Академический Документы
Профессиональный Документы
Культура Документы
363 370
Advance Access publication 15 July 2011
doi:10.1093/rpd/ncr315
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
# 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
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.
S I(T)fit
0.0003
0.0004
0.0005
jI(T)exp I(T)fitj
3.6E206
4.2E206
4.5E206
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
364
Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012
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
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
D. AFOUXENIDIS ET AL.
366
Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012
Peak 2
Peak 3
Peak 4
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
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
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
Peak 3
1316.279
0.914
3 318 872
1316.277
1.916
3 245 255
Deviations (%)
tmax (s)
0.086
Integral
21.806
0.000
2.218
0.000
20.858
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
368
Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012
Peak 5
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.
369
Downloaded from http://rpd.oxfordjournals.org/ at National Hellenic Research Foundation & National Documentation Centre on October 22, 2012
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