Ka

CD

a

ARRAA

KACKMM

1

tCisma

[bnm[dspta

s

0d

Journal of Hazardous Materials 190 (2011) 986–992

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journa l homepage: www.e lsev ier .com/ locate / jhazmat

inetic analysis of C.I. Acid Yellow 9 photooxidative decolorization by UV–visiblend chemometrics

ristina Fernández, M. Pilar Callao, M. Soledad Larrechi ∗

epartment of Analytical and Organic Chemistry, Rovira i Virgili University, C/Marcel·lí Domingo, s/n, Campus Sescelades, E-43007 Tarragona, Spain

r t i c l e i n f o

rticle history:eceived 1 December 2010eceived in revised form 9 March 2011ccepted 9 April 2011vailable online 15 April 2011

eywords:zo dye degradation

a b s t r a c t

A kinetic study of the C.I. Acid Yellow 9 photooxidative decolorization process, using H2O2 as oxidant, wascarried out by chemometric analysis of the UV–visible data recorded during the process. The number ofchemical species involved in the photooxidative decolorization process was established by singular valuedecomposition (SVD) and evolving factor analysis (EFA). Information about the different chemical speciesalong the process was obtained from the spectral and concentration profiles recovered by soft multivari-ate curve resolution with alternating least squares (MCR-ALS). This information was complemented bymass spectrometry (MS) to postulate a reaction pathway. The dye photooxidative decolorization pro-

.I. Acid Yellow 9inetic studyingass spectroscopyCR-ALS

cess involved consecutive and parallel reactions. The consecutive pathway consists of a first stage of dyeoxidation followed by the rupture of the azo linkage to form smaller molecules that are degraded in asubsequent stage. The parallel reactions form products that are undetectable in the UV–visible spectra.Kinetic constants of the reactions postulated in the photooxidative process were retrieved by applyinga hybrid hard and soft MCR-ALS resolution. All constants were similar for the consecutive stages andhigher than those obtained for the parallel reactions.

. Introduction

Organic azo dyes are involved in many industrial processes, sohey are frequently found in industrial wastewater [1]. Concretely,.I. Acid Yellow 9, is used in the textile industry [2–4]. Moreover,

t was recently reported that this dye has properties that make ituitable as a dispersing agent for carbon nanotubes [5,6] whicheans that their industrial use could increase in the future. Organic

zo dyes are contaminants and must be eliminated.Degradation processes of C.I. Acid Yellow 9 using Mn oxide

2], biodegradation [3] and electrochemical oxidation [4,7] haveeen reported. Nowadays, however, homogeneous and heteroge-eous dye removal techniques that allow fast degradations are theethods most commonly used for the degradation of organic dyes

1,8–14]. Regardless of which technique is used, the challenge ofetermining the optimal conditions for rapid dye removal is usuallyolved by determining the kinetic rate and using it as a referencearameter. This study focuses on the kinetic analysis of the pho-ooxidative decolorization process of C.I. Acid Yellow 9 using H2O2

s oxidant.

Usually, dye removal reactions are monitored by UV–visiblepectroscopy [1,8,10,15] and the degradation rate constant is estab-

∗ Corresponding author. Tel.: +34 977559559; fax: +34 977558446.E-mail address: mariasoledad.larrechi@urv.cat (M.S. Larrechi).

304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.jhazmat.2011.04.037

© 2011 Elsevier B.V. All rights reserved.

lished fitting the absorbance values obtained at one wavelength toa pseudo first-order reaction. During the degradation processes,however, as a consequence of the radical reaction, intermediates-some sensitive to UV–visible spectra and others not-can be formed[3,4,7,16–18]. This univariate approach can lead to erroneousresults if intermediates absorbing in the same wavelength rangethat the studied dye are formed during the process.

