t
2
m
=t
f
) suggesting a
more complex model than a summation of three
rst-order exponentials, despite the excellent t to
the data.
3.2.3. POSL
Mathematical models for the kinetics of POSL
are the same as those for CW-OSL. The POSL
technique is illustrated in Figs. 21(a) and (b). A
stream of pulses is incident on the sample. The
pulse widths are chosen so as to be less than the
intrinsic lifetime of the luminescence center [24].
The luminescence detector is gated so that it only
records the luminescence between the pulses, ra-
ther than during the pulses. Since the excitation
pulse width is less than the lifetime of the lumi-
nescence most of the OSL signal due to that pulse
emerges after the pulse. By using a series of pulses
and collecting the OSL signal only between each
pulse a strong OSL signal is built up. The timing of
Fig. 20. Examples of ts of LM-OSL curves (a) and CW-OSL
curves (b) in the same quartz sample (from [63]).
Fig. 21. (a) Schematic of the principle of the POSL technique on
which a series of optical stimulation pulses are incident on the
sample and the OSL is recoded between the pulses. (b) A detail
of the timing sequence. The key issue is to have the pulse width
less than the luminescence lifetime s. T
1
is the pulse width
(FWHM); T
2
is the period during which the OSL detector is
gated o; T
3
is the period during which the OSL is recorded. If
s T
3
, the luminescence during the measurement period is
approximately constant. Thus, by the time the next pulse arrives
(a) the luminescence from the rst pulse is still there, and the
two signals add. This repeats until such time as equilibrium
between the build-up of excited luminescence centers and the
decay of said centers occurs. Sometime after this the POSL
signal will decay slowly as the trapped charge is depleted (from
[24]).
48 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
the light collection sequence is illustrated in the
gure. Since the light collection is out of phase
with the stimulation, discrimination between the
two is achieved via time resolution rather than
wavelength resolution (as in CW-OSL and LM-
OSL). Thus, the need for lters is reduced and
signicant light collection eciency is possible.
So far the technique has been demonstrated
only for OSL from Al
2
O
3
:C for which the lumi-
nescence lifetime is long-lived (s = 35 ms at room
temperature) [2224]. In principle, however, other
materials can be used, but the timing sequence
needs to be ``tuned'' to the intrinsic response time
(specically, the luminescence lifetime) of the sys-
tem. An example of collected POSL data is shown
in Fig. 22 (from [65]) in which the POSL from
three samples of Al
2
O
3
:C is shown. Included in the
gure are data for samples normally used in TL
dosimetry, and samples containing either large or
small concentrations of shallow traps. If there are
large concentrations of shallow traps a signicant
fraction of the stimulated charge is retrapped into
these centers giving rise to a slow phosphorescence
component and a slowing down of the POSL re-
sponse. It is this latter component that is exploited
in Delayed OSL (DOSL) measurements, as noted
in Section 1. The best POSL material does not
have shallow traps, and the decay of the POSL
signal after the last pulse is characterized only by
the 3536 ms lifetime associated with the lumi-
nescence centers in this material (F-centers).
As the period of optical stimulation increases,
so the POSL signal decreases. The kinetics de-
scribing the shape of the depletion curve are the
same as those discussed above for the CW-OSL
decay curve. Similar models apply. A typical de-
pletion curve is shown in Fig. 23 where it is in-
teresting to observe that the depletion rate is dose
independent.
4. Radiation dosimetry using OSL
OSL has found application in three areas of
radiation dosimetry, namely:
1. retrospective dosimetry (the dosimetry of past
exposures of materials to radiation);
2. personal dosimetry;
3. environmental dosimetry.
