Applied Catalysis A: General 402 (2011) 218–223

Contents lists available at ScienceDirect

Applied Catalysis A: General

journal homepage: www.elsevier.com/locate/apcata

A simple visible spectrum deconvolution technique to prevent the artefact

induced by the hypsochromic shift from masking the concentration of
methylene blue in photodegradation experiments
Gregorio Marbán ∗ , Tan T. Vu, Teresa Valdés-Solís
Instituto Nacional del Carbón (CSIC), c/Francisco Pintado Fe 26, 33011 Oviedo, Spain

a r t i c l e i n f o a b s t r a c t

Article history: This paper analyzes the experimental evaluation of methylene blue concentration applied during photo-
Received 6 May 2011 catalytic experiments. In most studies a satisfactory evaluation of the concentration of methylene blue by
Received in revised form 6 June 2011 means of visible spectroscopy is masked by the presence of intermediates during the reaction, resulting
Accepted 8 June 2011
in an overestimation of the amount of methylene blue present. The deconvolution technique described in
Available online 15 June 2011
the present study solves this problem by estimating the contribution by intermediates to the visible spec-
trum. In the photodegradation of methylene blue by Degussa P25 TiO2 particles the main intermediate
Keywords:
detected in the visible spectral region was azure A, produced from the gradual demethylation of azure B.
Methylene blue
Photocatalysis
The methylene blue concentrations obtained if the artefact is not taken into account give lower values for
TiO2 the kinetic rate constants of photocatalytic degradation than those provided by this novel deconvolution
Visible spectroscopy technique.
Artefact © 2011 Elsevier B.V. All rights reserved.
Deconvolution

1. Introduction degradation process. It is known that this shift is produced by

Semiconductor metallic oxides such as TiO2 or ZnO are typically
used in the photocatalytic degradation of organic compounds in
water [1–4]. To test the photocatalysts in laboratory experiments,
dyes such as methylene blue are often employed [5–9] because
they represent a sizeable fraction (over 0.7 million tons) of the
organic compounds produced each year, whose disposal in efflu-
ent water streams has a considerable impact on the environment.
In most experimental strategies the concentration of methylene
blue is analyzed by applying absorption spectroscopy to the visible
spectral region of liquid samples from which the catalyst particles
have been previously separated. The relation between absorbance
and concentration is obtained by calibrating with methylene blue
solutions of different concentrations. To analyze the concentra-
tion of methylene blue during photodegradation experiments some
authors use the absorbance at the maximum absorption wave-
length though most researchers employ the absorbance at a fixed
wavelength, typically in the 664–668 range, which corresponds
to the maximum absorption wavelength for unreacted methylene
blue. This procedure is chosen because the maximum absorption
wavelength shifts to lower values (hypsochromic shift) during the
∗ Corresponding author. Tel.: +34 985119090; fax: +34 985297662.
E-mail address: greca@incar.csic.es (G. Marbán).
the formation of reaction intermediates that present absorption
maxima at lower wavelengths than methylene blue [10]. Thus,
according to some authors [11–13], the photocatalytic degradation
of methylene blue by TiO2 takes place via the sequential decompo-
sition of methyl groups, through the formation of azure B and azure
A (also azure C, thionine and phenothiazine). To take into account
this phenomenon some researchers use liquid chromatography to
separate the intermediate products (e.g. [11]), and then analyze
them by means of mass or visible spectrometry. Other authors rec-
ognize the formation of these intermediates but do not apply the
necessary corrections when evaluating the concentration of methy-
lene blue by means of visible spectroscopy (e.g. [9,12,14]). Most
authors however simply disregard the existence of intermediates
(e.g. [5,6,8]) and evaluate the concentration of methylene blue from
visible spectra that are almost certainly composed of spectra of dif-
ferent compounds. The formation of intermediate compounds in
methylene blue photodegradation can also be inferred from the
shape of the kinetic curves representing methylene blue concen-
tration decay, in which a shoulder can be observed at intermediate
reaction times (e.g. Fig. 6 in [15] (TiO2 –Cu films), Fig. 7 in [16]
(ZnO), Figs. 4 and 5 in [17] (Zn1−x Cux S)), a feature that is not men-
tioned in the works previously mentioned. Thus, in most cases
the methylene blue concentration evaluated by visible absorption
is wrongly estimated because of the presence in the spectra of
peaks corresponding to intermediate compounds. Yogi et al. [10]