The use of selective techniques such separation techniques cou-pled with mass spectrometry (MS) [19–22] are useful to avoid theproblem of the intermediates interference. With techniques of thissort, however, the samples usually need to be submitted to somesort of pre-treatment and the analysis time may be long. Anotherpossibility to overcome this problem is using UV–visible spec-troscopy coupled to multivariate curve resolution methods. Thisapproach was used in the present study.

In this study, the number of variability sources related to thenumber of chemical species present in the reaction was anal-ysed using singular value decomposition (SVD) and evolving factoranalysis (EFA) of the UV–visible spectra recorded during dye pho-tooxidative decolorization. Subsequently, a soft multivariate curveresolution with alternating least squares (MCR-ALS) was carried outto obtain information about the spectral profiles of each compound.

This information was complemented by MS analysis of some repre-sentative samples taken during the degradation process to proposea reaction pathway. In a second step a hybrid hard and soft mod-elling technique [23] was used to evaluate the kinetic constants of

zardou

dkkr

2

2

csfDf

2

2

o(pft0

2

tTbf

2

m(s17

2

LNa

2

istotdwo

Hr

C. Fernández et al. / Journal of Ha

ifferent reactions involved in the photooxidative process. To ournowledge, no strategies of this type have previously been used ininetic studies of dye photooxidative decolorization, so this studyepresents an advance in the field.

. Experimental

.1. Chemicals

The reagents used in this study were analytical-gradehemicals. C.I. Acid Yellow 9 (2-amino-5-([4-ulfophenyl]azo)benzenesulfonic acid) (95%) was purchasedrom Sigma–Aldrich and H2O2 (30%) was purchased from Scharlau.egradation solutions were prepared in ultrapure Milli-Q water

rom a system supplied by Millipore (USA).

.2. Instruments, measuring conditions and software

.2.1. ReactorThe cylindrical annular batch reactor consisted of a Pyrex glass

uter reactor with 0.7 L of capacity and a quartz immersion tube2.5 cm i.d. and 38 cm long). The tube contained a 15 W low-ressure mercury-vapour lamp (LPML) (TNN 15/32 56001721),rom Heraeus Noblelight, Germany. This UVC light source emit-ed at 254 nm and the incident photon flux of the UV reactor was.1 W cm−1.

.2.2. UV-visible spectrophotometerSpectral data were acquired using a diode-array spectropho-

ometer (Shimadzu MultiSpec-1501) and Hyper-UV 1.51 software.he UV–visible spectra of the dye-degradation samples and thelank, in which only H2O2 is present, were recorded at each nmrom 270 to 720 nm.

.2.3. Mass spectrometerThe MS spectra were acquired in negative centroid acquisition

ode by direct injection using a 6210 LC/TOF mass spectrometerAgilent Technologies, Palo Alto, USA), equipped with an electro-pray ionization source. MS spectra of six samples obtained at 0, 8,8, 25, 30 and 60 min of degradation process were recorded in the0–400 m/z range.

.2.4. SoftwareThe UV–visible spectra were exported and converted in MAT-

AB binary files. The data matrix was analysed by MCR-ALS [24].onlinear fitting was performed using home-made software withMATLAB 6.5 computer environment [25].

.3. Photooxidative decolorization procedure

A 0.5 L solution of 60 mg L−1 of C.I. Acid Yellow 9 and H2O2 0.05 Mn Milli-Q water was introduced into the cylindrical reactor. Thisolution was irradiated at 254 nm during the process. The concen-rations were selected from preliminary studies [8,26,27]. Samplesf 3 mL of the degradation solution were taken at the beginning ofhe experiment (t = 0 min) and after each minute throughout theegradation time (t = 60 min). Two repetitions of the experimentere carried out in order to calculate the standard deviation (SD)

f the results.In order to determine the noise in this process, degradation of

2O2 was carried out as the degradation blank under the sameeaction conditions.

s Materials 190 (2011) 986–992 987

2.4. Methodology and chemometric analysis

The methodology followed to obtain the spectral and concen-tration profiles of the chemical species and to calculate the rateconstants of the reactions involved in the photooxidative processwas: (a) evaluation of the number of chemical species along thephotooxidative decolorization process by SVD and EFA, (b) anal-ysis of the spectral and concentration profiles retrieved by softMCR-ALS and analysis of MS spectra to propose a reaction path-way, (c) calculation of the rate constants by hybrid hard and softMCR-ALS.

