Spectrochimica Acta Part A 54 (1998) 159170

From observed to corrected luminescence intensity of solution

systems: an easy-to-apply correction method for standard
spectrouorimeters
Alberto Credi, Luca Prodi *
Dipartimento di Chimica G. Ciamician, Uni6ersita` di Bologna, 6ia Selmi 2, 40126 Bologna, Italy
Received 2 April 1997; accepted 10 September 1997
Abstract
Although spectrouorimetry is a very sensitive analytical technique, its use is limited by the non linear relationship
between the concentration of the analyte of interest and the generated electric signal. Here we discuss a very easy
correction method which takes into account all the instrumental factors and transforms the observed intensity value
in a quantity, the corrected luminescence intensity, that is directly proportional to the concentration of the observed
luminophore. At the end of the discussion, a general formula is presented together with two examples illustrating the
method and its application. 1998 Elsevier Science B.V.
Keywords: Photoluminescence spectroscopy; Spectrouorimeter; Correction method; Inner lter effects; Fluorescence
1. Introduction
Photoluminescence is one of the most versatile
and sensitive techniques for the detection of
chemical species in solution. Nowadays, standard
UV-VIS-NIR spectrouorimeters are low-cost,
compact instruments usually interfaced to a PC,
very easy to use and largely employed in research
and analytical laboratories. Luminescence spec-
troscopy has been nding application in growing
elds such as molecular electronics and, particu-
larly with the development of luminescent labels
and sensors, environmental sciences, medical diag-
nostics, and cell biology [1,2]. Unfortunately, lu-
minescence measurements are often not as easy as
they could seem at a rst glance [37]. In fact,
while the electric signal produced by a spec-
trophotometer represents a physical quantity (the
absorbance) that can be expressed in an absolute
scale, the electric signal produced by a spec-
trouorimeter is related to the total luminescence
intensity (i.e. to the number of emitted photons)
through a number of instrumental factors (inten-
sity of the exciting source, instrument optics, sig-
nal amplication) and to the solution
characteristics. The observed intensity signal can
be related to sample concentration only if cor-
rected in order to take into account these instru-
mental and solution factors. It should be stressed
* Corresponding author. Fax: +39 51 259456; e-mail:
lprodi@ciam.unibo.it
1386-1425/98/$19.00 1998 Elsevier Science B.V. All rights reserved.
PII S1386- 1425( 97) 00224- 2
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 160
that luminescence intensity measurements carried
out with standard spectouorimeters are never
absolute, and the intensity values must be ex-
pressed in a relative scale even after corrections.
As a matter of fact, the observed luminescence
intensity has no need of corrections only for very
dilute solutions, where absorption effects can be
ignored [5]. Although this condition can some-
times be achieved (an analyte with a high lumines-
cence quantum yielde.g. anthracenecan be
detected at concentrations as low as 10
10
M
1
),
more concentrated solutions are normally to be
examined. This is unavoidable if, for example, the
emission quantum yield of the luminescent species
is very low or an equilibrium is undesirably af-
fected by dilution. Other problems may be present
if the solution contains other chromophores ab-
sorbing at the excitation and/or emission wave-
lengths.
2. Scope and limitations
As already pointed out, luminescence measure-
ments can hide subtle artifacts. Here we deal with
quantitative steady-state luminescence intensity
determination in solution, at a xed emission
wavelength, using a commercial spectrouorime-
ter with right-angle excitation (perpendicular ge-
ometry). These types of measurements are
particularly important in analyte detection, titra-
tions, quenching and sensitization experiments,
photoreaction and photoluminescence quantum
yield determination, and whenever a luminescence
signal is used to monitor a chemical process.
The determination of the corrected emission
intensity of a luminophore from the luminescence
values measured in an experiment implies the
knowledge of: (i) the fraction of incident photons
absorbed by the emitting species; and (ii) the
fraction of the actually emitted light that is de-
tected by the instrument, i.e. the part of emitted
photons lost on the way to the photomultiplier.
The corrections we are introducing take into ac-
count all the factors that affect the measured
luminescence intensity, thus answering the above
questions.
We call correction factor J a number that,
multiplied by the luminescence intensity I
L
obs
ob-
served for the luminophore at a xed emission
wavelength u
em
upon excitation at u
ex
, gives a
value of the corrected luminescence intensity I
L
0
(at u
em
) that is independent on the experimental
conditions (except for the concentration of the
considered emitting species):
I
0
L
(u
em
) =J(u
ex
, u
em
) I
obs
L
(u
ex
, u
em
) (1)
Then if C
1
and C
2
are the concentrations of a
luminescent analyte in two solutions, whatever the
experimental conditions, and I
1
0
and I
2
0
are the
corresponding corrected luminescence intensities,
in absence of quenching and sensitization phe-
nomena [8], we obtain
I
0
1
(u
em
)
I
0
2
(u
em
)
=
C
1
C
2
(2)
This relationship implies that the corrected lu-
minescence intensity is, as we need, linearly corre-
lated to the luminophore concentration.
Moreover, possible quenching or sensitization
processes can be quantitatively assessed if C
1
and
C
2
are known (see below).
