ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auE-mail address:

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

Mineralization of salicylic acid in acidic aqueous mediumby electrochemical advanced oxidation processes usingplatinum and boron-doped diamond as anode andcathodically generated hydrogen peroxide

Elena Guinea, Conchita Arias, Pere Lluıs Cabot, Jose Antonio Garrido,Rosa Marıa Rodrıguez, Francesc Centellas, Enric Brillas�

Laboratori d’Electroquımica dels Materials i del Medi Ambient, Departament de Quımica Fısica, Facultat de Quımica,

Universitat de Barcelona, Martı i Franques 1-11, 08028 Barcelona, Spain

a r t i c l e i n f o

Article history:

Received 31 May 2007

Received in revised form

19 July 2007

Accepted 24 July 2007

Available online 1 August 2007

Keywords:

Salicylic acid

Anodic oxidation

Electro-Fenton

Photoelectro-Fenton

Solar photoelectro-Fenton

Oxidation products

nt matter & 2007 Elsevie.2007.07.046

thor. Tel.: +34 93 4021223;brillas@ub.edu (E. Brillas)

a b s t r a c t

Solutions containing 164 mg L�1 salicylic acid of pH 3.0 have been degraded by

electrochemical advanced oxidation processes such as anodic oxidation, anodic oxidation

with electrogenerated H2O2, electro-Fenton, photoelectro-Fenton and solar photoelectro-

Fenton at constant current density. Their oxidation power has been comparatively studied

in a one-compartment cell with a Pt or boron-doped diamond (BDD) anode and a graphite

or O2-diffusion cathode. In the three latter procedures, 0.5 mM Fe2+ is added to the solution

to form hydroxyl radical (dOH) from Fenton’s reaction between Fe2+ and H2O2 generated at

the O2-diffusion cathode. Total mineralization is attained for all methods with BDD and for

photoelectro-Fenton and solar photoelectro-Fenton with Pt. The poor decontamination

achieved in anodic oxidation and electro-Fenton with Pt is explained by the slow removal

of most pollutants by dOH formed from water oxidation at the Pt anode in comparison to

their quick destruction with dOH produced at BDD. dOH generated from Fenton’s reaction

oxidizes rapidly all aromatic pollutants, but it cannot destroy final Fe(III)–oxalate

complexes. Solar photoelectro-Fenton treatments always yield quicker degradation rate

due to the very fast photodecarboxylation of these complexes by UVA irradiation supplied

by solar light. The effect of current density on the degradation rate, efficiency and energy

cost of all methods is examined. The salicylic acid decay always follows a pseudo-

first-order kinetics. 2,3-Dihydroxybenzoic, 2,5-dihydroxybenzoic, 2,6-dihydroxybenzoic,

a-ketoglutaric, glycolic, glyoxylic, maleic, fumaric, malic, tartronic and oxalic acids are

detected as oxidation products. A general reaction sequence for salicylic acid mineraliza-

tion considering all these intermediates is proposed.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Advanced oxidation processes (AOPs) are chemical, photo-

chemical, photocatalytic and electrochemical methods char-

r Ltd. All rights reserved.

fax: +34 93 4021231..

acterized by the in situ generation of hydroxyl radical (dOH).

They are promising environmentally friendly technologies for

the treatment of wastewaters containing low contents of

toxic and biorefractory organics (Tarr, 2003). This is feasible

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1500

because dOH is the second most strong oxidant known after

fluorine and has a very high standard potential (E1(dOH/

H2O) ¼ 2.80 V vs. NHE) that makes it able to non-selectively

react with organic pollutants giving dehydrogenated or

hydroxylated derivatives up to overall mineralization, i.e.,

total conversion into CO2 and inorganic ions. Recently, there

is increasing interest in the use of electrochemical methods

such as anodic oxidation (AO) and indirect electro-oxidation

methods with H2O2 electrogeneration, so-called electroche-

mical advanced oxidation processes (EAOPs), which can

produce dOH as the main oxidizing agent by different ways

(Brillas et al., 2000; Kraft et al., 2003).

The most popular EAOP is AO consisting in the destruction

of organics in an electrolytic cell under the action of hydroxyl

radical formed as intermediate from water oxidation to O2 at

the surface of a high O2-overvoltage anode (Marselli et al.,

2003; Panizza and Cerisola, 2005):

MþH2O!MðdOHÞ þHþ þ e�; (1)

where M(dOH) denotes the hydroxyl radical adsorbed on the

anode M or remaining near its surface. In the past years, AO

has attracted great attention for wastewater remediation due

to the use of a boron-doped diamond (BDD) thin-film anode

that has technologically important characteristics such as an

inert surface with low adsorption properties, remarkable

corrosion stability and a wide potential windows in aqueous

medium (Panizza and Cerisola, 2005). These properties confer

to the BDD electrode a much greater O2-overvoltage than a

conventional anode such as Pt, allowing the production of

higher amount of reactive BDD(dOH) than Pt(dOH) from

reaction (1) and enhancing the oxidation rate of most

organics. Different authors have shown that in aqueous

medium several organic pollutants can be completely miner-

alized by AO with BDD, whereas the use of a Pt anode under

comparable conditions leads to weak decontamination be-

cause of the formation of carboxylic acids that are hardly

oxidized with Pt(dOH) (Brillas et al., 2000, 2004, 2005; Montilla

et al., 2002; Kraft et al., 2003; Marselli et al., 2003; Canizares

et al., 2005; Panizza and Cerisola, 2005; Boulbaba et al., 2006;

Flox et al., 2006; Sires et al., 2007).

More potent indirect electro-oxidation methods with hy-

drogen peroxide electrogeneration are also being developed

for wastewater remediation. In these techniques, H2O2 is

continuously supplied to the contaminated solution from the

two-electron reduction of O2 usually at carbon-felt (Drogui

et al., 2001; Oturan et al., 2001; Gozmen et al., 2003; Hanna

et al., 2005; Irmak et al., 2006; Diagne et al., 2007) and carbon-

polytetrafluoroethylene (PTFE) O2-diffusion (Brillas et al.,

2000, 2004; Flox et al., 2006, 2007; Sires et al., 2006, 2007)

cathodes:

O2ðgÞ þ 2Hþ þ 2e� ! H2O2: (2)

When an undivided electrolytic cell is used, H2O2 is

oxidized to O2 at the anode with formation of hydroperoxyl

radical ðHOd2 Þ as intermediate, a much weaker oxidant than

dOH (Brillas et al., 2000):

H2O2 ! HO2d þHþ þ e�: (3)

Hydrogen peroxide is then accumulated in the medium up

to reaching a steady concentration directly proportional to

the applied current, just when the rates of reactions (2) and (3)

become equal. The direct use of this procedure, so-called

anodic oxidation with electrogenerated H2O2 (AO-H2O2),

involves the oxidation of organics mainly by M(dOH) formed

from reaction (1), although they can also be destroyed by

weaker oxidizing agents like H2O2 and HOd2 .

In acidic medium, the oxidizing power of H2O2 can be

strongly enhanced using the electro-Fenton method (EF),

where a small quantity of Fe2+ is added as catalyst to the

contaminated solution to generate dOH and Fe3+ from

Fenton’s reaction (Sun and Pignatello, 1993):

Fe2þ þH2O2 ! Fe3þ þ dOHþOH�: (4)

An advantage of EF is that reaction (4) is propagated from

Fe2+ regeneration that mainly takes place by reduction of Fe3+

at the cathode (Oturan et al., 2001). Our group has tested the

behavior of this method using an undivided cell with a Pt or

BDD anode and found that pollutants are destroyed by

Pt(dOH) or BDD(dOH) produced from reaction (1) and by dOH

formed in the medium from Fenton’s reaction (4) (Brillas et al.,

2000, 2004; Flox et al., 2006, 2007; Sires et al., 2006, 2007).

