Вы находитесь на странице: 1из 8

Radiation Protection Dosimetry (2012), Vol. 149, No. 4, pp.

363 370
Advance Access publication 15 July 2011

doi:10.1093/rpd/ncr315

D. Afouxenidis1,2,*, G. S. Polymeris1, N. C. Tsirliganis1 and G. Kitis2


1
Archaeometry Laboratory, Cultural and Educational Technology Institute (C.E.T.I.), R.C. ATHENA,
Tsimiski 58, 67100 Xanthi, Greece
2
Nuclear Physics Laboratory, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
*Corresponding author: dafouxen@ipet.gr
Received March 3 2011, revised May 12 2011, accepted June 17 2011
This paper exploits the possibility of using commercial software for thermoluminescence and optically stimulated luminescence
curve deconvolution analysis. The widely used software package Microsoft Excel, with the Solver utility has been used to
perform deconvolution analysis to both experimental and reference glow curves resulted from the GLOw Curve ANalysis
INtercomparison project. The simple interface of this programme combined with the powerful Solver utility, allows the analysis of complex stimulated luminescence curves into their components and the evaluation of the associated luminescence
parameters.

INTRODUCTION
The computerised curve deconvolution (CCD) analysis of thermoluminescence (TL) glow curves and
optically stimulated luminescence (OSL) decay
curves into their individual glow peaks and
components respectively have been recognised over
the last 30 y to be of major importance(1 5). The
information by the CCD concerning the trap energy
depth (E), frequency factor (s), kinetic order (b),
photoionisation cross section (s), etc, is very useful
in order to understand the luminescence mechanism
of materials.
The capabilities of several computer codes used by
the various research groups for CCD and their
assessment of the glow curve parameters were tested
by the GLOw Curve ANalysis INtercomparison
(GLOCANIN) project. Participants in the GLOC
ANIN project were asked to analyse the so-called
reference glow curves using their computer codes.
The results on synthetic as well as on experimentally
measured glow curves can be found in references(6, 7).
The application of the CCD consists essentially of
two steps: (i) deciding on the mathematical model that
describes a single TL/OSL peak and (ii) assessing the
values of the curve parameters that minimise the sum
of squares of the differences between the fitted model
and the experimental curve(8). Concerning the first
step, there is a number of well-established analytical
expressions in the literature describing single TL/OSL
peaks(6, 9, 10). Although a great amount of work exists
on deriving analytical single peak expressions and
despite the significant information that the CCD
analysis yield, the latter is not widely adopted by the
TL/OSL community as a basic tool, while the

majority of TL/OSL publications does not involve


any CCD analysis of relevant data. In authors
opinion, this could be partly attributed to the lack of
available commercial software; so every researcher has
to write his/her own program.
The aim of the present work is to offer a solution
to the aforementioned problem that will enable any
researcher to analyse easily and accurately even the
most complex TL/OSL curves consisting of many
overlapping individual peaks, employing the commonly used spreadsheet software package Microsoft
Excel, along with its Solver, add-in utility.
SINGLE TL/OSL PEAK ANALYTICAL
EXPRESSIONS
The analytical expressions used in the present work
are the well-known expressions for general and
mixed order kinetics, named GOK and MOK,
respectively hereafter(7, 11 14). The selection of these
expressions was based on the following:
(1) The GOK expression, although empirical is
almost exclusively used in the literature(7).
(2) The MOK expression, although rarely used in
the literature, is physically meaningful(11).
(3) The GOK expression for b 2 coincides with
second-order kinetics, whereas for b ! 1 coincides
with first-order kinetics.
(4) Similarly, the MOK expression for a ! 1
coincides with second-order kinetics and for a 0
coincides with first-order kinetics.
The advantage of those expressions is that
only one expression accounts for both first- and
second-order kinetics, and furthermore it can