The factors described in this note are usually
grouped in one chapter called inner lter effects
[6]. For the sake of clarity we prefer to divide
them in two classes, referring to as inner lter
effects only those related to the coabsorption of
the exciting light and the reabsorption of the
emitted light. Therefore, the procedure that
should be followed to make corrections on ob-
served luminescence intensities has been divided in
two steps: (i) correction for the (non-linear) lu-
minescence intensity response of the instrument
versus absorbance; and (ii) correction for inner
lter effects (coabsorption of the excitation light
and reabsorption of the emitted light). Other
problems [37,9] such as resolution of the spec-
trum, phototube response, and interference of ex-
citation and emission harmonics are out of the
scope of this paper, and will not be discussed.
1
This value was determined by diluting an air-equilibrated
CH
3
CN solution of anthracene until its uorescence intensity
had become of the same order of magnitude of the instrumen-
tal noise (spectrouorimeter: Perkin-Elmer LS-50 with photo-
tube Hamamatsu R928; excitation wavelength 260 nm;
T=298 K).
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 161
3. Instrument calibration curve
Even in the simplest case, when only one spe-
cies is present in solution, the relationship be-
tween the observed luminescence intensity and the
concentration (i.e. the absorbance) of the solution
at the excitation wavelength is not linear. This is
due to two reasons, one mathematical, and the
other of geometrical nature.
The rst reason is trivial. The emitted light is
proportional to the number of the excited states
present in solution, which are proportional to the
light intensity absorbed by the analyte at the
excitation wavelength, I
a
(u
ex
). The intensity of
light absorbed by a pathlength b of solution con-
taining a substance of concentration C
L
and mo-
lar absorptivity m
L
(u
ex
) is given by the well-known
Lambert Beer law:
I
a
(u
ex
) =I
i
(u
ex
) I
t
(u
ex
)
=I
i
(u
ex
) I
i
(u
ex
) 10
m
L
(u
ex
) C
L
b
=I
i
(u
ex
) (110
m
L
(u
ex
) C
L
b
) (3)
where I
i
(u
ex
) and I
t
(u
ex
) are the incident and trans-
mitted light intensity, respectively. By denition,
the luminescence quantum yield (F
L
) is the ratio
between the emitted (I
L
(u
ex
)) and absorbed
(I
a
(u
ex
)) photons; then, the total luminescence in-
tensity is
I
L
(u
ex
) =F
L
I
a
(u
ex
)
=F
L
I
i
(u
ex
) (110
m
L
(u
ex
) C
L
b
) (4)
From Eq. (4) it is evident that the relationship
between I
L
and the luminophore concentration is
not linear. The exponential term in Eq. (4) can
be expanded, and orders higher than 1 can be
neglected if m
L
(u
ex
) C
L
b is reasonably small
( B0.1). Under this condition, Eq. (4) can be
approximated to Eq. (5):
I
L
(u
ex
) =ln(10) I
i
(u
ex
) F
L
m
L
(u
ex
) C
L
b (5)
The second factor determining the non-linearity
of the I
L
versus A response of the instrument is
related to the geometry of the slits/monochroma-
tors/detector ensemble. In fact, in commercial
spectrouorimeters the light emitted by the sam-
ple is not collected over the whole space sur-
rounding the cuvette, but only on a window in
front of the cell surface at 90 with respect to the
excitation beam. This means that the light de-
tected by the photomultiplier depends on where
the luminescence is originated inside the cuvette.
If the solution is sufciently dilute, we can assume
that the excitation beam is able to cross the entire
cuvette before being considerably attenuated, and
that the excited states are uniformly produced
along the excitation beam path. Under this condi-
tion, the luminescence intensity is related to the
absorbance by Eq. (4), and no geometrical effect
is present. As an example, the excited-states distri-
bution along the beam path in a solution with
A(u
ex
) =0.1 is represented by black bars in Fig. 1.
When the absorbance at the excitation wavelength
is high, the analyte absorbs and emits in the rst
layers of the solution. Empty bars in Fig. 1 show
the excited-states distribution along the beam
path in a solution whose absorbance at the excita-
tion wavelength is 2.0. Due to the fact that the
detector monitors the central part of the observed
surface, as usually happens with perpendicular
geometry, it loses most of the emitted light; this
leads to the paradox that the luminescence inten-
sity decreases with increasing luminophore con-
centration for high values of the absorbance at
u
ex
. It is then evident that for solutions with large
absorbance at the excitation wavelength (typi-
cally, A\0.1) the observed luminescence intensity
is affected by a geometrical factor that must be
Fig. 1. Representation of the excited-states distribution along
the beam path for a solution with A(u
ex
) =0.1 (black bars)
and A(u
ex
) =2.0 (empty bars).
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 162
Fig. 2. Plot of the experimental instrument calibration curve
() and of its tangent for A(u
ex
) 0 ( ).
the above described calibration curve, Eq. (4) can
be rewritten as
I
obs
L
(u
ex
) =F
L
I
i
(u
ex
) F
g
(A(u
ex
)) (6)
Some authors [12] suggest that the correction
factor J (Eq. (1)) can be obtained by calculating
the tangent T(A(u
ex
)) of the F
g
(A(u
ex
)) function
for A(u
ex
) 0
T(A(u
ex
)) =k A(u
ex
) =k m
L
(u
ex
) C
L
,
where k= lim
A(u
ex
) 0
dF
g
(A(u
ex
))
dA(u
ex
)
(7)
and taking
J(u
ex
) =
T(A(u
ex
))
F
g
(A(u
ex
))
(8)
Then, according to Eqs. (1) and (6), one obtains
I
0
L
(u
ex
) =
T(A(u
ex
))
F
g
(A(u
ex
))
I
obs
L
(u
ex
)
=F
L
I
i
(u
ex
) T(A(u
ex
)) (9)
and, on the basis of Eq. (7),
I
0
L
(u
ex
) =F
L
I
i
(u
ex
) k m
L
(u
ex
) C
L
(10)
Eq. (10) shows that the corrected luminescence
intensity I
L
0
(u
ex
) is linearly related to the lu-
minophore concentration through constant terms
(F
L
, I
i
(u
ex
), k, m
L
(u
ex
)), and satises Eq. (2). The
correction factor expressed in Eq. (8) can be
simplied by taking A(u
ex
) instead of T(A(u
ex
)).