0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2011.06.009
 G. Marbán et al. / Applied Catalysis A: General 402 (2011) 218–223 219

tried to deal with this artefact in their analysis of the decompo-

sition of methylene blue by Au–TiO2 films by means of visible
spectroscopy. They started from the premise that the spectra at
different reaction times resulted from the absorption of several
species (methylene blue, azure A, azure B, azure C and thionine)
and that the absorbance of the multicomponent system would be
equal to the sum of the absorbances of each component. By solv-
ing the resulting equation system (3125 simultaneous equations
based on the Beer–Lambert law) they were able to estimate the
evolution of the concentration of the different species with the
reaction time. This procedure yielded some very interesting results
while at the same time it revealed the following flaws: the varia-
tion in the amount of dimeric methylene blue with the methylene
blue concentration was not considered; the azure C concentration
was estimated to be negative; and only the evolution of the species
inputted into the multicomponent system can be followed.
The dimeric form of methylene blue is apparent in the vis-
ible spectrum of methylene blue in the form of a shoulder at
∼616 nm [18]. The specific form of the equilibrium constant for the
dimer formation is K = (1 − X)/(2CX2 ), where X is the fraction of dye
molecules existing in monomeric form and C is the total concen-
tration of methylene blue that can be calculated as C = Cmon + 2Cdim .
This implies that the concentration of the dimeric form of methy-
lene blue increases with the increase in the total concentration
of methylene blue. However, some authors [19] consider that the Fig. 1. Visible spectra of methylene blue solutions in water (7.0 and 2.3 mg/L). P1 to
shoulder at ∼616 nm cannot be produced by the presence of the P6 are the peaks obtained by Gaussian deconvolution.
dimer since, according to their calculations, its concentration in
water is always about 106 times lower than that of the monomer to any reaction in which the concentration of the main species
(Fig. 2 in [19]). We suspect that these authors used the “dissocia- is evaluated by absorption spectroscopy. The calibration of the
tion” constant evaluated in [18] as a “dimerization” constant and, methylene blue concentration in aqueous solution was performed
because of this “misuse”, they obtained such an odd result. In fact, in the 0–12 mg/L range. Fig. 1 shows the visible spectra of the 7.0
if the dissociation constant provided in [18] is applied, the percent- and 2.3 mg/L solutions. To perform a Gaussian deconvolution of the
age of dimer in solution for an initial methylene blue concentration spectrum a linear baseline between the lowest point at the left of
of 10 mg/L is quite high (9.9%). the spectrum and the point at 794 nm was set. Six Gaussian peaks
In this study we propose a deconvolution procedure that (P1 to P6 in Fig. 1) were found to be necessary to obtain a cor-
permits an unambiguous evaluation of the methylene blue con- rect deconvolution of the original spectra. The convolved curves
centration during photodegradation experiments, allowing the match almost exactly the experimental curves, as can be appre-
absorption peaks ascribed to intermediate products to be evaluated ciated in Fig. 1. P1 and P2 are the main peaks that represent the
without the need for previous assumptions concerning peak posi- monomeric and dimeric methylene blue species, respectively [18].
tions and taking into account the variation of dimeric methylene Table 1 shows all of the parameters obtained from the Gaussian
blue with the concentration of methylene blue in solution.
Table 1
2. Experimental Gaussian parameters for P1 to P6 peaks (Fig. 1) obtained from the deconvolution of
the visible spectra for methylene blue solutions of different concentrations.