# The Author 2011. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

COMPUTERISED CURVE DECONVOLUTION OF TL/OSL


CURVES USING A POPULAR SPREADSHEET PROGRAM

D. AFOUXENIDIS ET AL.

advantage of the I (Im, Tm, E, b or a, T ) representation is not easily seen in the cases of experimental
curves with one single peak. However, it becomes
clear in cases of experimental curves consisting of
many overlapped peaks.
It must be noted that the proposed expressions
involving Im, Tm result as transformations of the
original expression involving n0 and s. The only
approximation in these expressions is the usual
approximation of the exponential integral appearing
in TL theory. The analytical TL expressions are
derived using two terms of the asymptotic series
approximation (ASA) of the exponential integral.
The transformed equations can include more terms
of the ASA through the term Dm, (see later text),
which should be simply replaced by the ASA
expressions above the third term. Details about the
ASA approximation can be found in reference(15)
SPREADSHEET PREPARATION
The steps needed to perform a CCD analysis with
an Excel spreadsheet are shown in Table 1. The
example used is that of a complex TL glow curve
consisting of two individual peaks using the GOK
expression.
In the first step, one ascribes the temperature and
TL intensity of the experimental glow curve to the
columns A and B. In the second step, the subsequent
columns C and D are ascribed to each one of the
individual peaks 1 and 2, respectively.

Table 1. Calculations for TL data under GOK.


A
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
a

S I(T)fit
0.0003
0.0004
0.0005

jI(T)exp I(T)fitj
3.6E206
4.2E206
4.5E206

Heat rate (K/s)


1
Imax
Tmax (K)
E (eV)
b
M.F.

Temp
301.01
302.02
303.02

Peak #1
0.04
417
1.3
1.0001
1000a
Integral
11 098
Freq factor (s 1)
7E15

FOM (%)

1.2

I(T)exp
0.0003
0.0004
0.0005

Peak 1
0.0003
0.0004
0.0005

Peak #2
0.05
456
1.5
1.0001
1000a
16 904
2.9E15

Peak 2
4.8E206
5.9E206
7.1E206

The order of magnitude of maximum intensity Imax.

364

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

additionally account for the intermediate orders.


Concerning the decision between GOK and MOK,
one has to know that their greater difference, as it
was shown by Kitis et al.(12), is observed at about
b1.6 and this difference is ,4 %. Moreover their
difference is mainly at the high temperature range of
the TL glow-peak. Taking into account that these
differences were obtained using numerically generated peaks, one can conclude that these differences
are rather difficult to be seen in an experimental TL
peak. Therefore, in most practical experimental
cases yielding an error higher than 1 %, there is no
real dilemma in choosing between GOK and MOK,
since they will both produce similar results.
Another feature of the selected analytical
expressions is that they are of the form of I (Im, Tm,
E, b or a, T )(7, 11) and I (Im, tm, b or a, t)(13, 14), for
TL and linearly modulated (LM)-OSL, respectively,
instead of the form I (n0, s, E, b or a, T ) and I (n0,
l, b or a, t) of the original equations(9, 10), where
variable T is the temperature, Tm is the temperature
at maximum TL intensity, Im is the maximum intensity, n0 is the initial concentration of trapped electrons, tm is the time at maximum LM-OSL intensity
and l is the LM-OSL stimulation wavelength. The
advantage of the proposed presentation is that the
quantities Im, Tm and tm can be evaluated directly
and accurately from the experimental TL/LM-OSL
curves, whereas it is not possible to have any knowledge about the probable values of n0, s and E
appearing in the original form of the equations. The

COMPUTERISED CURVE DECONVOLUTION

b=b1

IT Im  b

E T  Tm

 exp
kT
Tm


T2
 b  1  1  D  2 
Tm


b=b1
E T  Tm
Zm

 exp
kT
Tm
where
D

2kT
2kTm
; Dm
; Zm 1 b  1  Dm
E
E

has to be written in Excel format in cell C19. It will


be used to reproduce the first (lower Tmax) peak in
column C by dragging it to the entire column
C. Therefore a peak for the entire temperature
region is calculated, according to the temperature
T(K ) and its four trapping parameters that were preselected. This procedure is repeated for the next
glow peak in column D, by copying and pasting the
same expressions to cell D19 and dragging it to the
entire corresponding column.
In that way, two arbitrary glow peaks are
created without any other limitation. However, the
sum of these peaks should be similar to the total
experimental curve. This similarity is usually
checked by the linear regression coefficient (x 2),
which however does not provide an immediate
clue regarding the goodness of fit. In order to circumvent this problem, another mathematical index
was selected, which is termed the FOM(16) and is
defined as