Then Eq. (9) becomes
I
0
L
(u
ex
) =
A(u
ex
)
F
g
(A(u
ex
))
I
obs
L
(u
ex
)
=F
L
I
i
(u
ex
) m
L
(u
ex
) C
L
(11)
It should be noted that the luminescence intensi-
ties given by Eqs. (10) and (11) are proportional
through the constant term k. These intensity val-
ues, however, are completely equivalent since, as
stated in the introduction, they are expressed in
arbitrary scale.
From Eq. (11) it is clear that the total ab-
sorbance of the solution at u
ex
must be measured
for obtaining the corrected intensity. In some
analytical application, the absorbance of the solu-
tion can be too small to be measured with good
taken into account in the determination of the
corrected luminescence intensity. Obviously, any
species (not only the luminophore) absorbing at
the excitation wavelength enhances this effect.
While the rst reason of the non-linear I
L
ver-
sus A behavior could be accounted for by calcula-
tions, the geometrical factor has to be measured
since it depends on the instrument characteristics.
A simple correction formula has been proposed
[10] with the assumption that the luminescence is
originated in the central point of the cuvette. It is
recommended, however, to construct a calibration
curve for the instrument [5,6] by plotting the
luminescence intensity values (at a chosen u
em
)
obtained for several solutions of one emitting
species with different concentration (i.e. with dif-
ferent absorbance at u
ex
) versus the corresponding
absorbance at the excitation wavelength. Of
course, such a calibration graph depends on the
width of the window monitored by the detector,
i.e. on the emission slit width. A smaller depen-
dence of the calibration curve on the excitation
slit can also be observed. Fig. 2 shows the calibra-
tion graph obtained for a Perkin-Elmer LS-50
spectrouorimeter, which has been constructed by
measuring the luminescence intensity (u
em
=450
nm; bandpass=2.5 nm) of quinine sulphate in
H
2
SO
4
0.5 M [11] at different concentrations.
Notice that the relationship is practically linear
until A:0.1 (see Eq. (5) and above in the text).
Generally speaking, if F
g
(A(u
ex
)) is the function
containing both the mathematical and the geo-
metrical dependence of I
L
versus A, derived from
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 163
condence. In this case Eq. (11) cannot be used,
but, as discussed above and shown in Fig. 2, the
I
L
versus A trend is linear (Eq. (5)). This means
that luminescence intensities observed for solu-
tions whose absorbance at u
ex
lies in this region
can be directly compared, according to Eq. (2),
without any correction. This is also the reason
why a corrected excitation spectrum (i.e. in which
the emission intensity recorded at a xed u
em
is
proportional to the absorbance of the species, and
that can be compared with the absorption spec-
trum) is obtained only for solutions with low
absorbance (AB0.1) [13].
4. Inner lter effects
Among the chemical systems that can be stud-
ied by means of photoluminescence spectroscopy,
those where the species under examination is the
only one present in solution (besides the solvent)
are very few. For example, analytical chemistry
has usually to face complex matrices; in the eld
of supramolecular chemistry [14], that studies
multicomponent systems, the above condition is
never encountered. The presence of different sub-
stances in the same solution can affect the ob-
served luminescence intensity in different ways
(besides quenching and sensitisation phenomena),
namely by absorbing the incident light or by
reabsorbing the emitted light. For the same rea-
sons, correction of the observed luminescence in-
tensity values might be necessary also for a
molecule or a supramolecular system [14] that
contains many chromophoric and/or emitting
moieties.
4.1. Coabsorption of the exciting light
This phenomenon takes place when the incident
light is absorbed not only by the analyte, but also
by other chromophores present in solution, acting
as a lter at the excitation wavelength. If the
chromophores do not interact in the ground state,
the absorption spectrum of the mixture is exactly
the sum of the spectra of the separated species. At
the excitation wavelength u
ex
the fraction of light
F
L
(u
ex
) absorbed by the luminophore is given by
F
L
(u
ex
) =
m
L
(u
ex
) C
L
%
i
A
i
(u
ex
)
=
A
L
(u
ex
)
A(u
ex
)
(12)
where A
L
(u
ex
) is the absorbance of the lu-
minophore and A(u
ex
) is the total absorbance of
the solution at the excitation wavelength. The
term F
L
(u
ex
) of Eq. (12) should be introduced in
Eq. (6) when the coabsorption of the exciting light
is effective:
I
obs
L
(u
ex
) =F
L
I
i
(u
ex
) F
g
(A(u
ex
))
A
L
(u
ex
)
A(u
ex
)
(13)
The correction factor A(u
ex
)/F
g
(A(u
ex
)) introduced
in Eq. (11), however, takes already into account
this effect. In fact, by substituting in Eq. (11) the
above expression of I
L
obs
(u
ex
) (Eq. (13)), one ob-
tains
I
0
L
(u
ex
)
=
A(u
ex
)
F
g
(A(u
ex
))
F
L
I
i
(u
ex
) F
g
(A(u
ex
))
A
L
(u
ex
)
A(u
ex
)
=F
L
I
i
(u
ex
) m
L
(u
ex
) C
L
(14)
Eq. (14) demonstrates that the luminescence in-
tensity corrected with the factor A(u
ex
)/F
g
(A(u
ex
))
is effectively proportional to the concentration of
the examined luminophore even in case of coab-
sorption of exciting light.