Another EAOP is the photoelectro-Fenton (PEF) process in

which the solution treated under EF conditions is simulta-

neously irradiated with UVA light. The action of this irradia-

tion is complex and can be accounted for by: (i) the

production of greater amount of dOH from photoreduction

of Fe(OH)2+, the predominant Fe3+ species in acid medium

(Sun and Pignatello, 1993), by reaction (5) and (ii) the

photolysis of complexes of Fe(III) with generated carboxylic

acids, as shown, for example, in reaction (6) for oxalic acid

(Zuo and Hoigne, 1992):

FeðOHÞ2þ þ hn! Fe2þ þ dOH; (5)

2FeðC2O4Þnð3�2nÞ þ hn! 2Fe2þ þ ð2n� 1Þ C2O4

2� þ 2CO2: (6)

Oxalic acid is produced during the oxidation of most

organics, and the fast photodecarboxylation of Fe(III)–oxalate

complexes ðFeðC2O4Þþ; FeðC2O4Þ

�2 ; FeðC2O4Þ

3�3 Þ favors the de-

contamination process (Brillas et al., 2000; Irmak et al., 2006;

Sires et al., 2006, 2007). Under these conditions, it is also

feasible to use sunlight as an alternative inexpensive source

of UVA light using the solar photoelectro-Fenton (SPEF)

method (Flox et al., 2007).

Thousands of tons of pharmaceutical drugs are consumed

yearly worldwide in human and veterinary medicine and

agricultural products. Because of the inefficient destruction of

their wastewaters in sewage treatment plants (STPs), a fairly

large number of these compounds have been recently

detected in surface, ground and even drinking waters at low

contents of up to micrograms per liter (Daughton and Jones-

Lepp, 2001; Kummerer, 2001; Heberer, 2002a, b; Andreozzi

et al., 2003; Nakada, et al., 2006). The possible interactions of

these pollutants with living beings in the environment are not

well documented, although available data indicate that some

drugs can affect the endocrine system of fishes, can exert

toxic effects on algae and invertebrates and can favor the

development of multi-resistant strains of microorganisms

(Balcioglu and Otker, 2003). This makes necessary the

development of powerful oxidation methods to efficiently

remove drugs and their metabolites from wastewaters for

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 501

avoiding their potential adverse health effects on human

beings and animals.

In previous work, we have reported the degradation of

acidic solutions of the drug paracetamol (Sires et al., 2005,

2006) and the metabolite clofibric acid (Sires et al., 2007) by

AO, EF and PEF, showing that mineralization is more efficient

using a BDD anode than a Pt one. To gain a better knowledge

of the characteristics of EAOPs to decontaminate wastewaters

containing drugs and their by-products, we have undertaken

a comparative study on the destruction of salicylic acid

(2-hydroxybenzoic acid) by such procedures using both Pt and

BDD anodes, with special attention to the possible application

of SPEF that has not been previously tested for these

compounds. Salicylic acid is used in many pharmaceutical

and cosmetic formulations, being easily produced from

hydrolytic deacetylation of the common drug acetylsalicylic

acid (aspirin), which is the main source of its presence in STP

effluents and natural waters (Heberer, 2002a, b; Andreozzi et

al., 2003; Nakada et al., 2006). Several authors have treated

aqueous solutions of this product by AO. Thus, partial

decontamination with either a Pt or a C fiber anode

(Weichgrebe et al., 2004) and complete mineralization with a

BDD anode (Montilla et al., 2002; Marselli et al., 2003; Boulbaba

et al., 2006) have been found. In addition, partial degradation

of salicylic acid by AO-H2O2 with a DSA anode and a carbon-

felt cathode (Drogui et al., 2001) and generation of polyhy-

droxylated derivatives under the action of electrochemically

generated dOH (Oturan et al., 1992) have also been described.

This paper reports a study on the treatment of acidic

synthetic wastewaters of salicylic acid by AO, AO-H2O2, EF,

PEF and SPEF using a Pt or BDD anode under comparable

conditions. The influence of current density on the degrada-

tion rate, mineralization current efficiency and energy cost

for total mineralization of all methods was explored. The

effect of Fe2+ concentration and pH in the EF and PEF

processes was also examined. Oxidation products were

detected by gas chromatography–mass spectrometry

(GC–MS). The salicylic acid decay and the evolution of

intermediates were followed by chromatographic techniques.

2. Materials and methods

2.1. Chemicals

Salicylic acid was of analytical grade (499% purity), supplied

by Fluka, and used in the electrolytic experiments without

further purification. 2,3-Dihydroxybenzoic, 2,5-dihydroxyben-

zoic, 2,6-dihydroxybenzoic, a-ketoglutaric, glycolic, glyoxylic,

maleic, malic, fumaric, tartronic and oxalic acids were either

reagent or analytical grade, purchased from Sigma, Aldrich,

Merck, Panreac and Avocado. Analytical grade sulfuric acid

from Merck was used to adjust the initial solution pH.

Anhydrous sodium sulfate and heptahydrated ferrous sulfate

were of analytical grade from Fluka. All solutions were

prepared with pure water obtained from a Millipore Milli-Q

system, with resistivity 418 MO cm at 25 1C. Organic solvents

and other chemicals used were either HPLC or analytical

grade from Panreac and Aldrich.

2.2. Electrolytic systems

All electrolytic treatments were conducted in an open,

undivided and thermostated conic glass cell containing

100 mL of solution vigorously stirred with a magnetic bar.

Experiments were made at constant current density (j)

supplied with an Amel 2049 potentiostat-galvanostat. The

applied cell voltage was directly measured with a Demestres

605 BR digital multimeter. AO degradations (without H2O2 in

solution) were performed using either a 3 cm2 Pt sheet of

99.99% purity from SEMPSA (AO-Pt method) or a 3 cm2 BDD

thin film deposited on a conductive Si sheet from CSEM (AO-

BDD method) as anode and a 3 cm2 graphite bar from Sofacel

as cathode. A cathode of 3 cm2 carbon-PTFE cloth from E-TEK,

fed with pure O2 at 12 mL min�1, was used for continuous

H2O2 electrogeneration from reaction (2). The preparation and

characteristics of this O2-diffusion cathode are described

elsewhere (Brillas et al., 2000). AO-H2O2, EF, PEF and SPEF

treatments were then carried out using this cathode and

either the above Pt (AO-H2O2-Pt, EF-Pt, PEF-Pt and SPEF-Pt

methods) or BDD (AO-H2O2-BDD, EF-BDD, PEF-BDD and

SPEF-BDD methods) anode. PEF trials became operative

when the solution was irradiated with UVA light supplied

by a Philips 6 W fluorescent black light blue tube emitting

between 300 and 420 nm, with lmax ¼ 360 nm. The tube

was placed at the top of the open cell, at 7 cm above the

solution, giving a photoionization energy input to the

solution of 140mW cm�2, as detected with an NRC 820 laser

power meter working at 514 nm. In the SPEF treatments the

cell was directly exposed to solar irradiation, with a mirror at

its bottom to better collect the sun rays. These trials were

made in sunny and clear days during July 2006 in our

laboratory of Barcelona (longitude: 411210 N, latitude: 21100

E). The average solar irradiation intensity varied between 850

and 960 W m�2, with direct UV irradiation intensity between

20 and 23 W m�2, as measured by the weather station of our

center.

Solutions with 164 mg L�1 salicylic acid (corresponding to

100 mg L�1 total organic carbon (TOC)) and 0.05 M Na2SO4 as

background electrolyte of pH 3.0 were comparatively de-

graded by all methods at 33, 100 and 150 mA cm�2. The pH

value of 3.0 was chosen since it is close to the optimum pH of

2.8 for Fenton’s reaction (4) (Sun and Pignatello, 1993). The

effect of pH in the range 2.0–6.0 and Fe2+ concentration

between 0.2 and 2.0 mM in the EF and PEF processes was also

studied. All trials were made at 35 1C, which is the maximum

temperature to operate with the open cell without significant

water evaporation from solution (Sires et al., 2006).

2.3. Apparatus and analytical procedures

The solution pH was measured with a Crison 2000 pH-meter.