P 

p TLexp  TLfit
P
FOM% 100 
p TLfit

The FOM index value provides a measure for the


goodness of fit; the lowest its value, the best fit.
Therefore, every fitting attempt should result in
minimising the FOM index value, which is
achieved by changing the set of the parameter
values of each glow peak. This is achieved by
using certain optimisation software packages, the
Solver, the power full add in of Excel. A full
description for the latter is presented in the following section.
Returning to spreadsheet preparation procedure,
two more columns are required, for instance E and
F, in order for the sum of the fitted (calculated) glow
peaks and the absolute value of the difference
between each experimental and calculated data
point to be respectively presented. The main objective is to minimise the values in column F. Finally,
showing the FOM index value among the initial
cells of the spreadsheet is suggested.
The spreadsheet of Table 1 described earlier
stands as an example for the deconvolution of the
glow curve into its two overlapping glow peaks, in
order for the parameters associated with the individual peaks to be estimated. Only the first 21
rows are shown for the sake of brevity. Columns
(A19:A21) and (B19:B21) contain the experimental
data points for the TL glow curve, while columns
(C19:C21) and (D19:D21) contain the fitted data
points using the GOK model. Row nine holds the
value of a multiplier factor that is necessary for
further analysis and represents the order of peak
height intensity.
The heating rate used throughout the measurement is given in Cell B2 since it is a common parameter for all the glow peaks. Furthermore, the TL
integral of each glow peak (sum of the data in
column below row 19) is given in row 11 of the corresponding column, while in row 13, the corresponding frequency factors are shown. The frequency
factor values are calculated from the respective
equation(7) using the values of E, Tm and b obtained
from the curve fitting and using the heating rate
given in Cell B2.

ABOUT SOLVER
The Solver(17, 18) is an Excel Add-in, a software
program that could be found in the Tools menu; it
can be installed by checking the Solver Add-in following the path: Tools !Add-Ins. Solver is a
general-purpose optimisation package that is used in
order to find a maximum, minimum or specified
value of the target Cell. The Solver code is a
product of Frontline Systems Inc. The Solver can be
used to minimise the sum of squares of residuals
(differences between yobsd and ycalc) and thus
perform a least- square fitting. It can be used to

365

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

For the fitting procedure, one has to insert initial,


arbitrary but meaningful values for the parameters
Im, Tm, E and b for each single glow-peak. As can
be seen in Table 1, these values are inserted in the
rows 5 8 respectively in column C for the first peak
and column D for the second peak.
When the curve fitting procedure is completed, it
gives the net values of Im, Tm, E and b. However,
the evaluation of additional quantities such as the
integral of each glow-peak, the frequency factor and
figure-of merit (FOM) values (see later text) is also
feasible.
In the next step, the raw experimental data are
inserted in the same spreadsheet in a way that the
first data point appears in the 19th row of columns
A and B, corresponding to the temperature and
signal respectively.
Following that, the GOK analytical expression(7)

D. AFOUXENIDIS ET AL.

Figure 1. Solver parameter dialog box.

in the set target cell (on the Solver parameter dialog


box) be minimised.

CURVE FITTING EXAMPLES AND


DISCUSSION
TL case
The synthetic glow curve REFGLOW.002 as well as
the measured glow curve REFGLOW.009 of the
GLOCANIN project (19) are used next as examples
in order to estimate their parameters, using the
earlier-mentioned procedure in Excel.
The synthetic glow curve REFGLOW.002 is the
sum of four glow peaks obtained by solving directly
the differential equation with trapping parameters
such as the glow curve structure of TLD-100, which
describes the charge transport in the TL material
according to the well-known Randall-Wilkins model
with the Bulirsch-Stoer method. This glow curve is
included in order to evaluate the accuracy in the
determination of parameters in the case of overlapping peaks.
Figure 2 shows the reference glow curve
REFGLOW.002 in open cycles together with the calculated curve as a solid line passing through them,
using the GOK model.
Table 2 gives the calculated parameters for the
peaks 2, 3, 4 and 5 for the GLOCANIN curve
REFGLOW.002 as found by applying the solver Addin on the functions for both GOK and MOK(7, 11),
using the Microsoft Excel spreadsheet.
As can be seen from the low FOM values (0.0090
and 0.0093 % for the GOK model and the MOK
model, respectively), the glow curve deconvolution