Examples of the rst two are illustrated in the
following sections. Examples of the last (environ-
mental dosimetry) are much fewer and in any case
involve only the use of synthetic dosimeters
Fig. 22. Example POSL data from three samples of irradiated Al
2
O
3
:C. The stimulation is a 4 kHz and is incident on the sample for 1 s
(i.e. 4000 pulses). The OSL signal decays after the last pulse at a rate that depends upon the concentration of shallow traps and the
luminescence lifetime. Shallow traps in the samples give rise to a strong phosphorescence component in sample 3. The best sample for
POSL is one that contains few shallow traps i.e. sample 2 (from [65]). The timing parameters (refer to Fig. 21(b)) used in the
demonstrated measurements were T
1
= 300 ns, T
2
= 15 ls and T
3
= 235 ls, with s = 35 ms for Al
2
O
3
:C.
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 49
(notable Al
2
O
3
:C) with the OSL measured using
either the CW-OSL mode or the POSL mode.
4.1. Retrospective dosimetry
Examples of this fall in two major categories,
although the techniques used are identical. The
categories are:
1. Luminescence dating, in which one wishes to es-
tablish the dose delivered to natural materials
during exposure of the materials to natural radi-
ation over the lifetime of the sample in a certain
environment.
2. Accident dosimetry, in which one uses OSL
from natural or locally available materials to re-
construct the dose imparted to an area during a
radiation accident. The basic premise in each of
these is to measure the amount of luminescence
emitted from a ``natural'' sample and, through
calibration procedures, determine the dose of
absorbed radiation that gave rise to that lumi-
nescence signal. CW-OSL is the technique most
often used in these applications, and the latest
developments in the eld can be found in the
proceedings of the luminescence dating confer-
ences [25].
There are several methods used which dier in
important detail. Basically, the dierences are in
whether one uses multiple aliquots (i.e. one ali-
quot for each irradiation and OSL measurement)
or the same sample for all irradiations and
measurements. These are known as ``multiple-
aliquot'' and ``single-aliquot'' techniques, respec-
tively. Furthermore, one can choose between
adding calibration doses to the natural dose
(`additive dose techniques') or reading the natural
signal rst and regenerating the OSL by applying
calibration doses afterwards (`regeneration tech-
niques'). The most recent developments in this
eld have centered upon single-aliquot regenera-
tion (SAR) procedures and these will be briey
described here. For a more detailed description,
see Wintle [66].
In the SAR method those minerals which are
the focus of the study are extracted from the nat-
ural sample. Normally, these are quartz or feldspar
grains. Extraction is done using various acid wa-
shes, magnetic separation and sieving procedures.
A small aliquot of this material is then stimulated
with light of the appropriate wavelength. For
quartz (the most popular mineral studied in this
way) green or blue light is normally used, and the
CW-OSL signal from the sample is monitored.
The basic procedure is then to give the same
sample a known dose from a calibrated radia-
tion source in the laboratory and to record the
Fig. 23. Decay of the POSL signal due to repeated use. The depletion is shown here in terms of the number of 1 s measurements made
on the sample. The dierent symbols represent dierent absorbed doses, ranging from 10 mrad to 38 rad (from [24]).
50 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
CW-OSL signal again. This is repeated with sev-
eral calibration doses and an OSL-versus-dose plot
is constructed. By comparing the measured OSL
from the natural sample with this curve, an esti-
mate of the dose absorbed by the sample in its
natural environment is made (Fig. 24).
However, the process of re-using the same
sample over and over leads to changes in the
sensitivity (dened as the OSL output in response
to a small test irradiation) such that if the same
dose is administered twice, the OSL output is not
the same. Generally, it is larger, representing a
sensitivity increase. Such changes in sensitivity can
be caused by many factors but they should not be
surprising considering the complex charge transfer
processes that occur during optical stimulation in
materials that have many trap and recombination
centers. From the models for OSL discussed in the
previous sections it is clear that the response of a
material to radiation is sensitively dependent upon
many factors, including the concentrations of the
various traps and centers, and the population of
charge in those traps and centers at the start of the
next irradiation and OSL measurement. Changes
to these as a result of previous irradiation and
readout of the OSL may lead to sensitivity chan-
ges. Sensitivity changes in the SAR process were
modeled by McKeever et al. [67] and experimen-
tally studied by, for example, Richardson [68]. A
variety of eects were shown to be possible which
can lead to sensitivity changes, and to subsequent
non-linearities in the OSL-versus-dose response,
along with a concomitant error in the estimation
of the natural dose.
