THE SCINTILLATION MECHANISM IN
THALLIUM DOPED ALKALI HALIDES
R. Gilbert Kaufman, Wayne B. Hadley and Herbert N. Hersh
Zenith Radio Corporation
Chicago, Illinois
Summary
In the scintillation mechanism for
alkali iodides doped with thallous
iodide:
1. Most of the luminescence, and
possibly all of it, is from recombination
and therefore energy transfer is
predominantly from charge transport to
thallium.
2. There are two modes of
recombination at thallium which differ in
the sequence of charge trapping at Tl+.
The electron may be trapped before a
mobile hole recombines with it, or the
hole may be trapped before recombination
with a free electron.
3. Thermal hole diffusion is an
important process at room temperature
and determines which mode a recombination
event follows.
4. Both modes of recombination are
active.
5. There is non-thermal hole
trapping which is a significant part of
the scintillator mechanism.
1. Introduction
When ionizing radiation is absorbed
in NaI(TlI) doped with small amounts of
T1I, scintillations of blue light are
produced and the emission spectrum of
the blue light is similar to the
thallium ion fluorescence excited by *non-
ionizing radiation at wavelengths where
there is no optical absorption for NaI.
Since the maximum luminescence efficiency
occurs at low concentrations (.05 to .1
mole per cent) and the energy efficiency
is high (13%), it is inferred that the
ionizing radiation is absorbed by the
NaI host lattice and the deposited energy
subsequently transferred by some
mechanism to thallium giving thallium
luminescence.1'2 It is this mechanism,
in its parts and various processes, which
will be considered. With only trivial
differences we believe that the same
processes operate in all the alkali
iodides. Birks has exhaustively dis-
cussed the many possible processes and
so there is no particular need to repeat
a review here.1 It is enough for our
purpose to classify the energy transfer
processes into two categories, those
which involve charge transport followed
by recombination and those which do not.
We will restrict our discussion to those
processes which involve charge transport.
Work published over the past few years
82
from our laboratory and from several
others has provided much new information
on the trapping of electrons and holes
in thallium doped alkali halides at low
temperatures, and on the recombinatio_11
processes which produce luminescence.
These studies have in effect isolated
some individual steps in the scintil-
lation mechanism and from them a more
complete mechanism than that presently
in the literature can be constructed for
the scintillation mechanism at room
temperature.
2. Luminescence is Mainly From
Recombination
There is a very simple experiment
which shows that most of the luminescence
produced in a crystal of KI(TlI) irradi-
ated by x rays is from the recombination
of electrons and holes. When KI(T1I) is
irradiated at 780K two effects are
observed, the crystal gives off light and
new absorption bands rapidly appear which
can be bleached with infrared light after
irradiation to give additional light
having an emission spectrum similar to
that produced during irradiation.7 When
a light sum is made of the luminescence
produced during irradiation, I, and a
second light sum made of the luminescence
produced during infrared bleaching, II,
the sum of the two, I + II, increases
linearly with the irradiation time, or
the radiation dose. For short x-ray
exposures the ratio of the second light
sum, II, to the total light sum
approaches 0.50 for a TlI concentration
of 0.064 mole per cent. The radiation-
produced absorption bands have been
studied in great detail and assigned to
one kind of trapped elegtron and two
kinds of trapped holes. Infrared
bleaching stimulates the recombination
of electrons with the trapped holes
producing luminescence by processes dis-
cussed in the following section.
If we assume for the purpose of a
simple working model that an x-ray
photon absorbed by the photoelectric
effect produces a shower of electrons
and holes, then three categories of
processes can be considered:
I. Recombination of Separated Charges
Some of the electrons and holes
will be free, i.e. in the process of
being formed they are sufficiently
separated to be considered, by criteria