The photocatalytic methylene blue degradation experiment was Peak parameters Methylene blue concentration (mg/L)
carried out with 20 mg of commercial TiO2 particles (Degussa P25)
11.7 9.3 7.0 4.7 2.3
in a 400 mL quartz beaker under the illumination of two ring-type
1 (nm)a
666.0
UV 22 W lamps (Luzchem Ring-Illuminator) which predominantly
2 (nm) 615.9
emit radiation at 351 nm. The particles were introduced in 60 mL 3 (nm) 529.1
of an aqueous solution of methylene blue, of an initial concentra- 4 (nm) 564.5
tion of 10 mg/L. The suspension was magnetically stirred for 30 min 5 (nm) 642.9
under darkness to ensure the adsorption/desorption equilibrium 6 (nm) 708.3

between the dye and the photocatalyst. Next the suspension was 1 (nm)b 14.7
exposed to the UV lamp. Samples were extracted for measurement 2 (nm) 23.4
3 (nm) 43.0
after various reaction times, and centrifuged to remove the parti-
4 (nm) 15.4
cles before analysis. The visible absorption peaks of the analyzed 5 (nm) 9.6
samples were recorded in the 400–800 nm range by means of a 6 (nm) 11.7
UV–vis spectrometer (Shimadzu UV-2401PC).
f1 c 0.396 0.414 0.427 0.439 0.447
f2 0.443 0.429 0.417 0.406 0.396
f3 0.063 0.059 0.056 0.052 0.050
3. Deconvolution procedure
f4 0.035 0.035 0.035 0.035 0.035
f5 0.043 0.047 0.052 0.057 0.062
3.1. Methylene blue calibration f6 0.011 0.011 0.011 0.011 0.011
a
Mean wavelength for the Gaussian distribution.
The procedure described below has been tested with the methy- b
Standard deviation.
c
lene blue photodegradation reaction. However it can be applied Peak area for the normalized spectrum (fi = 1).
220 G. Marbán et al. / Applied Catalysis A: General 402 (2011) 218–223

1.15 0.18 Table 2

Parameters from the fitting of the values of fi -to-f1 ratios with the absorbance of
peak P1 from visible spectra obtained with methylene blue solutions of different
1.10 f5 /f1 0.15 concentrations.

fi to f1 ratios fi /f1 = A × A21 + B × A1 + C

1.05

f3 /f1, f4 /f1 , f5/f1 , f6 /f1

f3 /f1 0.12 A B C R2
−2 −3 −1
f2 /f1 5.22 × 10 −8.78 × 10 8.79 × 10 0.998
1.00
f3 /f1 1.06 × 10−2 −1.52 × 10−3 1.10 × 10−1 0.997
f2/f1

f4 /f1 0.09 f4 /f1 2.76 × 10−3 −1.72 × 10−3 7.92 × 10−2 0.997
0.95 f5 /f1 −8.29 × 10−4 −1.55 × 10−2 1.46 × 10−1 0.998
f6 /f1 8.58 × 10−4 −5.34 × 10−4 2.46 × 10−2 0.997
f2 /f1 0.06
0.90
f6 /f1 In conclusion, with this procedure it is possible to build up the
0.85 0.03 whole spectrum for methylene blue with only the values of f1 and
AT , since the rest of the parameters are known (Table 1) or have
0.80 0.00 known relations with f1 and AT (Eq. (1) and Table 2).
A classical calibration procedure would involve relating the
Methylene blue concentration (mg/L).