All samples withdrawn from treated solutions were filtered

with 0.45mm PTFE filters from Whatman before analysis.

The mineralization of salicylic acid solutions was monitored

from the decay of their TOC, determined on a Shimadzu 5050

TOC analyzer. Reproducible TOC values were obtained in all

cases, with an accuracy of 72%. These data allowed

calculating the mineralization current efficiency (MCE, in %)

for each treated solution at a given electrolysis time t (h) from

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1502

the equation:

MCE ¼nFVsDðTOCÞexp

4:32� 107 mIt� 100, (7)

where n is the number of electrons consumed in the

mineralization process, F is the Faraday constant

( ¼ 96,487 C mol�1), Vs is the solution volume (L), D(TOC)exp is

the experimental TOC decay (mg L�1), 4.32�107 is a conver-

sion factor ( ¼ 3600 s h�1�12,000 mg of C mol�1), m is the

number of carbon atoms in a salicylic acid molecule ( ¼ 7) and

I is the applied current (A). The n value was taken as 28,

considering that salicylic acid is completely mineralized to

CO2 as follows:

C7H6O3 þ 11H2O! 7CO2 þ 28Hþ þ 28e�: (8)

The energy cost for total mineralization (ETM, in kWh m�3)

of a given solution at time tTM (h) was determined from the

expression

ETM ¼VItTM

Vs, (9)

where V is the average cell voltage (V).

To identify the intermediates formed during salicylic acid

degradation, several solutions with 164 mg L�1 of this com-

pound at pH 3.0 were electrolyzed by AO-Pt at 33 and

150 mA cm�2 for 3 h. Direct extraction of resulting organics

with CH2Cl2 only allowed identifying the residual starting

compound by GC–MS. However, stable aromatics and some

carboxylic acids were detected by this technique after their

silylation. To do this, about 15 mL of each degraded solution

was lyophilized and the remaining solid was treated with

100mL of N,O-bis-(trimethylsilyl)acetamide under stirring and

heating at 60–80 1C for 10 min. The resulting trimethylsilyl

derivatives were diluted with 1 mL of ethyl acetate and further

separated and detected by GC–MS using a Hewlett-Packard

5890 Series II gas chromatograph fitted with a HP-5 0.25mm,

30 m�0.25 mm (i.d.), column, and a Hewlett-Packard 5989 A

mass spectrophotometer operating in EI mode at 70 eV. The

temperature ramp for this column was 50 1C for 3 min,

10 1C min�1 up to 300 1C and hold time 5 min, and the

temperature of the inlet, transfer line and detector was 250,

250 and 290 1C, respectively.

The evolution of salicylic acid and its aromatic products

was followed by reversed-phase HPLC chromatography using

a Waters 600 high-performance liquid chromatograph fitted

with a Spherisorb ODS2 5mm, 150 mm�4.6 mm (i. d.), column

at room temperature, and coupled with a Waters 996

photodiode array detector, which was selected for each

compound at the maximum wavelength of its UV absorption

band. Generated carboxylic acids were detected and quanti-

fied by ion-exclusion chromatography using the above HPLC

chromatograph fitted with a Bio-rad Aminex HPX 87H,

300 mm�7.8 mm (i.d.), column at 35 1C and selecting the

photodiode detector at l ¼ 210 nm. For these analyses, a

50:45:5 (v/v/v) methanol/phosphate buffer (pH ¼ 2.5)/penta-

nol mixture at 0.5 mL min�1 for reversed-phase HPLC chro-

matography and 4 mM H2SO4 at 0.6 mL min�1 for ion-

exclusion HPLC chromatography were used as mobile phases.

3. Results and discussion

3.1. Comparative degradation by EAOPs

A first series of trials was carried out by electrolyzing

164 mg L�1 of salicylic acid at pH 3.0, 33 mA cm�2 and 35 1C

for 6 h to clarify the comparative oxidation power of EAOPs

tested. The EF, PEF and SPEF treatments were operative using

0.5 mM Fe2+ as catalyst. In these experiments, the solution pH

dropped slightly with prolonging electrolysis time due to the

generation of strong acid by-products, and then it was

continuously regulated to pH 3.0 by adding small volumes

of 1 M NaOH. Treated solutions by AO and AO-H2O2 were

always colorless. In contrast, they acquired an intense violet

color after a few minutes of degradation by all EAOPs with

Fe2+, probably due to the formation of complexes of the

remaining salicylic acid with Fe(III) largely generated from

Fenton’s reaction (4) (Ogawa and Tobe, 1966). Violet solutions

were further decolorized, progressively changing from brown

to pale yellow color and becoming colorless again at

60–90 min of electrolysis when all aromatics were already

removed.

The TOC abatement with specific charge (Q, in Ah L�1)

found for the above trials using a Pt and BDD anode is

presented in Fig. 1a and b, respectively. Table 1 summarizes

the corresponding percentage of TOC removal determined

after 60 and 180 min of electrolysis. Comparison of these

results allows concluding that AO, AO-H2O2 and EF treat-

ments yield lower decontamination with Pt than with BDD,

but the catalytic action of UVA and solar light is so effective

that the PEF and SPEF methods show a quite similar

degradation rate for both anodes. The oxidation power of

EAOPs for a given anode increases in the order: AOpAO-

H2O2oEFoPEFoSPEF.

As can be seen in Fig. 1a, the AO-Pt method has the lowest

oxidation power because it only leads to 13% mineralization

at 6 h, indicating that reaction (1) produces a small quantity of

Pt(dOH) with ability to oxidize the organic pollutants to CO2.

The use of AO-H2O2-Pt yields a quicker degradation from

2 Ah L�1 (120 min), yielding 29% of final TOC decay, as

expected if some intermediates can be slowly mineralized

with H2O2 electrogenerated from reaction (2) and/or HOd2

formed from reaction (3). In contrast, Fig. 1b shows that the

analogous AO-BDD and AO-H2O2-BDD methods give a much

faster and similar TOC removal, attaining 85% of the final

decontamination in both cases. These results evidence that

contaminants are mainly destroyed by BDD(dOH), which is

produced in greater amount than Pt(dOH) by reaction (1), with

small participation of weak oxidants like H2O2 and HOd2 .

A different behavior can also be observed in Fig. 1a and b for

the EF process depending on the anode employed. In EF-Pt,

TOC is rapidly reduced by 57% for 180 min (see Table 1), that

is, up to 3 Ah L�1, but at longer time it is slowly removed to

67% of the decontamination at 6 h. The faster degradation

found at the early stages of EF-Pt in comparison to AO-H2O2-

Pt (see Fig. 1a) can then be explained by the quicker reaction

of pollutants with additional dOH produced in the medium

from Fenton’s reaction (4), whereas the strong inhibition of

the EF-Pt process at long times can be ascribed to the

formation of stable complexes of Fe(III) with final short-chain

ARTICLE IN PRESS

TO

C /

mg

L-1

0

20

40

60

80

100

120

0

20

40

60

80

100

0 1 2 3 4 5 6 7

Q / Ah L-1

a

b

Fig. 1 – TOC decay with specific charge for 100 mL of

164 mg L�1 salicylic acid solutions in 0.05 M Na2SO4 of pH

3.0 at 33 mA cm�2 and at 35 1C treated with: (a) Pt and (b) BDD

anode, both of 3 cm2 area. Method: (J,K) anodic oxidation

using a 3 cm2 graphite cathode (AO); (&,’) anodic oxidation

with electrogenerated H2O2 (AO-H2O2) using a 3 cm2 O2-

diffusion cathode; (W,m) electro-Fenton (EF) with 0.5 mM

Fe2+ in solution as catalyst; (X,.) photoelectro-Fenton (PEF)

with 0.5 mM Fe2+ and 6 W UVA light of kmax ¼ 360 nm as

catalysts; and (B,E) solar photoelectro-Fenton (SPEF) with

0.5 mM Fe2+ under direct solar irradiation.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 503

carboxylic acids that cannot be destroyed by dOH and Pt(dOH)

(Brillas et al., 2000, 2004; Sires et al., 2006). Results of Fig. 1b

also confirm the greater oxidation power of dOH than

BDD(dOH), since TOC decays much more rapidly in EF-BDD

than in AO-H2O2-BDD. However, the degradation rate of the

EF-BDD treatment gradually decreases up to reach 91% of

TOC abatement at 6 h, suggesting that some products

including complexes of Fe(III) with final carboxylic acids are

slowly oxidized with BDD(dOH) because they cannot be

removed by dOH. The parallel quicker photodecomposition

of these complexes to CO2 under an UVA irradiation (Zuo and

Hoigne, 1992) can be accounted for by the total mineralization

(498% TOC decay) achieved after 3 Ah L�1 (180 min) of PEF-Pt

and PEF-BDD, as well as after 2 Ah L�1 (120 min) of SPEF-Pt

and SPEF-BDD (see Fig. 1a and b). These findings allow

concluding that the oxidation power of both methods is

independent of the anode employed, the SPEF procedure

always being more potent than the PEF one. Solar light can

then be used as source of UVA irradiation for an effective and

complete mineralization of acidic wastewaters containing

salicylic acid.