Thus, the SAR procedure needs to test and
account for sensitivity changes and the procedures
adopted to do this do so by interspersing the ir-
radiations and OSL measurements with test irra-
diations and measurement of the OSL due to these
test doses. The details are summarized by Wintle
and Murray [69].
Since the ultimate goal is to use the OSL signal
as a measure of the length of time the sample has
been irradiated in nature, it is also necessary that
the OSL signal used is stable over the natural
lifetime of the sample (from as short as a few tens
of years to >10
5
years). To do this various preheat
procedures are adopted to remove unstable sig-
nals. In these procedures, the specimen is heated to
a dened temperature immediately after
irradiation but prior to readout of the OSL signal
[70].
The result of adopting these preheat and sen-
sitivity checks is that the actual SAR procedure
can be quite complex. A summary of a typical
procedure is given in Table 1. Example results are
Fig. 24. Schematic of the method of determining the natural
absorbed dose through calibration of the sample in the SAR
method. For more details see Wintle [66].
Table 1
Steps required in the production of a sediment date using the
SAR method
(a) Obtain sample
(b) Magnetically separate non-magnetic components
(c) Sort by sample grain size
(d) Preheat to an appropriate temperature (T
1
ph
) to remove
unstable components
(e) Stimulate sample with light; read natural luminescence
(L
n
)
(f) Irradiate with a small ``test dose'' (D
t
)
(g) Pre-heat to T
2
ph
(h) Stimulate with light to read luminescence (T
n
)
(i) Calculate ratio (L
n
=T
n
)
(j) Give sample rst calibration dose (D
1
)
(k) Preheat (T
1
ph
)
(l) Read luminescence L
1
(m) Repeat test dose
(n) Preheat (T
2
ph
)
(o) Read luminescence T
1
(p) Calculate L
1
=T
1
(q) Repeat for three more calibration doses
(r) Plot L
i
=T
i
ratios against calibration doses D
i
(s) Using least squares tting compare natural signal
L
n
=T
n
against calibration curve; obtain natural
equivalent absorbed dose D
n
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 51
illustrated in Fig. 25 for a wind-blown sedimentary
sample.
4.2. Personal dosimetry
Fig. 26 shows the POSL signal versus absorbed
dose for samples of Al
2
O
3
:C stimulated as de-
scribed above. The light output is the total inte-
grated output summed over 4000 laser pulses for a
stimulation period of 1 s. For doses greater than
3 10
2
Gy a ``weak'' stimulation beam (0.01 W)
was used in the measurement. For all other doses a
``strong'' beam (1.2 W) was used [24]. For the ar-
rangement described the minimum measurable
dose (dened as three times the standard deviation
of the background signal from an unirradiated
sample) was 5 lGy. The dose response is seen to
be linear over almost seven orders of magnitude.
Additional utility lies in the ability to re-mea-
sure the POSL signal. Since the sensitivity of the
technique is very high, it is only necessary to de-
plete a fraction of the trapped charge in any one
measurement. Thus, by simply re-exposing the
sample to the laser beam, and taking into account
the fraction of trapped charge depletion from the
previous measurement (or measurements) see
Fig. 23 one can arrive at a second (or third, etc.)
independent estimation of the dose. The standard
Fig. 25. Example results of using the SAR procedure to cal-
culate the natural absorbed dose of many (>100) single aliquots
from the same aeolian sediment (from an ancient sand dune in
Oklahoma, USA). Each measurement gave a similar estimate
for the natural dose and the tight clustering is typical of an
wind-blown (aeolian) sedimentary sample (from [71]).
Fig. 26. POSL versus absorbed dose from Al
2
O
3
:C (from [24]).
52 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954
deviation (1r) for ve re-estimated values (at
each dose) was found to be between 1.5% and
3.0%, for initial doses ranging from 0.3 mGy to 7
Gy [24].