The procedure begins with the arrangement of the UV–visiblespectra recorded during the blank and the photooxidation of C.I.Acid Yellow 9 in two data matrices (Db (m × n)) for the blank and (D1(m × n)) for the dye degradation, where m is the number of spectrarecorded (m = 61) and n is the number of wavelengths (n = 451).

In the two data matrices, Db and D1, the number of variabilitysources related to species active in the studied UV–visible rangewas analysed by SVD [28]. Taking into account that in Db only onechemical species is present (H2O2), the value of the second singularvalue from the analysis of Db, related to the variability representa-tive of the noise in the experimental process, was considered thethreshold for determining the number of significant factors in D1.

Information about the evolution of the different factors dur-ing the photooxidative decolorization process was obtained by EFA[29] of the data matrix D1. In this method, based on factor analysis,windows of linearly increasing size were subjected to SVD. In theforward analysis, we started by evaluating the SVD of the first twospectra of matrix D1 and then repeated this process by adding onespectrum at a time until we reached the total matrix. In the back-ward analysis, the process was the same but started with the lasttwo spectra of matrix D1.

After the determination of the number of chemical species asoft-modelling MCR-ALS [30] was applied to obtain informationabout the chemical species involved in the process. With this algo-rithm, the experimental matrix D1 was decomposed into two newmatrices according the following equation:

D1 = C1ST + E1

where C1 and ST1 contain, respectively, information about the

concentration and spectral profiles of the species involved in thephotooxidative decolorization process and E1 is a matrix of theresiduals that contains the variance unexplained by C1ST

1. MCR-ALS solved this equation iteratively in order to minimize theresidual matrix E1. This optimization requires an initial estimationof the concentration or spectral profiles, and some constraints canbe applied. In this step, the initial estimation was the EFA profilesand the constraints employed were non-negativity for the concen-tration and spectral profiles and unimodality for the concentrationprofiles.

The information obtained was complemented with the MS spec-tra of six representative samples taken during the degradationprocess and was employed to postulate a reaction pathway.

In a subsequent stage, rate constants of the reactions involved inthe C.I. Acid Yellow 9 photooxidative decoloration were obtainedby hybrid hard soft MCR-ALS. This procedure includes a hard mod-elling constraint that corresponds to the kinetic model postulated.In accordance with this postulated model, a function consisting ofthe differential equations of each compound present in the reactionwas constructed. This function was integrated numerically using anordinary differential equation solver (ODE23, a Runge-Kutta MAT-

LAB subroutine).

In this step, the concentration profiles retrieved by soft MCR-ALS and an estimation of the rate constants obtained through theseprofiles were used as initial estimation. Finally the spectral and

988 C. Fernández et al. / Journal of Hazardous Materials 190 (2011) 986–992

1.6

1.8 (a) (b)1.6

1.8

0.8

1

1.2

1.4

Abs

0.8

1

1.2

1.4

Abs

Degradation time

0

0.2

0.4

0.6

0

0.2

0.4

0.6

300 350 400 450 500 550 600 650 700Wavelength (nm)

300 350 400 450 500 550 600 650 700Wavelength (nm)

F I. Acids

kw

3

dsdp

ditmrc

Tn

tIiinyos1catn

da

TR

similarity coefficient (0.9990) between the real C.I. Acid Yellow 9spectra and the spectra retrieved by MCR-ALS.

Fig. 3a shows the spectral profiles retrieved by soft MCR-ALS, ofthe initial dye (1) and the chemicals that appear in the first minutes

ig. 1. UV–visible spectra recorded during the photooxidative decolorization of C.pectra are plotted.

inetic profiles and the rate constants that provided the best fitere retrieved.