3.2. Effect of current density

The influence of current density on the oxidation power

of all EAOPs was further examined by treating the above

salicylic acid solutions of pH 3.0 at 33, 100 and 150 mA cm�2.

Selected results listed in Table 1 show a gradual rise in TOC

removal with increasing j from 33 to 150 mA cm�2 in all

cases. This trend evidences an enhancement of the oxi-

dation power of all methods under these conditions, which

can be related to a quicker destruction of organic pollutants

due to the increase in rate of reaction (1) to form more

amount of Pt(dOH) or BDD(dOH), as well as the greater H2O2

electrogeneration at the O2-diffusion cathode via reac-

tion (2) producing more quantity of dOH from Fenton’s

reaction (4). The fact that final carboxylic acids are more

rapidly formed under the action of these oxidants

suggests the concomitant acceleration of the photodecompo-

sition of their Fe(III) complexes in PEF-Pt, PEF-BDD, SPEF-Pt

and SPEF-BDD, in agreement with the higher oxidation

power found for these procedures when j rises (see Table 1).

Overall mineralization is then achieved for all EAOPs

with BDD and for PEF-Pt and SPEF-Pt at decreasing times

with increasing current density, as can be deduced from

data given in Table 2. Note that a similar tTM-value is found

either for the PEF or for the SPEF treatment with both anodes

at each j, thus confirming the quite effective action of

UVA irradiation to photolyze all complexes between Fe(III)

and final carboxylic acids. At 150 mA cm�2, for example,

the degradation of 164 mg L�1 salicylic acid solutions in

SPEF-Pt and SPEF-BDD is so rapid by the combined action

of oxidants and UVA irradiation from solar light that they

can be completely mineralized in a time as short as 60 min

(see Table 2).

Note that all methods need the consumption of more

specific charge to achieve a given TOC decay when j rises, as

can be observed in Fig. 2a for the SPEF-BDD treatment.

In this case, total mineralization is achieved by consum-

ing greater Q-values of 2.0, 3.6 and 4.5 Ah L�1 at increasing

current densities of 33, 100 and 150 mA cm�2. This behavior

seems opposite to the faster degradation found for all

processes as more j is applied. This apparent contradiction

can be explained if higher current density causes the

production of larger amounts of Pt(dOH), BDD(dOH) and/ordOH favoring the removal of organics with time, but consum-

ing greater specific charge because these oxidants are

generated to a less relative extent due to the faster accelera-

tion of their non-oxidizing reactions. So, in the AO and AO-

H2O2 processes, Pt(dOH) or BDD(dOH) is expected to be more

rapidly oxidized to O2 from reaction (10) and recombined into

H2O2 from reaction (11) with increasing j, whereas when BDD

is used less relative production of BDD(dOH) takes place

due to the enhancement of the anodic generation of weaker

oxidants such as S2O2�8 and O3 via reactions (12) and (13),

respectively (Panizza and Cerisola, 2005). In EF, PEF and

SPEF, it is expected that a larger proportion of dOH formed

from Fenton’s reaction (4) is wasted by Fe2+ from reaction (14)

(Sun and Pignatello, 1993), along with the relative decay of

ARTICLE IN PRESS

Table 1 – Percentage of TOC removal and mineralization current efficiency determined from Eq. (7) for 100 mL of 164 mg L�1

salicylic acid solutions in 0.05 M Na2SO4 of pH 3.0 at 35 1C treated by different electrochemical advanced oxidationprocesses under selected conditions

Methoda Current density (mA cm�2) After 60 min of treatment After 180 min of treatment

% TOC removal MCE % TOC removal MCE

AO-Pt 33 5.3 4.7 8.5 2.5

100 6.5 1.9 12 1.2

150 7.6 1.5 14 0.9

AO-H2O2-Pt 33 4.5 4.0 14 4.2

100 6.0 1.8 17 1.7

150 7.4 1.5 23 1.5

AO-BDD 33 22 20 64 19

100 24 7.1 69 6.8

150 27 5.4 81 5.4

AO-H2O2-BDD 33 22 20 61 18

100 25 7.4 73 7.2

150 31 6.1 84 5.6

EF-Ptb 33 30 27 57 17

100 47 14 62 6.1

150 54 11 64 4.2

EF-BDDb 33 41 37 73 22

100 53 16 85 8.4

150 67 13 90 6.0

PEF-Ptb 33 44 39 97 29

100 75 22 –c –c

150 77 15 –c –c

PEF-BDDb 33 54 48 96 28

100 67 20 –c –c

150 73 14 –c –c

SPEF-Ptb 33 85 76 –c –c

100 95 28 –c –c

150 97 19 –c –c

SPEF-BDDb 33 82 73 –c –c

100 96 28 –c –c

150 98 19 –c –c

a AO-Pt: anodic oxidation with a Pt anode and a graphite cathode; AO-H2O2-Pt: anodic oxidation with electrogenerated H2O2 using a Pt anode;

AO-BDD: anodic oxidation with a BDD anode and a graphite cathode; AO-H2O2-BDD: anodic oxidation with electrogenerated H2O2 using a BDD

anode; EF-Pt: electro-Fenton with Pt; EF-BDD: electro-Fenton with BDD; PEF-Pt: photoelectro-Fenton with Pt; PEF-BDD: photoelectro-Fenton

with BDD; SPEF-Pt: solar photoelectro-Fenton with Pt; SPEF-BDD: solar photoelectro-Fenton with BDD.b The initial solution contained 0.5 mM Fe2+ as catalyst.c Overall mineralization (498% TOC decay) attained at smaller time.

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1504

Pt(dOH) or BDD(dOH):

2dOH! O2ðgÞ þ 2Hþ þ 2e�; (10)

2dOH! H2O2; (11)

2SO42� ! S2O8

2� þ 2e�; (12)

3H2O! O3ðgÞ þ 6Hþ þ 6e�; (13)

Fe2þ þ dOH! Fe3þ þOH�: (14)

The influence of current density on specific charge affects

notably the mineralization current efficiency and energy cost

of each degradation process, as will be discussed below.

3.3. Effect of Fe2+ content and pH

The degradation rate of the EF, PEF and SPEF treatments of

salicylic acid is expected to depend on the initial concentra-

tion of catalyst Fe2+ and solution pH, because they limit, at

least, the quantity of oxidant dOH produced in the medium

ARTICLE IN PRESS

Table 2 – Energy cost for total mineralization calculatedfrom Eq. (9) for 164 mg L�1 salicylic acid solutionsdegraded by electrochemical advanced oxidation pro-cesses

Method Currentdensity

(mA cm�2)

Averagecell

voltage(V)

tTMa

(min)ETM

(kWh m�3)

AO-BDD 100 14.2 360 256

150 18.1 300 407

AO-

H2O2-

BDD

100 17.3 330 285

150 20.9 270 423

EF-BDDb 100 17.1 360 308

150 21.0 300 472

PEF-Ptb 33 5.3 180 34c

100 13.2 150 114c

150 17.5 120 170c

PEF-

BDDb

33 10.2 180 49c

100 17.0 150 142c

150 20.8 130 219c

SPEF-Ptb 33 5.0 120 10

100 12.8 85 54

150 17.1 60 77

SPEF-

BDDb

33 9.9 120 20

100 16.8 75 61

150 20.7 60 93

a Time for total mineralization.b The initial solution contained 0.5 mM Fe2+ as catalyst.c Value including the energy consumption of the 6 W UVA lamp

used to irradiate the solution.