which will be discussed later, not
associated. Luminescence is produced
by recombination of charges.
II. Recombination of Paired Charges
Some of the electrons and holes will
be associated. In the process of being
formed they never completely escape from
their charge partner and recombination
can be considered parent ion recapture.
Luminescence is produced by recombination
of charges.
III. Hot Electron Excitation
Some of the electrons in the shower
will have sufficient kinetic energy to
excite the states which are responsible
for the luminescence without first pro-
ducing electrons and holes.
We believe that the light sum
experiment sets a lower limit on the
fraction of luminescence events which are
the result of recombination of free
electrons and holes. We have obtained a
ratio of I/(I + II) equal to 0.60 in
KI(TlI) 0.14 mole per cent and expect
that it could go even higher with more
TlI in the crystal. There is, however,
a practical limit. Increasing TlI also
increases the efficiency of electron re-
trapping and it is very difficult to
bleach the trapped electron at a reason-
able rate. With 1.3 mole%TlI infrared
bleaching is exceedingly slow; however,
the prompt luminescence at 780K is about
a quarter the intensity of that from
0.14 mole%TlI and one can estimate that
up to 90% of the luminescence from
KI(TlI) is the result of recombination.
From a comparison of the emission spectra
produced at 50K while x-raying and by
infrared stimulation after x-raying, it
has been estimated that at least 10% of
the luminescence during irradiation is
probably not fr ?m recombination
(category III).
3. Recombination Modes
Electrons and holes trapped at low
temperature in thallium and other heavy
metal doped alkali halides have been
studied extensively and we know with
remarkable detail the chemical identity
and structure of these centers and the
properties of the lumin synce produced
when they recombine.31 I The present
discussion will be restricted to KI(TlI),
the model material we use for the
proposed scintillation mechanism and the
material with which- y have the most
direct experience;7 there are, how-
ever, supportive studies in other
materials to which we have made refer-
83
ence which are the foundations of our
understanding of recombination modes in
KI(TlI). The following is a summary of
the recombination modes in KI(TlI) and
the conditions under which they occur at
low temperatures.
When KI(TlI) is irradiated below
800K electrons are trapped at substitu-
tional Tl+ ions as T10 centers and holes
are self-trapped as Vk centers (chemi-
cally a Vk center is I2 and occupies two
iodide ion lattice sites) or at Tl+ ions
as T1++ centers. The fraction of holes
trapped as Tl++ depends linearly on the
thallium concentration as will be dis-
cussed in a later section. The recombi-
nation of these trapped electrons and
holes can be produced either by ionizing
the trapped electrons with infrared light
(which gives a four-band emission
spectrum nearly identical to that meas-
ured during irradiation) or by warming
the crystal to room temperature. There
are two major glow peaks, at 1050K and
1750K, having different single-band
emissions. Optical ionization of trapped
electrons produces recombination of
electrons at Vk centers giving well char-
acterized emission bands at 300 nm and
370 nm, and recombination of electrons
at Tl++ centers gives the lower energy
emissions observed in T1+ fluorescence,
viz 335 nm and 430 nm at 50K or only the
430 nm band at 780K. During infrared
illumination some of the freed electrons
are retrapped at Tl+ ions, some of which
might be near Vk centers. Neighboring
Tl and Vk centers may undergo a radia-
tive recombination giving a single broad
emission band at 420 nm which is indis-
tinguishable from the emission of the
first thermal glow. In the thermal
recombination process at about 900K, Vk
centers start to move and they either
recombine with T10 centers producing the
radiative recombination emission at
420 nm or they are trapped at T1+ sites
as T1++ centers. In the second glow it
is believed that electrons are thermally
released from T10 centers and that
recombination occurs at Tl++ centers.
The emission of the second glow is iden-
tical to Tl+ ion fluorescence.