. Results and discussion

Fig. 1a and b shows the UV–visible spectra obtained during theegradation of C.I. Acid Yellow 9 and the blank, respectively. Fig. 1ahows a decrease in the absorbance values as a consequence of theye degradation, but it is not clear whether any intermediate isresent in the solution.

Table 1 shows the singular values obtained by SVD for both theye and the blank matrices. Because only one component is present

n the degradation blank, the second singular value obtained fromhe Db matrix is considered representative of the noise, which

eans that three eigenvalues are significant in the D1 matrix. Theseesults lead us to think that intermediates are present in this pro-ess.

Fig. 2 shows the EFA plot of the C.I. Acid Yellow 9 degradation.he horizontal line was retrieved by EFA and is associated withoise.

In the forward analysis, from the very first minutes of degrada-ion the variability in the spectral data is explained by two factors.t must be noted that the number of significant factors that containnformation on the spectral evolution recorded during degradations equal to the number of independent reactions +1 or equal to theumber of chemical species that contribute to the spectra anal-sed [31]. The presence of two factors from the very beginningf the reaction must be related to two chemical species corre-ponding to a reaction in which the initial dye is degraded. After5–20 min of reaction time, a new factor becomes relevant, indi-ating that another variability source has appeared. This could bessociated with the presence of another chemical species active inhe UV–visible region being studied. After 30 min of degradation,

o additional information is obtained.

Similar information is obtained in the backward analysis of theata matrix D1. In this direction, from 60 min since 40–45 min therere no factors of significant value because no changes are present

able 1ank analysis of D1 and Db matrices. Only the first five singular values are shown.

Factors Blank C.I. Acid Yellow 9

1 8,549 97,2622 2,205 13,5323 0,312 4,6744 0,068 1,8445 0,017 0,644

Yellow 9 (a) and the blank (b). For the decolorization of the dye, only the last 45

in the reaction, probably because the degradation has finished. At40–45 min, one factor which can be associated with a chemicalspecies appears. From 28 to 33 min, the variability is explained bytwo factors and the third significant factor is observed after 10 minof degradation.

At this point in the discussion, at least three chemical speciesactive in the UV–visible region can be thought to be involved inthe process. Information about these species can be obtained byMCR-ALS analysis of matrix D1. This process starts by using the con-centration profiles calculated by EFA as an initial estimation. Theretrieved spectral profiles were associated with the new chemicalspecies and these profiles jointly with the UV spectra of C.I. AcidYellow 9 were considered as the initial estimation during the opti-mization step when a subsequent MCR-ALS was applied to D1. Fig. 3contains the optimal spectral (Fig. 3a) and concentration (Fig. 3b)profiles, which show the behaviour of each species. The goodnessof the resolution can be evaluated by considering the fitting errorof the resolution (0.49%), the variance explained (99.92%) and the

Fig. 2. EFA of the spectral data matrix D1 . Forward (continuous lines) and backward(dotted lines) directions.

C. Fernández et al. / Journal of Hazardous Materials 190 (2011) 986–992 989

30

35

0.2

0.25 (b)(a)

15

20

25

Con

cent

ratio

n (a

.u.)

0.1

0.15

Abs

(u.a

.)

(2) Oxidized species

(1) C.I. Acid Yellow 9

(1) C.I. Acid Yellow 9(3) Aromatic compounds

10 20 30 40 50 600

5

10

00

0.05(3) Aromatic compounds

(2) Oxidized species

invol

ohtactthbad

tTpddtbismp

tTv

taotob[Ici

ttwtdntmnta[

300 350 400 450 500 550 600 650 70Wvelength (nm)

Fig. 3. Spectral (a) and concentration (b) profiles of the chemical species

f the reaction (2). The chemicals (2) have maximum absorbance atigher wavelengths than the initial dye. This profile suggests thathe azo dye degrades to form a first intermediate that contains thezo group because absorption at these wavelengths is typical ofompounds that contain an azo group. The shape of the third spec-rum (3) is related to a second intermediate that appear duringhe degradation corresponding to the smaller molecules that haveigh absorbance in the UV region, meaning that these are proba-ly aromatic molecules that are products of the breakdown of thezo linkage, at the end of the degradation these species are alsoegraded.