0

20

40

60

80

100

120

0

20

40

60

80

100

TO

C /

mg

L-1

0

20

40

60

80

100

0 1 2 3 4 5 6 7

Q / Ah L-1

a

b

c

Fig. 2 – Effect of experimental parameters on TOC removal of

100 mL of 164 mg L�1 salicylic acid solutions in 0.05 M

Na2SO4 at 35 1C. In plot (a), SPEF-BDD process with 0.5 mM

Fe2+ at pH 3.0 applying: (K) 33 mA cm�2; (’) 100 mA cm�2;

and (m) 150 mA cm�2. In plot (b), EF-BDD degradation with a

Fe2+ concentration of: (K) 0.2 mM; (’) 0.5 mM; (m) 1.0 mM;

and (.) 2.0 mM, at pH 3.0 and at 33 mA cm�2. In plot (c), EF-Pt

treatment with 0.5 mM Fe2+ at pH: (K) 2.0; (’) 3.0; (m) 4.0;

and (.) 6.0, and at 33 mA cm�2.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 505

from Fenton’s reaction (4) (Sun and Pignatello, 1993). The

effects of such parameters were studied for the EF and PEF

processes, being more clearly observed in the first method

due to their lower oxidation power. Fig. 2b presents the TOC–Q

plots obtained for the EF-BDD degradations of 164 mg L�1 of

salicylic acid in the presence of 0.2–2.0 mM Fe2+, at pH 3.0 and

33 mA cm�2. A similar TOC decay can be observed within the

Fe2+ range 0.2–1.0 mM, leading to 91–96% mineralization at

6 h. However, the use of 2.0 mM Fe2+ causes a strong inhibition

of the degradation process, mainly from 1.5 Ah L�1 (90 min), so

that only 82% of the initial TOC is removed at the end of this

trial. This tendency evidences the production of similar

amounts of oxidants BDD(dOH) and dOH up to 1.0 mM Fe2+,

whereas at higher contents of this ion both species are

significantly wasted by reaction (14) taking place in both the

vicinity of the anode, where BDD(dOH) formed from reaction

(1) is accumulated, and the solution bulk, where dOH is

generated from Fenton’s reaction (4). A similar influence of

Fe2+ concentration was found for the analogous EF-Pt

treatments of the same solution. According to these results,

a small amount of this catalyst, even as low as 0.2 mM, is

enough to obtain the fastest destruction of all contaminants

by Pt(dOH), BDD(dOH) and/or dOH in all indirect electro-

oxidation methods.

Fig. 2c shows a significant effect of pH on the EF-Pt process

of 164 mg L�1 salicylic acid solutions with 0.5 mM Fe2+ at

33 mA cm�2. These trials were made with continuous regula-

tion of the solution pH to its initial value with 1 M NaOH. As

can be seen, decontamination is very fast at pH 3.0 and 4.0,

where TOC is reduced by 67% and 74%, respectively, at 6 h. At

pH 2.0, however, the degradation rate drops slightly to yield

59% of final TOC removal, but at pH 6.0 it is so strongly

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1506

inhibited that a poor mineralization of 26% is attained. This

trend confirms that organics are more rapidly destroyed withdOH formed from Fenton’s reaction (4) than with Pt(dOH)

generated from reaction (1), since the rate of the first reaction

depends on pH (Sun and Pignatello, 1993), whereas the second

one is pH-independent (Sires et al., 2006). Similar treatments

by EF-BDD showed similar results but with greater removal of

pollutants, even at pH 6.0, because they can also be oxidized

with BDD(dOH). One can then establish that pH values of

3.0–4.0 are optimal to decontaminate salicylic acid waste-

waters by the indirect electro-oxidation methods.

3.4. Mineralization current efficiency and energy cost

The mineralization current efficiency for salicylic acid

degradation by all EAOPs was calculated from Eq. (7) and

selected values thus obtained are collected in Table 1. Fig. 3

shows the comparative MCE–Q plots determined for the trials

using a BDD anode at 33 mA cm�2 reported in Fig. 1b. As

expected, the efficiency increases in the same sequence as

the oxidation power of the methods, that is: AO-BDDEAO-

H2O2-BDDoEF-BDDoPEF-BDDoSPEF-BDD. For both AO pro-

cesses, a constant MCE value of 20% is reached approximately

up to 3 Ah L�1 (180 min), whereupon it decays slightly to 13%

at 6 h. This behavior suggests a practically constant destruc-

tion rate of all pollutants mainly by reaction with BDD(dOH) in

both cases. In contrast, all indirect electro-oxidation methods

present a maximum efficiency at 1 Ah L�1 (60 min), with

increasing values of 37% for EF-BDD, 48% for PEF-BDD and

73% for SPEF-BDD, which further fall to 14%, 29% and 44% at

the end of the corresponding runs. The increase in MCE at the

early stages of these processes can be ascribed to the initial

formation of large amounts of products easily mineralized

with BDD(dOH) and mainly with dOH, whereas the drop of this

parameter at longer electrolysis time can be associated with

the production of complexes of Fe(III) with final carboxylic

acids that are difficult to oxidize by BDD(dOH) and/or slowly

photolyze by UVA or solar light. The highest photoefficiency

of UVA irradiation supplied by solar light then explains the

0

10

20

30

40

50

60

70

80

0 1 2 3 4 5 6 7

MC

E /

%

Q / Ah L-1

Fig. 3 – Mineralization current efficiency calculated from Eq. (7)

vs. specific charge for the experiments reported in Fig. 1b.

greatest oxidation power and MCE of the SPEF-BDD method.

Similar relative efficiencies were found for the experiments

given in Fig. 1a using a Pt anode, although lower MCE values

were obtained for the AO-Pt, AO-H2O2-Pt and EF-Pt processes

due to the smaller oxidation power of Pt(dOH). On the other

hand, an inspection of Table 1 reveals a gradual decay of MCE

with rising j for all EAOPs, which can be related to the higher

increase in rate of non-oxidizing reactions of hydroxyl radical

involving its oxidation to O2 and reactions (10), (11) and/or

(14), as well as the enhancement of the formation of weaker

oxidants like S2O2�8 by reaction (12) and O3 by reaction (13).

This gives rise to the production of lower relative quantities of

Pt(dOH), BDD(dOH) and/or dOH with the consequent con-

sumption of more specific charge and loss of efficiency. This

effect is very notable even for the two SPEF treatments for

which the MCE value at 60 min decays from 73–76% at

33 mA cm�2 to 19% at 150 mA cm�2, with an increase in

mineralization from 82–85% to ca. 98%.

Table 2 summarizes the energy cost for total mineralization

calculated from Eq. (9) for the above trials. An analysis of

these data reveals an increase of ETM in each method when j

increases due to the higher specific charge consumed,

although this parameter strongly decreases for both PEF and

SPEF treatments since much more TOC is removed by their

higher oxidation power. In addition, indirect electro-oxidation

processes become more economic using Pt than BDD because

of the lower average cell voltage applied. This means that

SPEF-Pt is the most viable EAOP for the treatment of salicylic

acid solutions under the present conditions, since it yields

overall mineralization with the lowest energy consumption of

10 kWh m�3 at 33 mA cm�2 (see Table 2).