From these low temperature processes
it is evident that there are two possible
modes of recombination at thallium ions
depending on the sequence of trapping.
A. (1) e + Tl+ T10
(2) h + T10 +T1+ + hv'
hv' is a single broad band at
420 nm.
 B. (1) h + T1+ Tl++
(2) e + T1+++ T1+ + hv
hv is a single band at 430 nm (780K)
or two bands, 335 nm and 430 nm
(50K).
At sufficiently low temperatures these
two modes produce different emissions but
at room temperature the emissions are
probably not distinguishable; for
example in KCl(TlCl) the emission at 410
nm from the radiative recombination of
C12 and T10 is replaced by a thallium
emission at 298 nm around 2080K.4 The
existence of the two modes of recombina-
tion has been established in alkali
halides doped with other ions as well.
We believe that the two modes of recombi-
nation at thallium are an important part
of the scintillation mechanism and that
both operate at room temperature.
4. Spatial Distribution of Electrons
and Holes
We have demonstrated in a previous
study that the electrons and holes trap-
ped in KI(TlI) which has been irradiated
with soft x rays at 780K are spatially
distributed in irs rather than distri-
buted randomly. There are two inter-
esting reasons why pairing might occur
with trapped charges. Because an
electron is attracted to a hole it may be
more likely to be trapped nearby if it
doesn't recombine. Once it has been
trapped near a hole the pair forms a di-
pole which would have much less attrac-
tion for an electron than a free hole.
Alternatively, the electron and hole may
simply be trapped near their site of
generation. It is therefore evident that
the trapping of charges in pairs does not
necessarily reflect the distribution of
charges as they are formed in the charge
shower following the absorption of an
x-ray photon. Any kinetic treatment used
to describe the recombination of elec-
trons and holes at thallium by the two
modes must assume that electrons and
holes are either spatially correlated or
they are not. If there is spatial cor-
relation monomolecular kinetics will
describe the process and if there is not
spatial correlation bimolecular kinetics
will describe the process. There are
also other interesting consequences.
Even with spatial correlation of
electrons and holes, recombination for
each pair of charges could occur by
either mode A or mode B. Consider the
following processes.
A. (0) KI + e + h
(1) e + T1+ + T10
84
(2) h + T10 "
T1+ + hV
(3) e + h KI
B. (0) KI - e + h
(1) h + T1+ T1++
(2) e + T1+++ T1+ + hV
(3) e + h + KI
Ionizing radiation produces pairs of
electrons and holes, (0), i.e. each elec-
tron and hole is effectively isolated
from all other electrons and holes in the
system for purposes of recombination.
The electron, which has a much greater
mobility than the hole, can either be
trapped at Tl+, A(l), or recombine with a
free hole giving no luminescence (3).
Luminescence is produced if the hole re-
combines with the electron trapped as
T10; however, this is not the only pos-
sible fate of the hole even with highly
correlated pairing of charges. The hole
may alternatively be trapped at T1+ as
Tl+. The T10 is not a stable electron
trap at room temperature, although it
would, of course, have a definite life-
time, whereas the thallium trapped hole
appears to be stable at room tempera-
ture. Therefore, if the hole is trapped
at thallium, it will sit there until the
electron recombines with it. The mode of
recombination for any pair is determined
by the hole which can either combine with
the trapped electron, A(2), giving recom-
bination by mode A, or it can be trapped
at Tl+, B(1), eventually recombining by
mode B.
Consider the ratio 0l(Tl0)/02(Tl+)
where a1 is the cross section for hole
capture at T10, (T1°) the effective con-
centration of T10, 02 the cross section
for trapping of holes at T1+ and (Tl+)
the effective concentration of Tl+. This
ratio determines the relative importance
of mode A and B. It would not be un-
reasonable to assume that 01 = 02 and
therefore that the ratio of A/B is equal
to the ratio of effective concentrations
(Tlo)/(Tl+). That the effective concen-
tration of these species is not the same
as the average concentration can be
demonstrated from the processes in the
first glow where trapped holes are