Figure 2. Curve fitting of REFGLOW.002 using excel


spreadsheet and the Solver Add-in under GOK model. In
the lower graph residual TL intensity versus temperature,
after curve fitting is presented.

366

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

perform either linear or non-linear least-square


fitting.
The Solver uses the Generalised Reduced
Gradient non-linear optimisation code. For each of
the changing cells, the Solver evaluates the partial
derivative of the objective function (the target cell)
with respect to the changing cell ai, by means of the
finite-difference method.
The Solver uses a matrix including the partial
derivatives to determine the gradient of the response
surface and thus decides how to change the values
of the changing cells in order to approach the
desired solution.
The Solver Parameter dialog box is presented in
Figure 1. In the Set Target Cell box the cell
containing the quantity that is going to be minimised, here the FOM (%), must be typed. Because
of the minimisation process, the Min button must
be checked. In the By Changing Cells box, the
Cells that contain the fitting parameter values
must be typedhere the values of Imax, Tmax, E
and b.
With Solver, some physically imposed constrains
can be applied to the solution. For example, in the
case of GOK, the value of the kinetic parameter b is
limited by the inequality 1,b 2 and in the case of
MOK, the value of the kinetic parameter a is
limited by the inequality 0  a , 1. So this constrains must be used for the above cases.
Since the Solver operates by a search routine, a
solution would be found most rapidly and efficiently
if the initial estimates that have been provided are
close to the final values. A chart of the data that displays both yobsd and ycalc is often useful to be
created, and then a good set of initial parameters
can be estimated by varying the parameters
manually.
It must be noticed that the initial values of the
fitting parameters and the optimised parameter
values are placed in the same cell because the Solver
requires that the user should define the cells that
have to be modified (by changing cells, on the Solver
parameter dialog box) so that the quantity selected

COMPUTERISED CURVE DECONVOLUTION


Table 2. Estimated TL parameters for both GOK and
MOK using Microsoft Excel Add-in, Solver for
REFGLOW.002 as well as the references values as reported
in the CLOCANIN project.

Peak 2

Peak 3

Peak 4

General-order kinetic (FOM (%)0.008998)


456.5
484
Tmax (K) 417.1
E (eV)
1.383
1.483
1.584
b
1.0006
1.0009
1.0009
TL
11 100
16 914
27 382
3.9E16
1.6E16 2.1E16
s (s21)
Mixed-order kinetic (FOM (%)0.009339)
413.6
451.6
478.0
tmax (K)
E (eV)
1.383
1.483
1.583
a
7.2E205
0
0
TL
11 098
16 896
27 397
1
5.6E16
2.5E16 3.3E16
s (s )
(19)
CLOCANIN
tmax (K)
417.3
456.8
484
E (eV)
1.383
1.483
1.583
1
3.9 E16 1.6E16 2E16
s (s )
TL
111 001
16 898
27 401

Peak 5

511.6
2.0038
1.0006
47 302
4E19
503.7
2.0038
0
47309
8.5E19
511.9
2.0038
4E19
47 302

Note that Tmax for the CLOCANIN project is calculated


according to the Randall-Wilkins model from the given
heating rate, E and s values.