CMB = 0.407xA12 + 4.319xA1 values of methylene blue concentration to the absorbance of the
12.0
maximum in the original spectra (in our spectra the maximum
R2 = 0.9995 was located at 664 nm). For comparison purposes we deduced the
10.0 following equation for the classical calibration of methylene blue:

8.0 CMB = 0.466A2664 + 3.764A664 (3)

3.2. Deconvolution of visible spectra with the contribution of

6.0
unknown intermediates

4.0 The procedure for evaluating the methylene blue concentration

2.0
0.0
0.0 0.5 1.0 1.5
Peak P1 absorbance (A1 )
Fig. 2. Upper plot: variation in fi -to-f1 ratios with the absorbance of peak P1 from
visible spectra obtained with methylene blue solutions of different concentrations.
Lower plot: variation of methylene blue concentration with the absorbance of peak
P1 at the mean wavelength.
deconvolutions. As can be seen, the peak positions, i , and standard
deviations,  i , were independent of the solution concentration,
whereas the normalized areas of the peaks, fi (fi = 1), were found to
change with the concentration and follow different trends. Thus, f4
and f6 remained constant for all the methylene blue solutions, f1 and
f5 decreased with the increase in the methylene blue concentration
(the peaks being ascribed to monomeric methylene blue [18]) and
f2 and f3 increased with the increase in the methylene blue concen-
tration (the peaks being ascribed to dimeric methylene blue [18]).
Once the values of the peak areas were obtained for the different
solutions it was possible to construct the upper plot in Fig. 2. As can
be seen, fi -to-f1 ratios follow second order polynomial trends with
the absorbance of peak P1 (A1 ) at the mean wavelength (666 nm).
The polynomial and regression coefficients are shown in Table 2.
The lower plot in Fig. 2 displays the calibration data that allow the
methylene blue concentration to be related to the absorbance of
peak P1 (A1 ). This parameter is calculated as follows:
f1
A1 = √ AT
1 2
where AT is the total area of the visible spectrum (nm) before nor-
malization. A polynomial fitting of the data displayed in Fig. 2 yields
the following calibration equation:
CMB = 0.407A21 + 4.319A1
as well as the contribution of unknown intermediates to the visible
spectra obtained during the photodegradation experiment consists
in finding the peaks that, together with those of methylene blue,
best fit the spectra. To this end the following steps were followed:
2.0 2.5
1. A baseline was established between the lowest point at the left
end of the spectrum and the point at 794 nm.
2. The total area (AT ) between the profile of the spectrum and the
baseline was evaluated.
3. The spectrum was normalized so that the total area would be
equal to one (fi = 1).
4. The initial values for the unknown peaks (area fractions fi>6 ,
mean wavelengths, i>6 , and standard deviations,  i>6 ) and
for the methylene blue peaks were inputted. In the case of the
methylene blue peaks the only variable to be fitted was f1 , since
the rest of parameters are known (Table 1) or have known rela-
tions with f1 and AT (Table 2).
5. A minimization algorithm was set to fit f1 and the values of the
parameters for the new peaks (i > 6). Absorbance A1 was then
calculated by means of Eq. (1) and the methylene blue concen-
tration by means of Eq. (2).
In order to implement these steps on an easy-to-use interface we
designed a Microsoft Excel sheet with visual basic macros making
use of the Microsoft Solver complement to minimize the error.
4. Discussion of results
It was found that in most of the spectra obtained during the reac-
tion only three peaks were needed for deconvolution in addition to
those corresponding to methylene blue. Fig. 3 shows the spectrum
(1) obtained after 19 min of reaction together with the peaks obtained
by the deconvolution technique. The peaks ascribed to methylene
blue are represented by the thin lines, whereas the extra peaks (P7,
P8 and P9) are displayed using bold lines. P7, P8, P9 were located at
mean wavelengths of 530.9, 611.2 and 648.4 nm, respectively, and
they maintained these positions throughout the reaction, though
(2) their relative areas varied. This can be appreciated in Fig. 4, where
 G. Marbán et al. / Applied Catalysis A: General 402 (2011) 218–223 221

1.2 2.0
19 min P1 (666 nm; MB)
1.0
1.6 P8 (611 nm)
P9 (648 nm)