thermally activated to either recombine
with T10 or retrap at Tl+. T10 is less
than 1% of the total thallium whereas the
ratio of recombination to retrapping is
about 3/7 for KI(T1I) .064 mole per cent
rather than >1/100. We emphasize that
this ratio is for trapped charges which
are known to be paired and therefore
should not be applied without proper
caution to the kinetics of the scintil-
lator mechanism at room temperature. It
does, however, illustrate that
(TlO)/(Tl+) can be quite sizable when the
charge centers are spatially distributed
in pairs and probably can be considered
at most an upper limit for (Tlo)/(Tl+) if
electrons and holes are correlated in the
scintillation mechanism. The ratio of
effective concentrations of T10 and Tl+
is also an important parameter if elec-
trons and holes are not correlated in the
scintillation mechanism. When the ratio
(Tl°)/(Tl+) is small, mode B is expected
to be the predominant mode of recombina-
tion and the kinetics for such a proce4s
have been developed as discussed below.
5. Thermally Activated Hole Trapping
Gwin and Murray have suggested the
recombination of a hole with an electron
at thallium as the recombination process
in the scintillator mechanism and
recently Dietrich and Murray have made a
calculation using measured activation
energies for hole migration and electron
trapping at Tl+ which indicates that the
thallium-trapped electron has a much
longer lifetime than the free hole but
they do not consider both modes of recom-
bination.14
Pooley, in studying the inhibition
of F center formation in alkali halides
by heavy metals which act as recombina-
tion centers for electrons and holes,
has developed kinetics which we have
applied to the scintillator mechanism.15
They assume a random distribution of
electrons and holes in the stationary
state, as did Pooley, but allow recombi-
nation by either mode A or mode B but not
both. Such a kinetic treatment predicts
that the thallium luminescence intensity
would de end linearly on thallium concen-
tration. We have recently determined
experimentally that the thallium lumines-
cence intensity from KI(TlI) irradiated
with x rays depends on the square root of
the thallium concentration. We replotted
the Harshaw data on pulse height vs Tl+
concentration and found a (Tl )½ depend-
ance is a much better fit than (Tl+).
We are at present considering other
kinetic schemes and experiments which
will hopefully settle the question of
spatial correlation, and thereby the
question of whether to use a mono-
molecular or a bimolecular kinetic treat-
ment. It is, however, quite clear that
both modes of recombination must be used
since no kinetic scieme using only one
mode will give (Tl) .
Gwin and Murray noted an interesting
effect when they compared the emission
from CsI(TlI) excited by low L.E.T. radi-
ation, x ray, and high L.E.T. radiation,
a particles. In CsI the intrinsic host
emission which can be produced in process
85
(3) persists up to room temperature.
They found a particular thallium concen-
tration, 0.002 mole per cent, where high
L.E.T. radiation produced both thallium
emission and the intrinsic emission but
low L.E.T. radiation produced only
thallium emission. The a particle radi-
ation gives a higher ionization density
than x-radiation and therefore the shift
in recombination mode suggests that the
recombination kinetics with a particle
radiation are bimolecular. It seems
highly improbable that the free charges
produced, for example, in the scintil-
lation pulse of an absorbed x ray would
interact with the charges from other
absorbed x rays produced at the dose
rates of most laboratory x-ray machines.
Such is not the case for a particle
radiation.
6. Non-thermal Hole Trapping in KI(TlI)
Thus far we have considered what may
be characterized as that part of the
total scintillator mechanism which
depends on the thermal migration of
holes. We would now like to present
some new data and suggest that in the
recombination of electron and holes
there is a part which does not depend on
the thermal migration of holes, but on
the direct trapping of holes, depending
linearly on the thallium concentration
and being independent of temperature
between 50K and 780K.
In previous work we have establish-