According to Fig. 3b, the rate of dye and intermediates degrada-ion appears to be higher than the rate of intermediates formation.his leads us to think that undetectable UV–visible compounds areerhaps being formed during the degradation stages following aifferent pathway. These concentration profiles show that degra-ation could be considered finished at around 30 min and that athis time the compounds corresponding to the third spectrum haveeen degraded to compounds that are undetectable at the stud-

ed wavelength, such CO2, H2O and small acids that are the mainubproducts in reactions of this kind. These results are in agree-ent with that observed in the EFA profiles where no variability is

resent after 30 min of degradation.In order to propose the general degradation pathway, MS spec-

ra were recorded at 0, 8, 18, 25, 30 and 60 min of degradation.hese samples were selected according to the appearance of thearious factors in the EFA analysis.

Fig. 4 shows the most informative parts of the MS spectra, athe beginning of the reaction (t = 0), at an intermediate time (t = 18)nd at the final time (t = 60). These three times are representativef the different stages of the reaction. The MS spectrum of the ini-ial untreated C.I. Acid Yellow 9 (C12H9N3O6S2Na2) (Mr = 401) wasbtained both before and after the addition of H2O2 (time = 0). Inoth cases, two majority peaks were observed, corresponding toC12H9N3O6S2]−2 at m/z 177.50 and [C12H10N3O3S]− at m/z 276.04.n the sample containing H2O2, which corresponds to the initialonditions of degradation, a third peak was observed, correspond-ng to [C12H10N3O6S2]− at m/z 356.00. Fig. 4a shows these peaks.

At 8 min of degradation (not shown), two new peaks appeared,he larger one corresponding to [C12H9N3O7S2]−2 at m/z 185.49 andhe other corresponding to [C6H5O5S]− at m/z 188.99. These peaksere observed at 18 min (Fig. 4b). At subsequent degradation times,

hese two peaks grew and the peaks corresponding to the initial dyeecreased in size. At 25 min of degradation (not shown), these twoew peaks started to decrease in size and three new peaks startedo appear, corresponding to [C2H3SO5]− at m/z 138.97 [HSO5]− at/z 112.96, and [HSO4]− at m/z 96.96. At 30 min (not shown), the

ew peaks started to decrease, except for the one correspondingo [HSO4]−, which increased in size. These three peaks appearedfter 60 min of degradation (Fig. 4c), with the one corresponding toHSO4]− being the largest of the three.

0Time (min)

ved in the photooxidative decolorization process retrieved by MCR-ALS.

Taking these results into consideration, we propose the path-way shown in Fig. 5. In this scheme, C.I. Acid Yellow 9 (a) in afirst step is oxidized to (b) and then the azo linkage is broken toform (c) and (d). Subsequently, other reactions take place. At theend of the degradation time (60 min), C2H4SO3 (e), H2SO5 (f) andH2SO4 (g) can be observed and the other compounds are nearlydegraded. As it is indicated in this figure, in any step of the mainreaction undetectable compounds, in the UV–vis range studied, canbe formed.

These results are in agreement with those obtained by MCR-ALS(Fig. 3), where the dye (a) corresponds to compound (1), the oxi-dized species (b) corresponds to (2), and the species (c) and (d),which are aromatic compounds that absorb in the UV region, areconsidered to be included in the shape of spectra (3). The other com-pounds observed in MS (g, e and f) are not active in the UV–visibleregion studied. This is the reason why, in Fig. 3, no compoundsare observed at the end of degradation. This degradation pathwayagrees with the one obtained for the degradation of C.I. Acid Yel-low 9 using biodegradation and electrochemical degradation [3,4].Other intermediates, of low molecular weight and undetectable inthe studied wavelength range, could appear in each stage of degra-dation.