If some of the component of the solution do
interact with each other, the absorption spectrum
of the mixture may not be equal to the sum of the
spectra of the isolated species; in the worst cases,
when there is a strong interaction between the
chromophores in the ground state (e.g. charge-
transfer interaction), new bands appear or some
previously existing bands are modied in the ab-
sorption spectrum of the mixture [14]. In this
regard, we would like to notice that the lumines-
cence intensity values measured for different mul-
ticomponent solutions and corrected with Eq. (14)
can be compared only if m
L
(u
ex
) remains substan-
tially unperturbed. Thus the excitation should be
performed where the modication in the absorp-
tion spectrum are small.
It should be noted that the calculations done so
far are valid for any emission wavelength since the
mathematical and geometrical factors con-
tained in the F
g
(A(u
ex
)) function do not alter the
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 164
spectral dispersion characteristics of the lu-
minophore emission detected by the spec-
trouorimeter.
4.2. Reabsorption of the emitted light
In luminescence measurements it is rather com-
mon that the absorbance of the solution in the
spectral region of the luminophore emission is not
zero. In this case, part of the light emitted by the
luminophore is reabsorbed by the luminophore
itself and/or by other species. This problem is of
great importance in the determination of lumines-
cence quantum yields and of the correct shape of
a luminescence band [37]. Of course, in our xed
emission wavelength measurements, the best pre-
caution is to read the luminescence intensity at a
wavelength where the absorbance of the solution
is zero. This favorable situation, however, is
sometimes impossible to achieve, and corrections
for the reabsorption of the emitted light must be
done. The reabsorption effect depends on the
geometry of the cell, that in our case (perpendicu-
lar geometry and vertical slits) is that depicted in
Fig. 3.
The excitation beam, whose thickness will be
considered negligible with respect to the width of
the cell, incises perpendicularly to the excitation
window of the cuvette and passes through the
solution keeping parallel to the emission window.
We can therefore consider that the luminophore
emission takes place from a narrow strip of solu-
tion. The pathlength that the emitted photons
have to cross for getting out of the cuvette is not
unique and depends on the emission optics (Fig.
3). According to the Lambert Beer law, the frac-
tion of emitted light transmitted through the
pathlength b at the wavelength u
em
is given by
T
L
(u
em
) =10
A(u
em
) b
(16)
where A(u
em
) is the absorbance of the solution per
unit length at the monitored emission wavelength.
Thus, we obtain
I
obs
L
(u
ex
, u
em
) =I
0%
L
(u
ex
, u
em
) T
L
(u
em
) (17)
I
0%
L
(u
ex
, u
em
) =
I
obs
L
(u
ex
, u
em
)
T
L
(u
em
)
(18)
where I
L
0%
is the luminescence intensity free from
reabsorption effects (but has to be corrected for
the other effects, see above), and 1/T
L
(u
em
) is the
correction factor for the reabsorption of the emit-
ted light, or what we call the emission inner lter
(EIF) correction.
As one can see from Eq. (16), the b value must
be known in order to calculate the EIF correction.
This parameter is averaged over all possible path-
lengths, and would correspond to the geometri-
cal b only if parallel beams were detected (Fig. 3).
In most instruments with perpendicular geometry
the cell is illuminated centrally; then, in case of 1
cm1 cm cuvette, the geometrical b is supposed
to be 0.5 cm. With this assumption, Lakowicz has
proposed a simple correction formula [10] that,
however, gives only a rough estimate of the reab-
sorption effect. The value of the distance b can be
estimated with good condence by means of a
simple experiment. What is needed is a solution
containing the luminophore and a non-interacting
species absorbing in the spectral region where the
luminophore emits. The luminophore is excited,
and the luminescence intensity is measured at a
wavelength where the second species absorbs; the
measurement is repeated for solutions with in-
creasing concentration of absorbing species (in-
creasing absorbance at u
em
). The observed
intensity values have to be corrected for the in-
strumental response and for the coabsorption of
the exciting light (see Sections 3 and 4), i.e. for all
effects but internal emission lter. According to
Eqs. (16) and (17), from the plot of the logarithm
of these partially corrected luminescence values
versus the absorbance of the solution at u
em
one
Fig. 3. Schematic representation of a spectrouorimeter with
perpendicular geometry and vertical slits.
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 165
Fig. 4. Absorption spectra of naphthalene ( ) and an-
thracene (dotted line), and uorescence spectrum of naph-
thalene (; u
ex
=311 nm).