ed that when KI (TlI) .07 mole per cent
is x-rayed at 780K, about 7% of the
trapped holes are T1++.S The method is
summarized in the following. The light
sum of the f irst thermal glow is about
30% of the total thermally stimulated
luminescence. This means that of the
original I2 centers produced by
irradiation, 30% combine with trapped
electrons producing the first thermal
glow while the remainder are retrapped
as Tl++. There is independent evidence
showing that the retrapped holes are
Tl++ centers having an absorption band
at 306 nm which does not significantly
overlap the I2 and T1i absorption
bands. The increase in size of the 306
nm absorption band indicates a tenfold
increase in the number of trapped holes
at thallium. Let Nv represent the
initial number of I2- centers and Nt the
initial number of Tl++ centers. The
number of T1++ centers increases to 10
Nt while all the I2- centers disappear
during+the
in Tl
first glow. The net increase
centers, 9 Nt, is equal to the
number of I¢- centers that retrap rather
than recombining with electrons. This
equality is expressed as:
 9Nt = 0. 70 Nv.
From this Nt = 0.078 Nv
and Nt/(Nt + Nv) = 0.078/1.078 = 0.072.
Thus T1++ centers account for about 7% of
the total number of holes produced by
irradiation at 780K.
Using this method we find that the
fraction of holes which is directly
trapped as T1++ depends linearly on the
concentration of thallium. In crystals
which were later chemically analyzed for
thallium the fraction of Tl++ holes was
less than 1% at 0.009 mole per cent Tl+,
7% at 0.062 mole per cent Tl+, and 14%
at 0.14 mole per cent Tl+. By extra-
polation of these data the fraction of
holes at thallium should reach 100% at
1 mole per cent Tl+.
In KI(T1I) 1.3 mole per cent we
found no evidence for I2- centers after
irradiation both at 50K and at 780K. The
absorption spectra and thermal glows
which were measured indicated that there
was no I2 present, i.e. there was no
distinguishable 800 nm or 404 nm absorp-
tion bands and no first glow peak as the
crystal was warmed past 1050K. The
clearest evidence that there were no I2-
centers comes from stimulated emission
experiments of figure 1. Crystals of
KI (TI) with different concentrations of
T1+ were first x-rayed and then illumi-
nated with low intensity infrared as
described previously. The infrared light
bleaches electrons which then are either
retrapped or recombine at holes. When
they recombine with I2 an emission
having ba s at 300 nm and 370 nm is
produced. The other emission bands of
figure 1 are either from recombination
with thallium-trapped holes or from the
radiative recombination between nearby
trapped electrons and holes. The spectra
of figure 1 show that as the thallium
content increases the relative intensity
of emissions at 300 nm and 370 nm
decreases. At 1.3 mole per cent there
are no emission bands at 300 nm and 370
nm. From the estimated sensitivity of
the emission spectra less than one per
cent of the trapped holes are I2- centers.
From the above measurements it
appears that any hole formed within a
volume equal to 25 unit cell volumes
around the Tl+ ion will form T1++ before
recombination whether the self-trapped
holes are mobile or not, i.e. it is a
non-thermal process. In contrast to the
thermal migration of holes to either TI+
or T10, this non-thermal trapping of
86
holes at T1+ has not been recognized
previously as a part of the scintillator
mechanism.
7. Non-thermal Hole Trapping in NaI(TlI)
Similar experiments were carried out
on a sample of Harshaw Scintillator Grade
NaI(TlI). The chemically determined
thallium concentration was 0.14 mole per
cent. Figure 2 shows the absorption
spectrum measured at 50K. Before irradi-
ation there are two major absorption
bands, the larger centering around 288 nm
from Tl+ and a second centering around
303 nm from thallium dimers. After
irradiation at 50K with 125 keV x rays a
band appeared at 312 nm which has been
assigned to Tl++ from the same arguments
used to assign the 306 nm band to T1++
in KI (TlI) . The first glow peak, occur-
ring at 590K, was removed and the crys-
tal re-cooled to 50K. The 312 nm absorp-
tion band increased by a factor of 6.3.
The light sum for the first glo is about
10% of the total thermal glow . As