From a practical point of view, it is important not only to detectthe formation of intermediates but also to evaluate their rates ofdegradation. This information can be obtained if, during the opti-mization of the MCR-ALS step, a kinetic model is imposed as arestriction.

In accordance with the results discussed above, we propose thefollowing kinetic model:

Ak1−→B (1)

Ak2−→D (2)

Bk3−→C (3)

Bk4−→D (4)

Ck5−→D (5)

where A corresponds to the initial dye (a) in Fig. 5, B to the oxi-dized intermediate (b) in Fig. 5, C to a mixture of the (c) and (d)species of Fig. 5, and D to the small compounds undetectable in theUV–visible range, such (e), (f) and (g) in Fig. 5 or smaller compoundsundetectable by MS analysis in the selected m/z range.

In the selected model, the reactions (1), (3) and (5) are proposedon the basis of the MS spectra results and the pathway shown in

Fig. 5. We also included reactions (2) and (4) in which A and B formsmall undetectable products because this behaviour is common inradicalary reactions and, moreover, in the shape of the degradationprofiles obtained by MCR-ALS, the rate of the degradation of the

990 C. Fernández et al. / Journal of Hazardous Materials 190 (2011) 986–992

at diff

di

ttte

l

TM

Fig. 4. MS spectra of the photooxidative decolorization mixture

yes and intermediates is higher than the rate of the formation ofntermediates.

The initial estimations of the constants were calculated usinghe soft MCR-ALS concentration profiles, all of them were adjustedo first-order kinetics which is the common approximation in pho-ooxidation processes [8,10,15]. With this purpose the followingquation has been used:

n[dye]0

[dye]= kapt

he equations of the reactions used in the hybrid hard and softCR-ALS model were as follows:

dA

dt= −k1A − k2A

dB

dt= k1A − k3B − k4B

erent reaction times (a) at 0 min, (b) at 18 min and (c) at 60 min.

dC

dt= k3B − k5C

It has to be remarked that this is a first estimate of thereaction pathway because the mechanism of this reaction isunknown. Therefore, the evaluated rate constants must be con-sidered apparent values, informative of the main reaction stagesthat can be identified spectrophotometrically. The fitting errorof the resolution was 11.90% and the explained variance was98.50%. The dye spectra retrieved by MCR-ALS was comparedwith the real dye spectra and a correlation of 0.9983 wasobtained.

The values of the kinetic constants were: k1 = 0.092 (SD = 0.001),k2 = 0.050 (SD = 0.002), k3 = 0.099 (SD = 0.003), k4 = 0.044(SD = 0.005) and k5 = 0.122 (SD = 0.006). If the overall process

of degradation of A (reactions (1) and (2)), B (reactions (3) and(4)) and C (reaction (5)) is considered, these compounds are foundto degrade at the same rate. The values of k2 and k4 are lowerthan the values obtained for k1, k3 and k5, which means that the

C. Fernández et al. / Journal of Hazardou

NN

SO3Na

NaO3S (a)UC*

k2

H2N

NN

SO4H

HO3S (b)UC*

k1

k4

H2N

SO4HOHHO3S

k3

HOHO(d)(c)

k5

H2CCHSO5H

H2SO4H2SO5

(g)(f)(e)

compounds

UC*

UC* undetectable

Ft

rA

4

oTpmpi

iop

soo

[

[

[

[

[

[

[

[

[

[

[

[

[

ig. 5. Proposed reaction pathway for C.I. Acid Yellow 9 photooxidative decoloriza-ion.

eactions proposed in Fig. 5 make up the principal pathway of C.I.cid Yellow 9 photooxidative decolorization.

. Conclusions

Multivariate tools such EFA and soft MCR-ALS make it possible tobtain information about when intermediates appear in a reaction.his enables s to select the most appropriate samples during thehotooxidation for MS analysis, identify intermediates and deter-ine reaction pathways. Under our experimental conditions, the

hotooxidative decolorization of C.I. Acid Yellow 9 was completedn approximately 30 min.