Another problem related to the reabsorption of
the emitted light is the secondary emission from
the examined luminophore. In fact, if there is
overlap between the absorption and luminescence
spectra of the luminophore, a certain amount of
its luminescence is reabsorbed by itself and then
reemitted, and so on. The secondary emission is
important for concentrated solutions of strongly
luminescent molecules with large overlap between
absorption and emission spectra. However,
whereas this phenomenon is important in the
determination of luminescence quantum yields [7],
it can almost always be eliminated in xed-wave-
length luminescence measurements by selecting an
emission wavelength where the luminophore does
not absorb, since the overlap in this case can
never be complete. A more complicated situation
is presented when a chromophore absorbing at u
ex
is luminescent, and its emission spectrum overlaps
the absorption spectrum of the examined lu-
minophore. Then, one of the light absorbed by
the inner lter at u
ex
will be emitted and re-ab-
sorbed by the luminophore (a case of trivial
energy transfer phenomenon). The nal result is
that part of the light not directly absorbed by the
luminophore actually feeds the luminophore emis-
sion. This effect is difcult to take into account
because the amount of emitted light reabsorbed
by the luminophore is rather complicated to esti-
mate; however, if the luminophore absorbance in
the region where the inner lter molecule emits is
very high, one can consider that all the emitted
light is reabsorbed by the luminophore. In this
limit case F
L
(u
ex
) would become
F
L
(u
ex
) =
A
L
(u
ex
) +A
if
(u
ex
) F
if
A(u
ex
)
(19)
where A
if
(u
ex
) is the absorbance of the emitting
species acting as inner lter and F
if
is its lu-
minescence quantum yield.
We would like to stress, however, that correc-
tions of the emission inner lter effect are always
critical and introduce a rather high error on the
corrected luminescence values, mainly because of
the exponential relationship between I
L
0
% and
A(u
em
) (Eqs. (16) and (18)) and of the uncertainty
on the b value. For example, if we take A(u
em
) =
2.0 and b=0.5 cm, according to Eqs. (16) and
should obtain a linear correlation, whose slope is
the actual width of the layer of solution that the
emitted light must cross to exit the cuvette. In
principle, a dependence of b on A(u
ex
) can be
expected; however, for a wide range of ab-
sorbance (0BA(u
ex
) B2.8), we did not nd any
substantial change in the reabsorption pathlength.
In our experiment we have used naphthalene (2
10
4
M) as luminophore and anthracene (from 0
to 10
3
M) as an emission inner lter, in CH
3
CN
solution at room temperature. The excitation was
done at 311 nm, and the luminescence intensity
was read at 346 nm (see Fig. 4). We have also
veried that these species do not mutually interact
in our experimental conditions. The tting of the
experimental data is shown in Fig. 5.
Fig. 5. Plot of the logarithm of the naphthalene luminescence
intensity at 346 nm, corrected for the instrument response and
for the coabsorption of exciting light vs. the absorbance of the
solution at the same wavelength.
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 166
(18) we nd that the EIF correction 1/T
L
(u
em
) is 10;
if b=0.52 cm, the EIF correction becomes 11. This
means that a 4% error on the b value causes a 10%
error on the corrected luminescence intensity. The
reabsorption of the emitted light should always be
minimized with a wise choice of the monitored
emission wavelength; data corrected using Eqs.
(16) and (18) in a condition of strong absorption
at the emission wavelength (typically, A(u
em
) \
1.5) should be very carefully examined.
5. General formula
The general equation we propose for obtaining
a value of the luminescence intensity I
L
0
(u
em
) that
is proportional to the luminophore concentration
from the observed luminescence intensity
I
L
obs
(u
ex
, u
em
) is:
I
0
L
(u
em
) =
A(u
ex
)
F
g
(A(u
ex
))

1
T
L
(u
em
)
I
obs
L
(u
ex
, u
em
)
(20)
In Eq. (20), the rst term A(u
ex
)/F
g
(A(u
ex
)) repre-
sents the correction factor for the non-linear in-
strumental calibration curve described in Section 3.
It also takes into account the coabsorption of the
exciting light by other chromophores, when
present (Section 4.1). The second term 1/T
L
(u
em
) is
the correction factor for the reabsorption of the
emitted light at the monitored emission wave-
length, if any, by the chromophores present in
solution (Eq. (16), Section 4.2).
The propagation of errors on Eq. (20) gives the
following expression for the relative error on the
corrected luminescence intensity:
the symbols used have the same meaning as in Eq.
(20) and preceding ones; for the sake of clarity, the
identications (u
ex
) and (u
em
) have been omitted
where not explicitly required. The error on the
function F
g
(A(u
ex
)) is related to lA through its
derivative with respect to A(u
ex
). It is worth noting
that the largest contribution to the error on the
corrected luminescence intensity is usually given by
that on the emission inner lter correction (third
and fourth terms in Eq. (21); see also Section 4.2).
6. Examples
6.1. Experimental
Anthracene and pyrene were purchased from
Fluka and used as received. 1,5-dimethoxynapth-
talene, 2,7-dibenzyldiazapyrenium (PF
6

salt) and
1,5-dinaphtho-38-crown-10 were available from
previous investigations [15,16]. All the experiments
were carried out in acetonitrile (Merck Uvasol)
solutions at 298 K. Absorption and luminescence
spectra were recorded with a Perkin-Elmer u6
spectrophotometer and a Perkin Elmer LS-50 spec-
trouorimeter, respectively. Experimental errors:
absorption and emission maxima, 92 nm; ab-
sorbance, 90.002 units; observed luminescence
intensity, 5%. Under the experimental conditions
described below, according to Eq. (21), the upper
limiting value for the error on the corrected lu-
minescence intensity was estimated to be 10%.