above, let Nv represent the initial
number of I2 centers and Nt the initial
number of Tl++ centers. The number of
Tl++ increases to 6.3 Nt while all the
I2- centers disappear during the first
glow. The net increase in Tl++, 5.3 Nt,
is equal to the number of I2- centers
that retrap rather than recombining with
electrons.
5.3 Nt = 0.90 Nv
From this Nt = 0.17 Nv
and Nt/(Nt + Nv) = 0.17/1.17 = .146.
Thus Tl++ centers account for about 15%
of the total number of holes produced by
irradiation of NaI(TlI) 0.14 mole per
cent at 50K. Non-thermal hole trapping
in NaI is therefore similar to non-
thermal hole trapping in KI: any hole
formed within a volume of 25 unit cell
volumes about Tl+ will trap as Tl++.
As a consequence of the non-thermal
mode of hole trapping we would expect
that even at room temperature approx-
imately 15% of the ionization events
producing luminescence would not require
the thermal migration of holes. Figure
3 shows a plot of pulse height vs T1+
concentration taken from the Harshaw data
and a second curve from which we predict
what part of that pulse height is the
result of the direct non-thermal hole
trapping mechanism. It is not an insig-
nificant part of the scintillator
mechanism.
 Acknowledgements
We thank Prof. R. B. Murray for
several discussions on hole trapping and
showing us the results of experiments and
calculations by Dietrich and himself on
the lifetime of free holes and electrons
trapped at T1+ for room temperature. We
feel the work of Gwin and Murray is a
particularly significant and clear
exposition of the importance of recom-
bination in the scintillator mechanism.
We also thank Dr. M. Elango for a rather
extended discussion of recombination
luminescence studies pertinent to the
scintillator mechanism which are going on
in the Soviet Union. We hope to comment
on this work in a later paper.
References
1. J. B. Birks, The Theory and Practice
of Scintillation Counting, The Mac-
millan Company, New York 1964.
2. W. J. Van Sciver, IRE Trans. Nuc.
Sci. NS-3, 39 (1956) W. J. Van
Sciver and L. Bogart, IRE Trans. Nuc.
Sci. NS-5, 90 (1958).
3. C. J. Delbecq, W. Hayes, M.C.M.
O'Brien and P. H. Yuster, Proc. Roy.
Soc. (London) A271, 243 (1963).
4. C. J. Delbecq, A. K. Ghosh and P. H1.
Yuster, Phys. Rev. 151, 599 (1966).
5. H. Blume and P. Brauer, Z. Natur-
forsch. 20, 1236 (1965).
6. Ch. B. Lushchik, Proceedings of the
Luminescence Conference at the
University of Delaware, August 1969,
87
to be published. A great deal of
work has been done on recombination
luminescence in alkali halides at the
Institute for Physics and Astronomy,
Academy of Sciences of the Estonian
SSR, Tartu which is summarized in the
Lushchik review.
7. R. G. Kaufman and W. B. Hadley, J.
Chem. Phys. 44, 1311 (1965).
8. W. B. Hadley, S. Polick, R. G.
Kaufman and H. N. Hersh, J. Chem.
Phys. 45, 2040 (1966).
9. R. G. Kaufman and W. B. Hadley, J.
Chem. Phys. 46, 1598 (1967).
10. R. G. Kaufman and W. B. Hadley, J.
Chem. Phys. 47, 264 (1967).
11. R. G. Kaufman and W. B. Hadley,
Trans. IEEE Nuc. Sci. NS-15, 158
(1968).
12. W. Kansig, Phys. Rev. 99, 1890
(1955). T. Castner anUW. Kansig,
J. Phys. Chem. Solids 3, 178 (1957).
13. R. B. Murray and F. J. Keller, Phys.
Rev. 137, A942 (1965) M. N. Kabler,
Phys. Rev. 136, A-1296 (1964) .
14. R. B. Murray, private communication
of work by Dietrich and Murray.
15. D. Pooley, Proc. Phys. Soc. 89, 723
(1966).
16. The IHarshaw data has been reproduced
in several places including Birks1
and the Harshaw catalog for scin-
tillation phosphors.
 LUI

Z~~~~~~~~~~~~~. It IL %

.009 mOLE %

300 350 400 450 500

WAVELENGTH (nm)
Figure 1. Stimulated emission at 5°K
from KI(T1I) with several different con-
centrations of TlI, after first x-raying
at 50K.

Nal(Ttl) 5"K
15r
w

__

C.)
Go
az
0 Ii Ilo X
CO,)

(I 51 - _

0 300 310 330 D. 1 0.2 0

WAVELENGTH (nm) MOLE % TI IN NaI

Figure 2. The absorption spectrum of
NaI (TlI) 0.14 mole per cent at 50K be-
fore x-raying, (1). The absorption
spectrum after 30 minutes x irradiation
at 50K, (2). The absorption spectrum at
50K after the first glow was removed,
(3).
Figure 3. A replot of the Harshaw data
on pulse height from the Cs 17 y ray vs
Tl+ concentration, (1) .16 The effect of
thallium content on the pulse height
expected from non-thermal hole trapping.

88