A further advantage is the ability to ``image''
the dose distribution from a lter pack placed over
the sample. By using a lter with a known spatial
periodicity it is possible to decide whether or not
the badge was used while moving or while sta-
tionary, thus yielding information relevant to the
intentional irradiation of the badge under a radi-
ation source. An example of this application was
described by Akselrod et al. [25] who used POSL
from thin powder lms of Al
2
O
3
:C to image the
pattern of radiation absorption through a spatially
periodic copper lter and provided estimates of the
``condence'' in whether or not the detector was
stationary or moving during the irradiation pe-
riod. An example image is shown in Fig. 27 in
which is also shown the fourier Transform of the
image. By comparing the Fourier Transform from
an obtained image with that shown in Fig. 27 one
can obtain information on the degree of ``blur-
ring'' of the original image caused by movement of
the dosimeter during exposure. POSL from
Al
2
O
3
:C is now the basis of a new personal dosi-
meter marketed by Landauer Inc. under the name
Luxele.
5. Summary
In this paper we have attempted to provide a
historical background and conceptual basis for the
development of OSL as a means of radiation
dosimetry. Models describing OSL and related
phenomena (PTTL, photoconductivity and
bleaching of TL) have been described in order to
provide a theoretical framework from which we
may understand the myriad of OSL properties
displayed by real materials. An important pa-
rameter is the photoionization cross-section of a
trap. This parameter dictates the optical sensitivity
of a trap at a given wavelength and dictates which
wavelength is the optimum for stimulation during
OSL measurement. Several OSL readout modes
are available, each providing advantages including
ease of use (CW-OSL), separation of the individ-
ual traps contributing to the OSL (LM-OSL), and
speed and sensitivity (POSL). In general, OSL
provides a radiation dosimetry system of great
utility and sensitivity.
References
[1] D.J. Huntley, D.L. Godfrey-Smith, M.L.W. Thewalt,
Nature 313 (1985) 105.
[2] J. Fa

in, D. Miallier, M.J. Aitken, I.K. Baili, R. Gr un, A.
Mangini, V. Mejdahl, H.M. Rendell, P.D. Townsend, G.
Valladas, R. Visocekas A.G. Wintle (Eds.), Proceedings of
the 6th International Specialist Seminar on Thermolumi-
nescence and ESR Dating, Clermont-Ferrand, 1990, Nucl.
Tracks Radiat. Meas. 18 (1991) 1.
[3] I.K. Baili, J. Fa

in, J.R. Prescott, P.D. Townsend, N.
Vana, R. Visocekas (Eds.), Proceedings of the 7th Inter-
national Specialist Seminar on Thermoluminescence and
ESR Dating Krems, 1993, Radiat. Meas. 23 (1994) 267.
[4] S.W.S. McKeever (Ed.), Proceedings of the 8th Interna-
tional Conference on Luminescence and ESR Dating
Conference, Canberra, 1996, Radiat. Meas. 27 (1997) 75.
[5] S.W.S. McKeever (Ed.), Proceedings of the 9th Interna-
tional Conference on Luminescence and ESR Dating
Conference, Rome, 2000, Radiat. Meas. 32 (2000) 385.
[6] V.V. Antonov-Romanovskii, I.F.Keirum-Marcus, M.S.Po-
roshina, Z.A.Trapeznikova, in: Conference of the Academy
Fig. 27. POSL image from an Al
2
O
3
:C Luxel
TM
detector after
exposure to 1 mGy of M30 X-rays through a 1-mm-thick per-
forated copper lter in a xed (i.e. stationary) position relative
to the source. The inset shows a Fourier transform of this image
(from [25]).
S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954 53
of Sciences of the USSR on the Peaceful Uses of Atomic
Energy, Moscow, 1955, USAEC Report AEC-tr-2435 (Pt.
1), 239, 1956.
[7] P. Braunlich, D. Shafer, A. Sharmann, in: Proceedings of
the 1st International Conference on Luminescence Dosi-
metry, Stanford, June, 1965, USAEC, Vol. 57, 1967.