3.5. Salicylic acid decay

The kinetics for salicylic acid decay by the different EAOPs

tested was followed by reverse-phase HPLC chromatography,

in which it exhibited a well-defined absorption peak with a

retention time (tr) of 5.4 min. No change in the concentration

of this compound was previously found when 20 mM H2O2,

0.5 mM Fe2+ or 0.5 mM Fe3+ were separately added to a

164 mg L�1 salicylic acid solution in 0.05 M Na2SO4 of pH 3.0

and further exposed either to UVA or to solar light for 60 min

without applying current. This brings us to consider that in all

EAOPs the electroactive species (salicylic acid and/or mainly

its complexes with Fe(III)) is degraded neither by electro-

generated H2O2 nor by incident light, and hence it can

react only with Pt(dOH) or BDD(dOH) formed from reaction

(1) and/or dOH generated in the medium from reactions (4)

and/or (5).

Fig. 4a shows the change of salicylic acid concentration

with time for the AO-Pt, AO-H2O2-Pt, AO-BDD and AO-H2O2-

BDD methods at 33 mA cm�2. This compound is so slowly

removed under these conditions that it still remains bet-

ween 7 and 17 mg L�1 after 6 h of these treatments. A similar

decay rate can be observed for both AO and AO-H2O2

processes at each anode, as expected if salicylic acid

does not react with electrogenerated H2O2, although its

removal is slightly enhanced using BDD instead of Pt.

This behavior evidences that it is destroyed by a similar

amount of Pt(dOH) or BDD(dOH) during the AO procedures.

ARTICLE IN PRESS

[sal

icyl

ic a

cid]

/ m

g L

-1

0

50

100

150

200

0 60 120 180 240 300 360 420

time / min

ln (

C0/C

)0.0

0.5

1.0

1.5

2.0

0 60 120 180 240 300

0

50

100

150

200

0 10 20 30 40

time / min

time / min

ln (

C0/C

)

0.0

1.0

2.0

3.0

0 5 10 15 20 25

a

b

Fig. 4 – Salicylic acid decay with electrolysis time under the

same conditions as given in Fig. 1. In plot (a), (J) AO-Pt; (K)

AO-BDD; (&) AO-H2O2-Pt; and (’) AO-H2O2-BDD. In plot (b),

(W) EF-Pt; (m) EF-BDD; (X) PEF-Pt; and (.) PEF-BDD. The

corresponding kinetic analysis assuming a pseudo-first-

order reaction for salicylic acid is given in the inset panels.

Table 3 – Aromatic intermediates and generated car-boxylic acids detected during the degradation of salicylicacid by electrochemical advanced oxidation methods

Compound Analyticaltechniquea

Retentiontime(min)

Molecularmassb

(g mol�1)

2,3-

Dihydroxybenzoic

acid

GC-MS 17.7 385

HPLC 3.5

2,5-

Dihydroxybenzoic

acid

GC-MS 18.0 385

HPLC 3.8

2,6-

Dihydroxybenzoic

acid

HPLC 3.0

a-Ketoglutaric

acid

GC-MS 16.1 362

Glycolic acid HPLC 12.3

Glyoxylic acid HPLC 9.2

Malic acid GC-MS 14.6 350

Maleic Acid GC-MS 12.1 260

HPLC 8.0

Fumaric acid GC-MS 12.6 260

HPLC 15.5

Tartronic acid GC-MS 13.3 336

HPLC 7.7

Oxalic acid HPLC 6.5

a GC–MS analysis was carried out after treating 164 mg L�1 salicylic

acid solution of pH 3.0 by anodic oxidation at 33 and 150 mA cm�2

for 180 min, followed by silylation.b Value corresponding to the trimethylsilyl derivative.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 507

However, Figs. 1 and 4a show that only the BDD anode is able

to remove salicylic acid and simultaneously mineralize the

treated solution at practically the same rate for 360 min

(Q ¼ 6 Ah L�1), suggesting that the major part of reactive

BDD(dOH) produced by this anode reacts rapidly with its

oxidation products to yield CO2.

As can be seen in Fig. 4b, the comparative EF-Pt, PEF-Pt, EF-

BDD and PEF-BDD treatments with 0.5 mM Fe2+ at 33 mA cm�2

yield a very fast and overall removal of salicylic acid in about

30 min in all cases. The quite similar decay rate observed for

electro-Fenton and photoelectro-Fenton agrees with the fact

that Fe(III)–salicylate complexes are not photolyzed by UVA

irradiation, as stated above, and also evidences the existence

of a slow production of dOH via reaction (5). The large

enhancement of the decay rate of salicylic acid in these

indirect electro-oxidation methods in relation to the AO ones

corroborates that the main oxidant is dOH generated from

Fenton’s reaction (4), at least at short electrolysis times.

Good linear correlations were obtained by fitting the above

concentration decays with a pseudo-first-order kinetics, as

presented in the inset of Fig. 4a and b. This analysis allows

determining an analogous apparent first-order rate constant

(k1) of 9.2�10�5 s�1 (square regression coefficient (R2¼ 0.997)

for AO-Pt and 8.7�10�5 s�1 (R2¼ 0.992) for AO-H2O2-Pt, which

slightly rises to 1.1�10�4 s�1 (R2¼ 0.994) and 1.2�10�4 s�1

(R2¼ 0.993) for the same methods with BDD). In contrast,

much greater and similar k1-values of 2.2�10�3 s�1 (R2¼

0.989), 2.1�10�3 s�1 (R2¼ 0.992), 1.8�10�3 s�1 (R2

¼ 0.990) and

2.0�10�3 s�1 (R2¼ 0.990) are found for EF-Pt, PEF-Pt, EF-BDD

and PEF-BDD, respectively. The increase in k1 for these

procedures is due to the much faster degradation with dOH

than with Pt(dOH) or BDD(dOH), as pointed out above. Since

the absolute second-order rate constant for the reaction

between salicylic acid and dOH in acidic medium is

k2 ¼ 2.7�1010 M�1 s�1 (Amphlett et al., 1968), one can esti-

mate, as the first approach, a constant dOH generation of

about 10�13 M ( ¼ k1/k2) from reaction (4) in the indirect

electro-oxidation methods at 33 mA cm�2.

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1508

3.6. Identification and evolution of oxidation products

Table 3 collects the aromatic intermediates and generated

carboxylic acids identified by GC–MS and HPLC chromato-

graphy during the degradation processes of salicylic acid.

Several dihydroxybenzoic acids were detected as primary

products, similar to those previously reported from its

treatment by electrogenerated Fenton’s reagent (Oturan

et al., 1992) and AO-BDD in 1 M HClO4 (Marselli et al., 2003).

The time course of the concentration of these compounds

was determined by reversed-phase HPLC chromatography via

calibration with external standards. In AO-Pt and AO-H2O2-Pt

they were largely accumulated in a similar way due to the

small oxidation power of Pt(dOH). As exemplified in Fig. 5a for

the latter method at 33 mA cm�2, 2,5-dihydroxybenzoic acid is

formed to a larger extent than 2,6-dihydroxybenzoic and 2,3-

dihydroxybenzoic acids, attaining ca. 16, 7 and 3 mg L�1,

respectively, at the end of electrolysis. For AO-BDD and AO-

H2O2-BDD, however, all these products were detected only up

conc

entr

atio

n /

mg

L-1

0

10

20

30

40

0 10 20 30 40 50

time / min

0

5

10

15

20

0 60 120 180 240 300 360 420

a

b

Fig. 5 – Time course of the concentration of

dihydroxybenzoic acids detected during the treatment of

164 mg L�1 salicylic acid solutions in the same conditions as

in Fig. 1. Plot (a) corresponds to (K) 2,5-dihydroxybenzoic

acid; (’) 2,6-dihydroxybenzoic acid; and (m) 2,3-

dihydroxybenzoic acid for AO-H2O2-Pt. Plot (b) corresponds

to 2,5-dihydroxybenzoic acid for (J) EF-Pt and (K) PEF-Pt;