6.2. Determination of different analytes in a
single matrix
The following example shows how the spectro-
uorimetric techniques can be used for the deter-
mination of analyte concentrations in a com-
plex matrix. We prepared a CH
3
CN solution con-
taining three different luminophores, namely 1,5-
dimethoxynaphthalene (2.410
5
M), anthra-
cene (1.010
5
M), and pyrene (1.010
5
M).
Matrices with several polycyclic aromatic hydro-
carbons are often encountered in analytical
chemistry [17]. In general, when several chro-
mophores are present in the same sample, at least
a partial overlap exists both in the absorption
and in the emission spectra of the examined com-
pounds. As it can be seen from Fig. 6 (dashed
lI
0
L
I
0
L
=
' lA
A(u
ex
)
+
lF
g
F
g

2
+[ln(10) b lA]
2
+[ln(10) A(u
em
) lb]
2
+
lI
obs
L
I
obs
L

2
(21)
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 167
Fig. 6. Luminescence spectra (dilute CH
3
CN solution, 298 K)
of 1,5-dimethoxynaphthalene ( ; u
ex
=327 nm), pyrene
(; u
ex
=337 nm), and anthracene ( ; u
ex
=378 nm).
solution, since irradiation at u\375 nm leads to
the excitation of only this species. From the ex-
perimental parameters for the sample and refer-
ence (anthracene 2.010
5
M) solutions (Table
3), a concentration of 9.810
6
M is found for
anthracene.
The concentrations determined with the spec-
trouorimetric method are in good agreement
with the actual values, showing that spec-
trouorimetry can provide a method for the reso-
lution of complex matrices. Of course, the
detection limit of a species depends ultimately on
its luminescence quantum yield. In this regard, we
would like to stress that the more dilute the
solution, the less important the correction factors.
6.3. Determination of the formation constant of a
[2]pseudorotaxane
[2]Pseudorotaxanes are supramolecular (multi-
component) systems comprising one macrocyclic
ring threaded by an acyclic component [18]. Their
most important feature is that changes in the
relative position of their molecular components
can be induced by external stimuli [15,19]; this
could be of interest for processing and storing
information at the molecular level [20,21].
The [2]pseudorotaxane 1 [17] is obtained in
CH
3
CN solution by self-assembling of the 2,7-
dibenzyldiazapyrenium dication 2 with the crown
ether 3 based on 1,5-dioxynaphthalene units (Fig.
7). The main driving force of this self-assembly
process is a charge-transfer (CT) interaction from
the y-electron donor naphthoxy units of 3 to the
y-electron acceptor diazapyrenium group of 2.
Upon formation of the [2]pseudorotaxane 1,
the CT interaction leads to: (i) the appearance of
a weak and broad absorption band in the visible
region (Fig. 8); and (ii) the disappearance of the
luminescence characteristic of the two compo-
nents (Fig. 8, inset). The latter result is due to the
presence of low-energy charge-transfer excited
states which offer fast radiationless decay of the
upper-lying luminescent states of the molecular
components 2 and 3 [15].
The complete quenching of the luminescence of
the molecular components when they are threaded
in the pseudorotaxane structure can be used to
line), 1,5-dimethoxynaphthalene has the highest
energy emission. This means that there is a region
(330365 nm) where its emission spectrum does
not overlap those of the other two components of
the solution. We decided therefore to excite the
sample solution at 327 nm, where the 1,5-
dimethoxynaphthalene has an absorption maxi-
mum, and to measure the luminescence intensity
at 360 nm. The same measurement has been per-
formed on a reference solution containing C
R
=
2.910
5
M 1,5-dimethoxynaphthalene only.
The data and correction factors used for obtain-
ing the corrected intensity emission (Eq. (20)) are
listed in Table 1 (a value of b=0.85 was used in
Eq. (16)).