analysis process was very accurate. The similar


values of peak temperature (Tmax), activation energy
(E) and Integral, for each individual glow curve presented, as well as the same order of magnitude of
frequency factor (s), indicate the agreement between
the used models. These values are in great agreement
with those reported in the GLOCANIN project (19),
indicating the computational power of Solver utility.
The value of the calculated parameter (b) that
stands for kinetic order according to the GOK has
been calculated as b1.00065 for peak 2, very close
to unity (b1) that stands for the first-order kinetics
model that has been used in order to obtain the
REFGLOW.002, according to the Randall-Wilkins
model. On the other hand, the value of the calculated parameter (a) that stands for kinetic order
according to the MOK has been calculated as
a 7.261025. For this value (a0), the MOK
model is equivalent with the first-order kinetic
model as has been described by Kitis et al.(12).
Reference glow curve REFGLOW.009 stems for a
highly irradiated (D600 Gy) TLD 700 sample.
Apart from the well-known glow peaks 2 5, it features a complex group of high temperature peaks.
These overlapping peaks make the fitting procedure
much more complicated.
Figure 3. Curve fitting of REFGLOW.009 using
excel spreadsheet and the Solver Add-in under
General-Order Kinetics model. In the lower graph

Figure 3. Shows the reference glow curve REFGLOW.009


in open cycles together with the calculated curve as a solid
line passing through them, using the GOK model.

residual TL intensity versus temperature, after curve


fitting is presented.
Table 3 gives the calculated parameters for the
peaks 1, 2, 3, 4, 5, 5a, 6, 7, 8, 9, 10 and 11 of the
GLOCANIN curve REFGLOW.009 as they were
estimated by applying the Solver add-in utility on
the function of the GOK(7), in the Microsoft Excel
spreadsheet. These values are in great agreement
with those reported in the GLOCANIN project (19),
indicating the computational power of Solver utility.

OSL case
As opposed to the case of TL, there is no project
such as the GLOCANIN for OSL. So, there are not
any reference LM-OSL curves with their corresponding parameters in order to evaluate the capabilities
of computer codes for fitting in LM-OSL cases.
Due to the lack of any reference LM-OSL curves,
simulation was necessary in order to obtain data for
reliable curve fitting. The latter was performed using
the equation(10)

sg b=1b
I tOSL n0 sgt 1 b  1 t2
2
where n0 is the initial concentration of electrons in
traps, s the photoionisation cross section, g the
stimulation increase rate and b the kinetic order parameter. To obtain numerical values for I(t) OSL, the
values for n0 100 000, s.g0.0001 and
b1.00000001 were used. The resulting data were
used to perform a curve fitting using the equations
for GOK and MOK of LM-OSL. Figure 4 shows
the simulated LM-OSL curve as open cycles and the
calculated curve as a solid line passing through them
under the GOK model.

367

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

TL parameters

D. AFOUXENIDIS ET AL.
Table 3. Estimated TL parameters for General-Order Kinetics using Microsoft Excel Add-in, Solver for REFGLOW.009 as
well as the reference values as reported in the CLOCANIN project.
Peak 1

Peak 2

Peak 3

Peak 4

Peak 5a

Peak 6

Peak 7

General-order kinetic (FOM0.78 %)


E (eV) 0.95
1.23
1.30
1.56
2.08
s (s 1) 11013 11015 21014 11016 51020

1.77
91016

1.46
1.37
1.86
1.97
11013 91011 71015 11016

1.75
31013

1.41
11010

GLOCANIN(19)
E (eV)
1.25
1.3
1.6
2.02
11015 21014 81016 81019
s (s 1)

Peak 8

Peak 9

Peak 10 Peak 11

Table 5. Estimated parameters for both GOK and MOK


using Microsoft Excel Add-in, Solver for Al2O3:C LM-OSL
curve structure.
Al2O3:C

Figure 4. Curve fitting of simulated LM-OSL curve using


Microsoft Excel spreadsheet and the Solver Add-in
according GOK model. In the lower graph values for
residual LM-OSL intensity versus stimulation time, after
curve fitting are presented.