After an initial approach based on soft MCR-ALS results, thentroduction of a hard restriction in MCR-ALS makes it possible tobtain information about all the kinetic constants involved in thehotooxidative decolorization process.

C.I. Acid Yellow 9 photooxidative decolorization involves con-ecutive and parallel reactions. The consecutive pathway—the mainne—consists of a first stage of dye oxidation followed by a rupturef the azo linkage to form smaller molecules, which are degraded in

[

[[

s Materials 190 (2011) 986–992 991

a subsequent stage. All of the identified compounds are degradedat similar rates.

Acknowledgements

The authors would like to thank the Spanish Ministry of Sci-ence and Innovation (Project CTQ2007-61474/BQU) for economicsupport and for providing Cristina Fernández with a doctoral fel-lowship (AP2007-03788).

References

[1] C. Fernández, M.S. Larrechi, M.P. Callao, An analytical overview of processesfor removing organic dyes from wastewater effluents, Trends Anal. Chem. 29(2010) 1202–1211.

[2] C.E. Clarke, F. Kielar, H.M. Talbot, K.L. Johnson, Oxidative decolorization of acidazo dyes by a Mn oxide containing waste, Environ. Sci. Technol. 44 (2010)1116–1122.

[3] B.V. Pandey, R.S. Upadhyay, Spectroscopic characterization and identification ofPseudomonas fluorescens mediated metabolic products of Acid Yellow-9, Micro-biol. Res. 161 (2006) 311–315.

[4] D. Vanerkova, A. Sakalis, M. Holcapek, P. Jandera, A. Voulgaropoulos, Analy-sis of electrochemical degradation products of sulphonated azo dyes usinghigh-performance liquid chromatography/tandem mass spectrometry, RapidCommun. Mass Spectrom. 20 (2006) 2807–2815.

[5] S.A. Kumar, C. Tang, S. Chen, Poly(4-amino-1-1′-azobenzene-3,4′-disulfonicacid) coated electrode for selective detection of dopamine from its interfer-ences, Talanta 74 (2008) 860–866.

[6] S.A. Kumar, S. Wang, T. Yang, C. Yeh, Acid Yellow 9 as a dispersing agent forcarbon nanotubes: preparation of redox polymer–carbon nanotube compositefilm and its sensing application towards ascorbic acid and dopamine, Biosens.Bioelectron. 25 (2010) 2592–2597.

[7] A. Sakalis, D. Vanerková, M. Holcapek, P. Jandera, A. Voulgaropoulos,Electrochemical treatment of a simple azodye and analysis of the degrada-tion products using high performance liquid chromatography–diode arraydetection–tandem mass spectrometry, Chemosphere 67 (2007) 1940–1948.

[8] C. Fernández, M.S. Larrechi, M.P. Callao, Modelling of the simultaneous pho-todegradation of Acid Red 97, Acid Orange 61 and Acid Brown 425 usingfactor screening and response surface strategies, J. Hazard. Mater. 180 (2010)474–480.

[9] F.H. Abdullah, M.A. Rauf, S.S. Ashraf, Photolytic oxidation of Safranin-O withH2O2, Dyes Pigments 72 (2007) 349–352.

10] M.A. Behnajady, N. Modirshahla, Kinetic modelling on photooxidative degra-dation of C.I. Acid Orange 7 in a tubular continuous-flow photoreactor,Chemosphere 62 (2006) 1543–1548.

11] B.K. Körbahti, M.A. Rauf, Application of response surface analysis to the pho-tolytic degradation of Basic Red 2 dye, Chem. Eng. J. 138 (2008) 166–171.

12] M.A. Rauf, N. Marzouki, B.K. Körbahti, Photolytic decolorization of Rose Bengalby UV/H2O2 and data optimization using response surface method, J. Hazard.Mater. 159 (2008) 602–609.