From Eq. (2) the concentration C
S
of the 1,5-
dimethoxynaphthalene in the sample is
C
S
=
I
0
L,S
(360)
I
L,R
(360)
C
R
=2.510
5
M (22)
Moving to lower energy in the luminescence spec-
trum, we nd the band of pyrene (Fig. 6, full line),
whose highest energy feature peaks at 373 nm. If
we excite the solution at a wavelength (u\330
nm) where the 1,5-dimethoxynaphthalene does
not absorb, the emission intensity observed at 373
nm is due only to the pyrene component. From
the data listed in Table 2 (the reference solution in
this case contains 3.210
5
M pyrene only) we
calculated a pyrene concentration in the sample
solution of 1.010
5
M. Finally, it is easy to
nd the concentration of anthracene in the sample
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 168
Table 1
Data for correction of the luminescence intensity of 1,5-dimethoxynaphthalene
I
L
obs
(327, 360) A(360) T
L
(360) I
L
0
(360) A(327) F
g
(327)
30.3 0.063 0.88 Sample 69.5 0.316 0.82
0.001 1.00 Reference 0.168 0.57 35.3 120
Table 2
Data for correction of the luminescence intensity of pyrene
T
L
(373) I
L
0
(373) A(337) F
g
(337) I
L
obs
(337,373) A(373)
0.040 0.92 Sample 0.578 0.98 46.8 73.0
0.095 150.6 0.83 Reference 69.9 1.431 0.80
Table 3
Data for correction of the luminescence intensity of anthracene
I
L
obs
(378, 402) A(402) T
L
(402) A(378) I
L
0
(402) F
g
(378)
9.7 0.000 1.00 Sample 40.2 0.072 0.30
0.000 1.00 Reference 0.147 0.52 19.8 70.0
estimate their association constant. In fact, the
residual luminescence of 2 and 3 in a solution
containing both these species is related to the
concentration of free 2 and 3, respectively,
through Eqs. 23, that are directly derived from
Eq. (2):
I
0
2
(u
em
)
I
0
2,i
(u
em
)
=
C
2
C
2,i
I
0
3
(u
em
)
I
0
3,i
(u
em
)
=
C
3
C
3,i
(23)
where C
2
and C
3
are the concentration of the
free (i.e. unthreaded) molecular components, and
C
2,i
and C
3,i
are the total (i.e. analytical) concen-
tration. Of course, C
2,i
and C
3,i
are known from
the preparation of the solution. I
2,i
0
(u
em
) and
I
2
0
(u
em
) are the corrected luminescence intensities
of 2 in a solution containing 2 alone at a concen-
tration of C
2,i
and in a solution containing C
2,i
2
and C
3,i
3, respectively. I
3,i
0
(u
em
) and I
3
0
(u
em
) are
Fig. 8. Absorption spectrum (CH
3
CN, 298 K) of 5.010
5
M 2 ( ), of 6.410
5
M 3 ( ), and of their mixture
(). Note the charge-transfer band (u
max
=510 nm) due to
formation of the pseudorotaxane 1. The inset shows the ob-
served luminescence spectra of the above mentioned solutions
of 2 (u
ex
=371 nm; ) and 3 (u
ex
=295 nm; ).
Fig. 7. Self-assembly of the 2,7-dibenzyldiazapyrenium dica-
tion (2) and the crown ether 3 to give the pseudorotaxane 1.
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 169
Table 4
Data for correction of the luminescence intensity of 2
T
L
(453) I
L
0
(453) I
L
obs
(371, 453) A(453) A(371) F
g
(371)
0.98 69.0 5.010
5
M 2 0.123 0.46 273 0.012
37.5 0.076 5.010
5
M 2+6.410
5
M 3 0.86 0.123 11.7 0.46
Table 5
Data for correction of the luminescence intensity of 3
A(344) T
L
(344) A(295) F
g
(295) I
L
obs
(295, 344) I
L
0
(344)
357 0.98 0.008 6.410
5
M 3 287 1.121 0.92
0.481 0.39 6.410
5
M 3+5.010
5
M 2 1.190 0.90 37.3 126
the corrected luminescence intensities of 3 in a
solution containing 3 alone at a concentration of
C
3,i
and in a solution containing C
3,i
3 and C
2,i
2,
respectively. The thermodynamic constant of the
association equilibrium between 2 and 3 is
K
ass
=
C
1
C
2
C
3
=
C
1
(C
2,i
C
1
)(C
3,i
C
1
)
(24)
where C
1
is the concentration of pseudorotaxane.
It should be noticed that it is sufcient to
measure the luminescence quenching of only one
component in order to calculate K
ass
: the concen-
tration C
1
of pseudorotaxane can in fact be evalu-
ated independently as both C
2,i
C
2
and
C
3,i
C
3
. In this case we are discussing, both
components of the pseudorotaxane are lumines-
cent, and K
ass
can be estimated from two indepen-
dent series of luminescence measurements. The
comparison of the results of these two experi-
ments, which of course must be consistent, can be
useful for checking the validity of the corrections
that have been made. As one can see from Fig. 8,
the absorption spectrum of the solution contain-
ing 1 and its unthreaded components 2 and 3 is
rather complex; in particular, it is impossible to
excite 3 selectively, and luminescence of both 2
and 3 (Fig. 8, inset) is reabsorbed by the solution.
Therefore, considerable corrections of the ob-
served luminescence intensities are necessary.
In our experiment, C
2,i
and C
3,i
were 5.010
5
and 6.410
5
M, respectively; emission of 2 was
observed at 453 nm upon excitation at 371 nm,
whereas emission of 3 was observed at 344 nm
upon excitation at 295 nm. The observed ab-
sorbance and luminescence intensity values are
gathered in Tables 4 and 5 along with the correc-
tion parameters and the corrected luminescence
intensity calculated according to Eq. (20).
From the measurement of the luminescence in-
tensity characteristic of 2 (Table 4) one obtains
C
2
=
I
0
2
(453)
I
0
2,i
(453)
C
2,i
=
11.7
69.0
5.010
5
=8.510
6
M (25)
C
1
=C
2,i
C
2
=4.1510
5
M (26)
C
3
=C
3,i
C
1
=2.2510
5
M (27)
K
ass
=
C
1
C
2
C
3
=2.210
5
M
1
(28)
while the measurement of the luminescence inten-
sity characteristic of 3 (Table 5) leads to
C
3
=
I
0
3
(344)
I
0
3,i
(344)
C
3,i
=
126
357
6.410
5
=2.2610
5
M (29)
C
1
=C
3,i
C
3
=4.1410
5
M (30)
C
2
=C
2,i
C
1
=8.610
6
M (31)
K
ass
=
C
1
C
2
C
3
=2.110
5
M
1
(32)
The values of pseudorotaxane concentration so
determined (Eqs. (26) and (30)) are substantially
A. Credi, L. Prodi / Spectrochimica Acta Part A 54 (1998) 159170 170
identical, indicating that the correction factor is
reasonable. This example demonstrates that lu-
minescence spectroscopy can provide a simple and
quick method for studying equilibria in high-dilu-
tion conditions.