Absorbance
0.8
Absorbance

1.2 P7 (531 nm)

0.6
P8 P9 P10 (667 nm)
0.8
0.4
P7
0.4
0.2

0.0
0.0
400 450 500 550 600 650 700 750 800 0 50 100 150 200
Wavelength (nm)
time (min)
Fig. 3. Visible spectrum of a sample obtained after 19 min of reaction together with
Fig. 5. Temporal evolution of the absorbances at the mean wavelengths for P1
the peaks obtained by the deconvolution technique. Bold peaks (P7, P8, P9) are
(methylene blue) and the peaks derived for intermediate compounds (P7 to P10).
ascribed to intermediates; thin peaks are ascribed to methylene blue.

0.60 only the extra peaks and their convolved profiles are represented
for different reaction times. As can be seen, the decreasing area of
0.50 8 min peak P9 during the reaction produces a hypsochromic shift in the
Absorbance

0.40 convolved spectrum of the intermediate products. Fig. 5 shows the

P7 P8 P9
temporal evolution of the absorbances at mean wavelengths for
0.30 the extra peaks (P7, P8 and P9) together with those for peak P1,
corresponding to methylene blue, and for a peak (P10) that started
0.20 to emerge in the longest reaction times at a mean wavelength of
0.10 667 nm, although its significance was always minimal. It can be
seen that the absorbances of P7, P8 and P9 vary in a way that would
0.00 be expected of intermediate products; they increase at the begin-
ning of the reaction and start to decrease when the methylene blue
0.50 19 min concentration has fallen to almost zero.
Absorbance

0.40 Having managed to discriminate between the spectrum cor-

responding to methylene blue and that due to the formation of
0.30 intermediates, we are now in a position to calculate the concen-
tration of methylene blue during the reaction. However we still
0.20
know nothing about the nature of the intermediates. These can be
0.10 expected to be mainly azure A and azure B and, to a lesser extent,
azur C, thionine and phenothiazine [11–13]. Fig. 6 shows the visi-
0.00 ble spectra of aqueous solutions of azure A, B and C taken from the
0.50 38 min literature [20]. The deconvolution of these spectra yields the peaks
that are displayed in the figure. As can be seen the azure A and azure
Absorbance

0.40 B spectra contain three peaks, whose Gaussian parameters (mean

wavelengths and standard deviations) are very similar to those of
0.30 P7, P8 and P9 (see Fig. 4). The ratios of the absorbance of the peak at
0.20 ∼650 nm to the absorbance of the peak at 609–614 nm are 1.15, for
azure B, and 0.34, for azure A. Assuming that peaks P8 (at 611.2 nm)
0.10 and P9 (648.4 nm) are composites of the peaks at 609–614 nm and
at ∼650 nm for azure A and azure B we can use the following equa-
0.00 tion to estimate the proportion of these compounds in the spectra
0.50 54 min shown in Fig. 4:

0.40 1.15 − A9 /A8

%Azure A = 100 − %Azure B = 100 × (4)
Absorbance

1.15 − 0.34
0.30
where A8 and A9 are the values of absorbance at the mean wave-
0.20 length corresponding to peaks 8 and 9, respectively. Fig. 7 shows
the estimated speciation of azure A and B obtained by means of Eq.
0.10 (4). From the figure it can be deduced that the main intermediate
0.00 product is azure A that is formed by the gradual demethylation of
azure B, which in turn is a product of the demethylation of methy-
400 500 600 700 800 lene blue. At reaction times of over ∼100 min both azure A and B
Wavelength (nm) have been degraded to other products that do not absorb radia-
tion in the visible spectral region (Fig. 5). The results displayed in
Fig. 4. Fraction of the visible spectra for samples obtained at different reaction Figs. 5 and 7 are congruent with those obtained by Yogi et al. [10].
times corresponding to intermediate compounds as evaluated by the deconvolution Nevertheless, the main purpose of the deconvolution procedure
technique. is to avoid the artefact arising from the presence of these intermedi-
222 G. Marbán et al. / Applied Catalysis A: General 402 (2011) 218–223