2,6-dihydroxybenzoic acid for (&) EF-Pt; and (’) PEF-Pt and

2,3-dihydroxybenzoic acid for (W) EF-Pt; and (m) PEF-Pt.

to 9 mg L�1 as maximum for 180 min. These results indicate

that dihydroxybenzoic acids are much more rapidly destroyed

by BDD(dOH) than Pt(dOH), contrary to salicylic acid that

reacts at a similar rate with both oxidants (see Fig. 4a). In

contrast, Fig. 5b shows a quick accumulation of all these

compounds using EF-Pt and PEF-Pt, which persist in the

medium for less than 40 min, similar to the starting product

(see Fig. 4b), as expected if they undergo a faster reaction withdOH. The similar evolution of each dihydroxybenzoic acid in

both methods evidences that their Fe(III) complexes are not

directly photolyzed by UVA light. Under these conditions,

Fig. 5a also shows a greater generation of 2,3-dihydroxyben-

zoic acid, suggesting the preferential attack of dOH on the C(3)

position of salicylic acid, whereas Pt(dOH) hydroxylates

mainly its C(5) position to yield 2,5-dihydroxybenzoic acid

(see Fig. 5a). Surprisingly, only traces of these compounds

were detected in EF-BDD and PEF-BDD, confirming their

parallel quick reaction with BDD(dOH), as found in the

corresponding AO treatments.

Linear short-chain carboxylic acids given in Table 3 are

generated from the oxidative breaking of the aryl moiety of

aromatic products. Since these acids are composed of five to

two C atoms, two parallel degradation paths can be envis-

aged involving the opening and rupture of aromatics as

0

1

2

3

4

5

6

7

[mal

eic

acid

] / m

g L

-1

0

20

40

60

80

100

120

0 60 120 180 240 300 360 420

[oxa

lic

acid

] / m

g L

-1

time / min

a

b

Fig. 6 – Evolution of the concentration of (a) maleic acid and

(b) oxalic acid during the degradation of 164 mg L�1 salicylic

acid solutions under the conditions reported in Fig. 1.

Method: (J) AO-Pt; (K) AO-BDD; (&) AO-H2O2-Pt; (’) AO-

H2O2-BDD; (W) EF-Pt; (m) EF-BDD; (X) PEF-Pt and (.) PEF-

BDD.

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 509

dihydroxybenzoic acids containing seven C atoms to give two

carboxylic acids either with (i) five and two C atoms or with

(ii) four and three C atoms. It is then expected that

a-ketoglutaric and glycolic acids are formed in the first

pathway, whereas malic, maleic and fumaric acids along

with tartronic acid are produced in the second one. Glycolic

acid is further transformed into glyoxylic acid (Brillas et al.,

2000). All these acids are independently oxidized to oxalic

acid as the ultimate by-product (Brillas et al., 2005; Flox et al.,

2006, 2007), which is directly mineralized to CO2.

Ion-exclusion chromatograms of treated solutions revealed

the generation of small contents of maleic, fumaric and

tartronic acids, as well as traces of glycolic and glyoxylic

acids, for most processes. The time course of these com-

CO

HO

CO 2

COOH

COOH

O

2,3-dihydroxybenzoic acid

P

salicylic

OH

COOH

OH2,5-dihydroxyb

CO

HO

oxalic a

B

B

.

HOOC

α-ketoglutaric acid

HO

CH2OH

COOH

glycolic acid

COH

COOH

glyoxylic acid

HOOCOH+

OHPt( OH)BDD( OH).

..

+

OPB

.OHPt( OH)

BDD( OH)..

.

OH Pt( OH)

BDD( OH)...

OHPt( OH)

BDD( OH)..

.

Fig. 7 – Proposed reaction sequence for salicylic acid mineraliza

oxidation processes. Pt(dOH) and BDD(dOH) are the oxidant hydro

from water oxidation, whereas dOH is the hydroxyl radical form

pounds at 33 mA cm�2 was similar to that depicted in Fig. 6a

for maleic acid. A continuous accumulation of this acid takes

place in AO-Pt due to the low oxidation power of Pt(dOH),

although it is slightly removed in AO-H2O2-Pt probably by its

slow reaction with H2O2 and/or HOd2 . In contrast, maleic acid

was undetected for both AO treatments with BDD, indicating

its efficient destruction by BDD(dOH). For EF-Pt, PEF-Pt, EF-

BDD and PEF-BDD, this acid reaches a maximum accumula-

tion of 0.9–1.5 mg L�1 at 20 min and disappears at 60 min. This

evidences the most effective reaction of Fe(III)-maleate

complexes with dOH in the medium, without significant

photolysis under UVA irradiation.

A very different behavior can be seen in Fig. 6b for oxalic

acid detected in the trials at 33 mA cm�2. In both AO methods

OH

OH

COOH

OCHOOC

COOH

H

Fe3+

-Fe2+

t( OH)

acid

enzoic acid

OH

OH

2,6-dihydroxybenzoic acid

OH

COOH

HO

cid

Fe(III)-oxalate complexes

DD( OH)

hν

BDD( OH).

.

DD( OH)..

+

maleic acid fumaric acid

OC COOH

malic acidOH

HO

COOH

COOH

tartronic acid

+

Ht( OH)DD( OH).. OH

Pt( OH)BDD( OH).

..

OHPt( OH)BDD( OH).

..

tion in acidic aqueous medium by electrochemical advanced

xyl radicals produced at the Pt and BDD anode, respectively,

ed from Fenton’s reaction.

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1510

with Pt this acid is progressively accumulated to 10 mg L�1 at

360 min, as expected if it does not react with Pt(dOH). The low

oxidation power of these methods can then be accounted for

by the inability of this oxidant to degrade most carboxylic

acids. In contrast, this acid shows a maximum accumulation

of 18 mg L�1 after 120 min of AO-BDD and AO-H2O2-BDD

treatments, decreasing at longer time because it is miner-

alized with BDD(dOH). The ability of this species to oxidize all

carboxylic acids explains the great oxidation power of the AO

methods using BDD, yielding the total mineralization of

salicylic acid at high current density. As can be seen in Fig.

6b, the use of EF-Pt leads to a steady concentration of oxalic

acid close to 120 mg L�1 from 180 min of electrolysis, as

expected if Fe(III)–oxalate complexes are not destroyed either

by dOH or by Pt(dOH). The remaining oxalic acid in the final

solution treated by this method corresponds to 32 mg L�1 of

TOC, in good agreement with the value of 33 mg L�1 experi-

mentally found (see Fig. 1a). This confirms the overall removal

of all generated products by this method, except Fe(III)–ox-

alate complexes that remain stable in the final solution.

Fig. 6b also shows that these complexes are slowly destroyed

by BDD(dOH) in EF-BDD and much more rapidly photodecom-

posed by UVA light in PEF-Pt and PEF-BDD. In the first case,

oxalic acid disappears in 360 min while the final solution still

has 9 mg L�1 of TOC (see Fig. 1b), suggesting that it contains

some undetected products that are hardly oxidized by

BDD(dOH) and dOH. These products are also efficiently

mineralized under the action of UVA light in PEF-Pt and

PEF-BDD, where solutions are completely decontaminated in

180 min (see Figs. 1a and b) when all Fe(III)–oxalate complexes

are already destroyed (see Fig. 6b). The faster photodecompo-

sition of all these final products by UVA irradiation supplied

by solar light then explains the highest oxidation power of the

SPEF-Pt and SPEF-BDD treatments.

3.7. Mineralization sequence

Taking into account the intermediates detected in the present

work, a plausible general reaction sequence for salicylic acid

mineralization is proposed in Fig. 7. This path only remarks

the formation of final Fe(III)–oxalate complexes in the indirect

electro-oxidation methods for sake of simplicity, although the

other carboxylic products also form large amounts of Fe(III)

complexes that are oxidized in parallel following the same

sequence. The main oxidants (dOH, Pt(dOH) and BDD(dOH))

are only specified, also being possible parallel and slower

reactions of some intermediates with weaker oxidizing

agents such as H2O2, HOd2 ; S2O2�

8 , O3, etc., as stated above.