[8] E.N. Sanborn, E.L. Beard, in: Proceedings of the 1st
International Conference on Luminescence Dosimetry,
Stanford, June, 1965, USAEC, Vol. 183, 1967.
[9] R.P. Rao, M. de Murcia, J. Gasiot, Radiat. Prot. Dosim. 6
(1984) 64.
[10] E. Tochilin, N. Goldstein, W.G. Miller, Health Phys. 16
(1969) 1.
[11] C.R. Rhyner, W.G. Miller, Health Phys. 18 (1970) 681.
[12] J. Hanniger, B. Horlbeck, K. Hubner, K. Prokert, Nucl.
Instr. and Meth. 204 (1982) 209.
[13] R. Bernhardt, L. Herforth, in: Proceedings of the 4th
International Conference on Luminescence Dosimetry,
Krakow, Poland, Vol. 1091, 1974.
[14] A.S. Pradhan, Int. J. Appl. Radiat. Isotopes 28 (1977) 534.
[15] A.S. Pradhan, R.C. Bhatt, Phys. Stat. Sol. (a) 68 (1981) 405.
[16] R.C. Yoder, M.R. Salasky, Health Phys. 72 (1997) S18.
[17] S.D. Miller, G.W.R. Endres, J.C. McDonald, K.L. Swinth,
Radiat. Prot. Dosim. 25 (1988) 2010.
[18] D.F. Regulla, Health Phys. 22 (1975) 491.
[19] S.D. Miller, G.W.R. Endres, Radiat. Prot. Dosim. 33
(1990) 59.
[20] E. Piesch, B. Burgkhardt, M. Vilgis, Radiat. Prot. Dosim.
33 (1990) 215.
[21] E. Piesch, B. Burgkhardt, M. Vilgis, Radiat. Prot. Dosim.
47 (1993) 409.
[22] B.G. Markey, L.E. Colyott, S.W.S. McKeever, Radiat.
Meas. 24 (1995) 457.
[23] S.W.S. McKeever, M.S. Akselrod, B.G. Markey, Radiat.
Prot. Dosim. 65 (1996) 267.
[24] M.S. Akselrod, S.W.S. McKeever, Radiat. Prot. Dosim. 81
(1999) 167.
[25] M.S. Akselrod, N.A. Larsen, S.W.S. McKeever, Radiat.
Meas. 32 (2000) 215.
[26] M.S. Akselrod, N.A. Larsen, V. Whitley, S.W.S. McKe-
ever, J. Appl. Phys. 84 (1998) 3364.
[27] G.P. Summers, Radiat. Prot. Dosim. 8 (1984) 69.
[28] G. H utt, I. Jaek, J. Tchonka, Quat. Sci. Rev. 7 (1988) 381.
[29] G. H utt, I. Jaek, Nucl. Tracks Radiat. Meas. 21 (1993) 95.
[30] J.L. Merz, P.S. Pershan, Phys. Rev. 162 (1967) 217, 235.
[31] N.F. Mott, R.W. Gurney, Electronic Processes in Ionic
Crystals, Clarendon Press, Oxford, 1948.
[32] H.G. Grimmeis, L.-

A. Ledebo, J. Appl. Phys. 46(1974) 2155.


[33] H.G. Grimmeis, L.-

A. Ledebo, J. Phys. C: Sol. Stat. Phys.


8 (1975) 2615.
[34] V. Whitley, S.W.S. McKeever, A. Appl. Phys. 87 (2000)
249.
[35] C. Ditlefsen, D.J. Huntley, Radiat. Meas. 23 (1994) 675.
[36] K.W. B oer, Survey of Semiconductor Physics, Van No-
strand Reinhold, New York, 1990.
[37] A.M. Stoneham, Theory of Defects in Solids, Clarendon
Press, Oxford, 1975.
[38] P. Braunlich, in: P. Braunlich (Ed.), Thermally Stimulated
Relaxation in Solids, Vol. 1, Springer, Berlin, 1979.
[39] L. Btter-Jensen, G.A.T. Duller, N.R.J. Poolton, Radiat.
Meas. 23 (1994) 613.