Mixed-order kinetics [FOM (%)1.389]


Peak 1
Peak 2
93.552
415.368
tmax (s)
a
0.950
0.909
Integral
1 932 742
4 516 100

Peak 3
1316.279
0.914
3 318 872

General-order kinetics [FOM (%)1.410]


tmax (s)
93.471
415.367
b
2.000
2.000
Integral
1 967 657
4 554 843

1316.277
1.916
3 245 255

Deviations (%)
tmax (s)
0.086
Integral
21.806

0.000
2.218

0.000
20.858

Table 4. Estimated LM-OSL parameters for both GOK


and MOK using Microsoft Excel Add-in, Solver for
simulated LM-OSL curve.
LM-OSL parameters
GOK
tmax (s)
b
Integral
FOM (%)

100.000013
1.000000001
99998.75394
0.00004929

MOK

99.99991708
0.000002696
99998.80336
0.000054287

Table 4 gives the peak parameters for the simulated curve as found by applying the Solver add-in
utility on the function of the GOK(13) and on the
function of the MOK(14), using Microsoft Excel
spreadsheet.
As can be seen from Table 4, the accuracy of the
fitting procedure using Microsoft Excel Add-in,
Solver, is extremely good, as described by the FOM

Figure 5. Shows an experimental LM-OSL curve of the


widely used Al2O3:C dosemeter (open cycles) together with
the fitted curve (solid line) using the MOK model. The
individual components of the Al2O3:C LM-OSL curve are
also shown in the same figure.

368

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

Peak 5

COMPUTERISED CURVE DECONVOLUTION

CONCLUSIONS
The deconvolution process for TL as well as linear
modulated optically stimulated luminescence curves
using a common Microsoft Excel spreadsheet was
outlined in this paper. Both GOK and MOK functions can be used for deconvolution. It should be
noted that the same process can be used for analysing the sum of a number of exponentials and is not
limited to TL and LM-OSL cases presented explicitly in this paper. For instance, CW-OSL curves can
be analysed in a similar way by applying the transformations proposed by Bulur.
The process uses the Solver utility, a Microsoft
Excel add-in, that is also briefly presented here;
Specific examples of both TL and OSL computerised curve deconvolution analysis using the
spreadsheet along with the Solver add-in, are given.
These examples show that the luminescence trapping
parameters can be estimated with a high accuracy
with a simple procedure without the need of complicated and specialised computer codes. The accuracy
of the results was further confirmed in the case of
TL by comparison with the results of the
GLOCANIN project.
This work shows the power of using a simple
interface commercial software such as Microsoft
Excel, in order to carry out complex scientific problems, such as computerised curve deconvolution
analysis. It should be noted that the authors strongly
recommend the usage of the Solver utility because it
is used within the familiar Excel environment; so
new commands and procedures do not have to be
learnt.

Microsoft Excel spreadsheets for multi-component


luminescence computerised curve deconvolution
analysis can be obtained on-line from the web site
(www.ipet.gr) of the Archaeometry Department, of
the Cultural and Educational Technology Institute
by filling a registration form with the contact information. The spreadsheets are offered free of charge,
as long as they are not used for commercial purposes
or as a part of a system for dose assessment (21).
REFERENCES
1. Horowitz, Y. S. and Moscovitch, M. Computerized
glow curve deconvolution applied to high dose (102 105
Gy) TL dosimetry. Nucl. Instrum. Methods. Phys. Res.
A 243(1), 207 214 (1986).
2. Horowitz, Y. S, Moscovitch, M. and Wilt, M.
Computerized glow curve deconvolution applied to ultralow
dose LiF thermoluminescence dosimetry. Nucl. Instrum.
Methods. Phys. Res. A 244(3), 556564 (1986).
3. Horowitz, Y. S. and Yossian, D. Computerized glow
curve deconvolution: the case of LiF TLD-100. J. Phys.
D: Appl. Phys. 26(8), 1331 1332 (1993).
4. Horowitz, Y.S. and Yossian, D. Computerized glow
curve deconvolution: application to thermoluminescence
dosimetry. Radiat. Prot. Dosim. 60(1), 1 114 (1995).
5. Furetta, C., Kitis, G. and Kuo, C. -H. Kinetics parameters of CVD diamond by computerized glow-curve
deconvolution (CGCD). Nucl. Instrum. Methods. Phys.
Res. B: Beam Interact. Mater. At. 160(1), 6572
(2000).
6. Pagonis, V., Furetta, C. and Kitis, G. Numerical and
Practical Exercises in Thermoluminescence. Springer
(2006).
7. Kitis, G., Gomes-Ros, J. M. and Tuyn, J. W. N.
Thermoluminescence glow curve deconvolution functions
for first, second and general orders of kinetics. J. Phys.
D: Appl. Phys. 31 26362641 (1998).
8. Van Dijk, J. W. E. Thermoluminescence glow curve
deconvolution and its statistical analysis using the flexibility of spreadsheet programs. Radiat. Prot. Dosim.
119, 332 338 (2006).
9. Chen, R. and McKeever,, S. W. S. Theory of
Thermoluminescence and Related Phenomena. World
Scientific (1997).
10. Btter-Jensen, L., McKeever, S. W. S. and Wintle,,
A. G. Optically Stimulated Luminescence Dosimetry.
Elsevier (2003).
11. Kitis, G. and Gomes-Ros, J. M. Thermoluminescence
glow curve deconvolution functions for mixed order of
kinetics and continuous trap distribution. Nucl. Instrum.
Methods Phys. Res. A 440, 224231 (2000).
12. Kitis, G., Chen, R. and Pagonis, V. Thermoluminescence
glow-peak shape methods based on mixed order kinetics.
Physica Status Solidi (a) 205(5), 11811189 (2008).
13. Kitis, G. and Pagonis, V. Computerized curve deconvolution analysis for LM-OSL. Radiat. Meas. 43, 737 741
(2008).
14. Kitis, G., Furetta, C. and Pagonis, V. Mixed order kinetics model for optically stimulated luminescence. Mod.
Phys. Lett. B, 23(27), 31913207 (2009).
15. Kitis, G., Chen, R., Pagonis, V., Carinou, E. and
Kamenopoulou,, V. Thermoluminescence under an