0.3 12

Methylene blue concentration (mg/L)

Azure A Eq. (3) - Maximum
λmax=609.12 nm Eq. (3) - 664 nm
9
Eq. (2) - P1 absorbance
0.2
Absorbance

6
k = 0.034 s-1
k = 0.041 s-1
0.1
λmax =651.2 nm 3 k = 0.062 s-1

λmax =543.2 nm

0
0.0 0 20 40 60 80 100 120 140 160
time (min)
0.6 Azure B
Fig. 8. Evolution of the methylene blue concentration with the reaction time a eval-
λmax =650.5 nm uated by: (i) using Eq. (3) with the values of absorbance at the maximum absorption
0.5
wavelength in the visible spectra, (ii) using Eq. (3) with the values of absorbance at
Absorbance

λmax =614.2 nm 664 nm and (iii) using Eq. (2) with the values of absorbance of peak P1 at its mean
0.4
wavelength (deconvolution technique). Dashed lines represent the curve fittings to
Eq. (5) with the values of k also shown in the figure.
0.3

0.2
of absorbance at the maximum point of the visible spectra, (ii) by
λmax =527.7 nm
using Eq. (3) fed with the values of absorbance at 664 nm and (iii)
0.1
by using Eq. (2) fed with the values of absorbance of peak P1 at the
0.0 mean wavelength. There are significant differences between the
three procedures, which proves that evaluation of the concentra-
tion of methylene blue without taking into account the presence of
Azure C
0.3
intermediates yields higher values than those obtained by means
of Eq. (2) (deconvolution technique). Very often the concentra-
Absorbance

tion values are used to derive kinetic rate constants (k) from first
λmax =590.1 nm
λmax =624.0 nm order reaction equations derived from the Langmuir–Hinshelwood
0.2
kinetic model [21]. The simplest form of this model yields the fol-
lowing equation:

0.1
CMB = C0 e−kt (5)
λmax=504.4 nm
where C0 is the methylene blue concentration at t = 0. The dashed
0.0 lines in Fig. 8 represent the curve fittings to Eq. (5) with the values
400 450 500 550 600 650 700 750 800
of k also shown in the figure. It can be seen that by ignoring the arte-
Wavelength (nm) fact when evaluating the methylene blue concentration the kinetic
rate constants turn out to be significantly lower than that obtained
Fig. 6. Visible spectra of azure A, azure B and azure C (10 ␮M aqueous solutions);
original data taken from [20]. Curve deconvolutions performed in this study. by employing the methylene blue concentration as calculated by
means of our deconvolution technique. In the view of the different
results shown in Fig. 8, future studies devoted to methylene blue
ate compounds in the evaluation of the concentration of methylene degradation should clearly state whether the artefact masking the
blue. Fig. 8 shows the evolution of the concentration of methylene evaluation of methylene blue concentration has been taken into
blue with the reaction time as evaluated by three different meth- account or not.
ods: (i) by using the classical calibration Eq. (3) fed with the values

5. Conclusions

100 %Azure A
The presence of intermediates during the photocatalytic
%Azure B degradation of methylene blue in aqueous solution masks the
Intermediates (%)