7. Conclusion
In this article we have discussed the relevant
problems related to quantitative luminescence
measurements in solution using standard spec-
trouorimeters with perpendicular geometry. An
expression for the correction of luminescence data
(Eq. (20)) has been proposed, and two examples
of application of our method have been illus-
trated.
This correction method is very easy to apply
since it requires the measurement of only three
quantities: (i) the absorbance of the solution at
the excitation wavelength; (ii) the absorbance of
the solution at the monitored emission wavelength
and, of course; (iii) the luminescence intensity.
The instrument calibration graph (Fig. 2) and the
reabsorption pathlength b (Eq. (16) and Fig. 5)
must be determined, but once they have been
obtained, they are valid for the entire lifetime of
the instrument (unless changes in the geometry of
the instrument occur). A simple computer pro-
gram containing the expression of the correction
factors, including the instrument calibration
graph, would allow the systematization of these
calculations. We believe that the correction steps
presented here give a simple answer to most prob-
lems related to spectrouorimetric techniques,
which are scarcely treated in a systematic way in
the literature, and sometimes underevaluated by
users.
Acknowledgements
This work was supported by the Ministero del-
lUniversita` e della Ricerca Scientica e Tecnolog-
ica. We are indebted to Professors Luca Moggi,
Vincenzo Balzani, Maria Teresa Gandol and to
Dr Roberto Ballardini, who encouraged us to
write this contribution. We also thank all the
members of the Bologna Photochemistry Group
for carefully reviewing our manuscript.
References
[1] A.W. Czarnik (Ed.), Fluorescent Chemosensors for Ion
and Molecule Recognition, ACS, Washington, 1992.
[2] A. Mayer, S. Neuenhofer, Angew. Chem. Int. Ed. Engl.
33 (1994) 1044.
[3] J.N. Miller (Ed.), Standards in Fluorescence Spectrome-
try, Chapman and Hall, London, 1981.
[4] K.L. Mielenz (Ed.), Optical Radiation Measurements,
Academic, New York, 1982.
[5] D. Rendell, Fluorescence and Phosphorescence, Wiley,
Chichester, 1987.
[6] D.A. Harris, C.L. Bashford (Eds.), Spectrophotometry
and Spectrouorimetry, IRL, Oxford, 1987.
[7] J.N. Demas, G.A. Crosby, J. Phys. Chem. 75 (1971) 991.
[8] N.J. Turro, Modern Molecular Photochemistry, Ben-
jamin/Cummings, Menlo Park, 1978.
[9] J.W. Hofstraat, M.J. Latuhihin, Appl. Spectrosc. 48
(1994) 436.
[10] J.R. Lakowicz, Principles of Fluorescence Spectroscopy,
Plenum Press, New York, 1983, pp. 4447.
[11] S.R. Meech, D. Phillips, J. Photochem. 23 (1983) 193.
[12] J.F. Holland, R.E. Teets, A. Timnick, Anal. Chem. 45
(1973) 145.
[13] J.N. Miller (Ed.), Standards in Fluroescence Spectrome-
try, Chapman and Hall, London, 1981, Ch. 7; K.L.
Mielenz (Ed.), Optical Radiation Measurements, Aca-
demic, New York, 1982, Ch. 3.
[14] V. Balzani, F. Scandola, Supramolecular Photochemistry,
Horwood, Chichester, 1991.
[15] R. Ballardini, V. Balzani, A. Credi, M.T. Gandol, S.J.
Langford, S. Menzer, L. Prodi, J.F. Stoddart, M. Venturi,
D.J. Williams, Angew. Chem. Int. Ed. Engl. 35 (1996)
978.
[16] P.R. Ashton, R. Ballardini, V. Balzani, A. Credi, M.T.
Gandol, S. Menzer, L. Perez-Garc a, L. Prodi, J.F.
Stoddart, M. Venturi, A.J.P. White, D.J. Williams, J.
Am. Chem. Soc. 117 (1995) 11171.
[17] T. Vo-Dish (Ed.), Chemical Analysis of Polycyclic Aro-
matic Compounds, Wiley, New York, 1989.
[18] D.B. Amabilino, J.F. Stoddart, Chem. Rev. 95 (1995)
2725.
[19] P.R. Ashton, R. Ballardini, V. Balzani, S.E. Boyd, A.
Credi, M.T. Gandol, M. Go mez-Lo pez, S. Iqbal, D.
Philp, J.A. Preece, L. Prodi, H.G. Ricketts, J.F. Stoddart,
M.S. Tolley, M. Venturi, A.J.P. White, D.J. Williams,
Chem. Eur. J. 3 (1997) 152.
[20] A.C. Benniston, Chem. Soc. Rev. 25 (1996) 427; V.
Balzani, A. Credi, F. Scandola, Chim. Ind. (Rome) 79
(1997) 751.
[21] A. Credi, V. Balzani, S.J. Langford, J.F. Stoddart, J. Am.
Chem. Soc. 119 (1997) 2679.
.