369

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

index. The respective FOM values are 0.000049 %


for the GOK model and 0.000054 % for the MOK
model. It should be noted that the estimated values
of the parameters in both cases are almost the same
with very small deviations.
In Table 5, the calculated peak parameters for the
experimental LM-OSL curve of Al2O3:C are given, as
they were estimated by applying the Solver add-in
utility on the function of the GOKand on the function
of the MOK, in the Microsoft Excel spreadsheet.
Figure 5. LM-OSL Curve fitting of Al2O3:C using
excel spreadsheet and the Solver Add-in under
General-order kinetics model. In the lower graph
values for residual LM-OSL intensity versus stimulation time, after curve fitting is presented.
As can be seen from the low FOM values (1.410
and 1.389% for the GOK model and the MOK
model, respectively), the curve deconvolution analysis process was very accurate. Deviations between
the estimated values of each of the components presented, in stimulation time at peak height (tmax), in
Integral, is also presented. These values are in great
agreement with those reported in the literature(20).

D. AFOUXENIDIS ET AL.
19. Bos, A. J. J., Piters, T. M., Gomez Ros, J. M. and
Delgado,, A. An intercomparison of glow curve analysis
computer programs: I. synthetic glow curves. Radiat.
Prot. Dosim. 47, 473 477 (1993).
20. Dallas, G. I., Polymeris, G. S., Stefanaki, E. C.,
Afouxenidis, D., Tsirliganis, N. C. and Kitis, G. Sample
dependent correlation between TL and LM-OSL in
Al2O3:C. Radiat. Meas. 43(2 6), 335 340, (2008).
21. Bos, A. J. J., Piters, T. M., Gomez Ros, J. M. and
Delgado,, A. An intercomparison of glow curve analysis
computer programs: II. Measured glow curves. Radiat.
Prot. Dosim. 51, 257 264 (1994).

370

Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012

exponential heating function: I. Theory. J. Phys. D:


Appl. Phys. 39, 1500, (2006).
16. Balian, H. G. and Eddy,, N. W. Figure-of-merit
(FOM), an improved criterion over the normalized chisquared test for assessing goodness-of-fit of gamma-ray
spectral peaks. Nucl. Instrum. Methods 145(2),
389 395 (1977).
17. Billo, E. J. Excel for Scientists and Engineers. Wiley
(2007).
18. Fylstra, D., Lasdon, L., Watson, L. and Waren, A.
Design and use of the Microsoft Excel solver. Interfaces
28, 29 55 (1998).