13] M.G. Dias, E.B. Azevedo, Photocatalytic decolorization of commercial acid dyesusing solar irradiation, Water Air Soil Pollut. 204 (2009) 79–87.

14] A.J. Julson, D.F. Ollis, Kinetics of dye decolorization in an air–solid system, Appl.Catal. B 65 (2006) 315–325.

15] N. Daneshvar, A. Aleboyeh, A.R. Khataee, The evaluation of electrical energyper order (Eeo) for photooxidative decolorization of four textile dye solutionsby the kinetic model, Chemosphere 59 (2005) 761–767.

16] S. Vajnhandl, A.M. Le Marechal, Case study of the sonochemical decolourationof textile azo dye Reactive Black 5, J. Hazard. Mater. 141 (2007) 329–335.

17] S. Song, H. Ying, Z. He, J. Chen, Mechanism of decolorization and degradationof CI Direct Red 23 by ozonation combined with sonolysis, Chemosphere 66(2007) 1782–1788.

18] L. Lei, Q. Dai, M. Zhou, X. Zhang, Decolorization of cationic red X-GRL by wetair oxidation: performance optimization and degradation mechanism, Chemo-sphere 68 (2007) 1135–1142.

19] M. Constapel, M. Schellenträger, J.M. Marzinkowski, S. Gäb, Degradation of reac-tive dyes in wastewater from the textile industry by ozone: analysis of theproducts by accurate masses, Water Res. 43 (2009) 733–743.

20] F. Zhang, A. Yediler, X. Liang, Decomposition pathways and reaction interme-diate formation of the purified, hydrolyzed azo reactive dye C.I. Reactive Red120 during ozonation, Chemosphere 67 (2007) 712–717.

21] M. Zhou, J. He, Degradation of azo dye by three clean advanced oxidationprocesses: wet oxidation, electrochemical oxidation and wet electrochemicaloxidation – A comparative study, Electrochim. Acta 53 (2007) 1902–1910.

22] W. Feng, D. Nansheng, H. Helin, Degradation mechanism of azo dye C.I. Reac-tive Red 2 by iron powder reduction and photooxidation in aqueous solutions,Chemosphere 41 (2000) 1233–1238.

23] A. de Juan, M. Maeder, M. Martínez, R. Tauler, Combining hard- and soft-modelling to solve kinetic problems, Chemom. Intell. Lab. Syst. 54 (2000)123–141.

24] A. de Juan, R. Tauler, J. Jaumot. http://www.mcrals.info/ (accessed 24.02.11).25] Matlab 6.5, The Mathworks, South Natick, MA, USA.

9 zardou

[

[

[

[

[30] R. Tauler, Multivariate curve resolution applied to second order data, Chemom.Intell. Lab. Syst. 30 (1995) 133–146.

92 C. Fernández et al. / Journal of Ha

26] C. Fernández, M.S. Larrechi, M.P. Callao, Study of the influential factors in thesimultaneous photocatalytic degradation process of three textile dyes, Talanta79 (2009) 1292–1297.

27] V. Gómez, J. Font, M.P. Callao, Sequential injection analysis with second-order

treatment for the determination of dyes in the exhaustion process of tanningeffluents, Talanta 71 (2007) 1393–1398.

28] D.L. Massart, B. Vandeginste, L. Buydens, S. De Jong, P. Lewi, J. Smeyers-Verbeke,Handbook of Chemometrics and Qualimetrics. Part A: Data Handling in Scienceand Technology, Elsevier, Amsterdam, 1997.

[

s Materials 190 (2011) 986–992

29] H.R. Keller, D.L. Massart, Evolving factor analysis, Chemom. Intell. Lab. Syst. 12(1992) 209–224.

31] M. Amrhein, B. Srinivasan, D. Bonvin, M.M. Schumacher, On the rank deficiencyand rank augmentation of the spectral measurement matrix, Chemom. Intell.Lab. Syst. 33 (1996) 17–33.