[40] F.D. Walker, L.E. Colyott, N.A. Larsen, S.W.S. Mc-
Keever, Radiat. Meas. 26 (1996) 711.
[41] M.F. Morris, S.W.S. McKeever, Radiat. Meas. 23 (1994)
323.
[42] R. Chen, W.F. Hornyak, V.K. Mathur, J. Phys. D: Appl.
Phys. 23 (1990) 724.
[43] S.W.S. McKeever, Nucl. Tracks Radiat. Meas. 18 (1991) 5.
[44] C.M. Sunta, S. Watanabe, J. Phys. A: Appl. Phys. 9 (1976)
1271.
[45] N.R.J. Poolton, L. Btter-Jensen, G.A.T. Duller, Radiat.
Meas. 24 (57) (1995) 531.
[46] N.R.J Poolton, L. Btter-Jensen, O. Johnsen, J. Phys. A:
Condens. Matter 7 (1995) 4751.
[47] N.A Spooner, Radiat. Meas. 23 (1994) 593.
[48] W.G. Buckman, M.R. Payne, Health Phys. 31 (1976) 501.
[49] V.K. Jain, Phys. Stat. Sol. (a) 60 (1980) 1.
[50] L.E. Colyott, M.S. Akselrod, S.W.S. McKeever, Radiat.
Prot. Dosim. 72 (1997) 87.
[51] R. Chen, S.W.S. McKeever, Theory of Thermolumines-
cence and Related Phenomena, World Scientic, Singa-
pore, 1997.
[52] S.C. Alexander, M.F. Morris, S.W.S. McKeever, Radiat.
Meas. 27 (1997) 153.
[53] S.C. Alexander, S.W.S. McKeever, J. Phys. D: Appl. Phys.
31 (1998) 2908.
[54] N.A. Spooner, J.R. Prescott, J.T. Hutton, Quat. Scit. Rev.
7 (1976) 1271.
[55] A.G. Wintle, A.S. Murray, Radiat. Meas. 27 (1997) 611.
[56] B.W. Smith, E.J. Rhodes, Radiat. Meas. 23 (1994) 541.
[57] I.K. Baili, N.R.J. Poolton, Nucl. Tracks Radiat. Meas. 18
(1991) 111.
[58] S.W.S. McKeever, L. Btter-Jensen, N.A. Larsen, G.A.T.
Duller, Radiat. Meas. 27 (1997) 161.
[59] E. Bulur, Radiat. Meas. 26 (1996) 701.
[60] E. Bulur, H.Y. G oksu, Radiat. Meas. 30 (1999) 505.
[61] L. Btter-Jensen, G.A.T. Duller, A.S. Murray, D. Baner-
jee, Radiat. Prot. Dosim. 84 (1999) 335.
[62] E. Bulur, L. Btter-Jensen, A.S. Murray, Radiat. Meas. 32
(2000) 407.
[63] C.K. Kuhns, N.A. Larsen, S.W.S. McKeever, Radiat.
Meas. 32 (2000) 413.
[64] V.H. Whitley, Ph.D. Thesis, Oklahoma State University,
2000.
[65] M.S. Akselrod, A.C. Lucas, J.C. Polf, S.W.S. McKeever,
Radiat. Meas. 29 (1998) 391.
[66] A.G. Wintle, Radiat. Meas. 27 (1997) 769.
[67] S.W.S. McKeever, N.A. Larsen, L. Btter-Jensen, V.
Mejdahl, Radiat. Meas. 27 (1997) 75.
[68] C.A. Richardson, Radiat. Meas. 23 (1994) 587.
[69] A.G. Wintle, A.S. Murray, Radiat. Meas. 32 (2000) 387.
[70] A.G. Wintle, A.S. Murray, Radiat. Meas. 29 (1998) 81.
[71] K. Lepper, N. Agersnap Larsen, S.W.S. McKeever,
Radiat. Meas. 32 (2000) 603.
54 S.W.S. McKeever / Nucl. Instr. and Meth. in Phys. Res. B 184 (2001) 2954

You might also like