80
determination of its concentration when classical evaluation meth-
ods by means of visible spectroscopy analyses are employed. The
60
deconvolution procedure described in this work circumvents this
40
artefact by estimating the contribution of intermediates to the vis-
ible spectrum. In the photodegradation of methylene blue with
20 Degussa P25 TiO2 particles the main intermediate detected in
the visible spectral region was azure A resulting from the grad-
0 ual demethylation of azure B. The methylene blue concentrations
8 19 38 54 obtained if the artefact was not taken into account were higher
time (min) than those evaluated by the deconvolution technique and provided
lower values of kinetic rate constants for the photocatalytic degra-
Fig. 7. Speciation of azure A and azure B during the reaction obtained by Eq. (4). dation.
 G. Marbán et al. / Applied Catalysis A: General 402 (2011) 218–223
Acknowledgement
Tan T. Vu is grateful to CSIC for the award of a JAE predoc grant.
References
[1] U.I. Gaya, A.H. Abdullah, J. Photochem. Photobiol. C: Photochem. Rev. 9 (2008)
1–12.
[2] C.M. Teh, A.R. Mohamed, J. Alloy Compd. 509 (2011) 1648–1660.
[3] F. Han, V.S.R. Kambala, M. Srinivasan, D. Rajarathnam, R. Naidu, Appl. Catal. A:
Gen. 359 (2009) 25–40.
[4] K. Rajeshwar, M.E. Osugi, W. Chanmanee, C.R. Chenthamarakshan, M.V.B.
Zanoni, P. Kajitvichyanukul, R. Krishnan-Ayer, J. Photochem. Photobiol. C: Pho-
tochem. Rev. 9 (2008) 171–192.
[5] L. Xiong, W. Sun, Y. Yang, C. Chen, J. Ni, J. Colloid Interface Sci. 356 (2011)
211–216.
[6] A.B. Gambhire, M.K. Lande, B.R. Arbad, S.B. Rathod, R.S. Gholap, K.R. Patil, Mater.
Chem. Phys. 125 (2011) 807–812.
[7] C.H. Kwon, H. Shin, J.H. Kim, W.S. Choi, K.H. Yoon, Mater. Chem. Phys. 86 (2004)
78–82.
[8] J. Matos, A. García, S.E. Park, Appl. Catal. A: Gen. 393 (2011) 359–366.
[9] M.A. Rauf, M.A. Meetani, A. Khaleel, A. Ahmed, Chem. Eng. J. 157 (2010)
373–378.
223
[10] C. Yogi, K. Kojima, N. Wada, H. Tokumoto, T. Takai, T. Mizoguchi, H. Tamiaki,
Thin Solid Films 516 (2008) 5881–5884.
[11] Y. Tominaga, T. Kubo, K. Hosoya, Catal. Commun. 12 (2011) 785–789.
[12] L. Yu, S. Yuan, L. Shi, Y. Zhao, J. Fang, Micropor. Mesopor. Mater. 134 (2010)
108–114.
[13] T. Zhang, T. Oyama, A. Aoshima, H. Hidaka, J. Zhao, N. Serpone, J. Photochem.
Photobiol. A: Chem. 140 (2001) 163–172.
[14] L. Mi, P. Xu, H. Shen, P.N. Wang, J. Photochem. Photobiol. A: Chem. 193 (2008)
222–227.
[15] H.W.P. Carvalho, A.P.L. Batista, P. Hammer, T.C. Ramalho, J. Hazard Mater. 184
(2010) 273–280.
[16] O. Mekasuwandumrong, P. Pawinrat, P. Praserthdam, J. Panpranot, Chem. Eng.
J. 164 (2010) 77–84.
[17] S. Naghiloo, A. Habibi-Yangjeh, M. Behboudnia, Appl. Surf. Sci. 257 (2011)
2361–2366.
[18] K. Patil, R. Pawar, P. Talap, Phys. Chem. Chem. Phys. 2 (2000) 4313–
4317.
[19] P.A.R. Tafulo, R.B. Queirós, G. González-Aguilar, Spectrochim. Acta A: Mol.
Biomol. Spectrosc. 73 (2009) 295–300.
[20] N. Narband. Nanoparticles and Photosensitisers; Their Interactions and
Antibacterial Properties, PhD Thesis, University College London, 2009, Ref Type:
Thesis/Dissertation.
[21] Y. Zhang, J.C. Crittenden, D.W. Hand, D.L. Perram, Environ. Sci. Technol. 28
(1994) 435–442.