2,3-Dihydroxybenzoic, 2,5-dihydroxybenzoic and 2,6-dihy-

droxybenzoic acids are initially formed from hydroxylation

on the C(3), C(5) and C(6) positions of salicylic acid by dOH,

Pt(dOH) and BDD(dOH), respectively. Further breaking of such

products by these oxidants leads to two mixtures of

carboxylic acids, one of them corresponding to a-ketoglutaric

and glycolic acids and the other to malic, maleic and fumaric

acids along with tartronic acid. Glycolic acid is subsequently

oxidized to glyoxylic acid. All these acids are then rapidly

transformed into oxalic acid by dOH and BDD(dOH), but are

much more slowly oxidized by Pt(dOH). In the AO treatments,

oxalic acid is converted into CO2 only by BDD(dOH) but not by

dOH and Pt(dOH), whereas in all indirect electro-oxidation

methods it forms complexes with Fe(III) that persist up to the

end of electrolysis. These Fe(III)–oxalate complexes remain

stable under EF-Pt conditions, being mineralized by BDD(dOH)

in EF-BDD and more quickly photodecarboxylated under UVA

irradiation in the PEF and SPEF processes, with loss of Fe2+

according to reaction (6).

4. Conclusions

Solutions containing 164 mg L�1 salicylic acid can be totally

mineralized by all EAOPs using a BDD anode, with increasing

oxidation power in the order AO-BDDEAO-H2O2-BDDoEF-

BDDoPEF-BDDoSPEF-BDD. The same sequence is followed

by the same procedures with a Pt anode, although the AO-Pt,

AO-H2O2-Pt and EF-Pt methods yield much lower decontami-

nation under comparable conditions. This behavior can be

accounted for by the slow removal of most pollutants by

Pt(dOH) in comparison to their quick destruction with

BDD(dOH). dOH oxidizes rapidly all aromatic pollutants, but

it cannot mineralize final Fe(III)–oxalate complexes. The very

fast photolysis of these complexes under UVA irradiation

explains the higher oxidation power found for PEF and SPEF.

Solar light is then an inexpensive and useful source of UVA

irradiation for an efficient treatment of acidic wastewaters

containing salicylic acid. The increase in current density

causes a gradual rise in the degradation rate of all methods

due to the higher production of Pt(dOH), BDD(dOH) and dOH.

Their efficiency, however, undergoes a progressive decay

because these oxidants are generated in less relative propor-

tion by the greater acceleration of non-oxidizing reactions

and the enhancement of parallel anodic reactions giving

weaker oxidants such as S2O2�8 and O3. Indirect electro-

oxidation processes operate in optimum conditions at pH

values 3.0–4.0 and with Fe2+ contents between 0.2 and 1.0 mM.

The lowest energy cost for total mineralization is attained for

the SPEF-Pt treatment, of 10 kWh m�3 at 33 mA cm�2. Salicylic

acid always follows a pseudo-first-order decay with a much

lower rate constant for the AO methods than for the indirect

electro-oxidation ones. 2,3-Dihydroxybenzoic, 2,5-dihydrox-

ybenzoic and 2,6-dihydroxybenzoic acids are detected as

primary aromatic products, which are further oxidized to a

mixture of carboxylic acids. The ultimate by-product is oxalic

acid, which can be mineralized only by BDD(dOH) under AO

conditions. In the indirect electro-oxidation methods, it forms

Fe(III) complexes that remain stable in EF-Pt, being miner-

alized by BDD(dOH) in EF-BDD and more quickly photodecar-

boxylated under UVA irradiation in the PEF and SPEF

processes.

Acknowledgments

The authors are grateful to MEC (Ministerio de Educacion y

Ciencia, Spain) for the financial support and the grant given to

E. Guinea to do this work under project CTQ2004-01954/BQU.

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 4 9 9 – 5 1 1 511

R E F E R E N C E S

Amphlett, C.B., Adams, G.E., Michael, B.D., 1968. Pulse radiolysisstudies of deaerated aqueous salicylate solutions. In: Gould,R.F. (Ed.), Radiation Chemistry. I. Aqueous Media, Biology,Dosimetry. Proc. Advances in Chemistry Series 81. AmericanChemical Society, Washington, pp. 231–250.

Andreozzi, R., Marotta, R., Nicklas, P., 2003. Pharmaceuticals inSTP effluents and their solar photodegradation in aquaticenvironment. Chemosphere 50 (10), 1319–1330.

Balcioglu, A.I., Otker, M., 2003. Treatment of pharmaceuticalwastewater containing antibiotics by O3 and O3/H2O2 pro-cesses. Chemosphere 50 (1), 85–95.

Boulbaba, L., Nasr, B., Abdellatif, G., 2006. Electrochemicaloxidation of benzoic acid derivatives on boron doped dia-mond: voltammetric study and galvanostatic electrolyses.Chem. Eng. Technol. 29 (8), 944–950.

Brillas, E., Calpe, J.C., Casado, J., 2000. Mineralization of 2,4-D byadvanced electrochemical oxidation processes. Water Res. 34(8), 2253–2262.

Brillas, E., Boye, B., Sires, I., Garrido, J.A., Rodrıguez, R.M., Arias, C.,Cabot, P.L., Comninellis, Ch., 2004. Electrochemical destructionof chlorophenoxy herbicides by anodic oxidation and electro-Fenton using a boron-doped diamond electrode. Electrochim.Acta 49 (25), 4487–4496.

Brillas, E., Sires, I., Arias, C., Cabot, P.L., Centellas, F., Rodrıguez, R.M.,Garrido, J.A., 2005. Mineralization of paracetamol in aqueousmedium by anodic oxidation with a boron-doped diamondelectrode. Chemosphere 58 (4), 399–406.

Canizares, P., Lobato, J., Paz, R., Rodrigo, M.A., Saez, C., 2005.Electrochemical oxidation of phenolic wastes with boron-doped diamond anodes. Water Res. 39 (12), 2687–2703.

Daughton, C.G., Jones-Lepp, T.L., (Eds.), 2001. Pharmaceuticals andpersonal care products in the environment. Scientific andregulatory issues. ACS Symposium Series, Washington.

Diagne, M., Oturan, N., Oturan, M.A., 2007. Removal of methylparathion from water by electrochemically generated Fenton’sreagent. Chemosphere 66 (5), 841–848.

Drogui, P., Elmaleh, S., Rumeau, M., Bernard, C., Rambaud, A.,2001. Oxidising and disinfecting by hydrogen peroxide pro-duced in a two-electrode cell. Water Res. 35 (13), 3235–3241.

Flox, C., Ammar, S., Arias, C., Brillas, E., Vargas-Zavala, A.V.,Abdelhedi, R., 2006. Electro-Fenton and photoelectro-Fentondegradation of indigo carmine in acidic aqueous medium.Appl. Catal. B Environ. 67 (1–2), 93–104.

Flox, C., Cabot, P.L., Centellas, F., Garrido, J.A., Rodrıguez, R.M.,Arias, C., Brillas, E., 2007. Solar photoelectro-Fenton degrada-tion of cresols using a flow reactor with a boron-dopeddiamond anode. Appl. Catal. B Environ. 75 (1–2), 17–28.

Gozmen, B., Oturan, M.A., Oturan, N., Erbatur, O., 2003. Indirectelectrochemical treatment of bisphenol A in water viaelectrochemically generated Fenton’s reagent. Environ. Sci.Technol. 37 (16), 3716–3723.

Hanna, K., Chiron, S., Oturan, M.A., 2005. Coupling enhancedwater solubilization with cyclodextrin to indirect electroche-mical treatment for pentachlorophenol contaminated soilremediation. Water Res. 39 (12), 2763–2773.

Heberer, T., 2002a. Tracking persistent pharmaceutical residuesfrom municipal sewage to drinking water. J. Hydrol. 266 (3–4),175–189.

Heberer, T., 2002b. Occurrence, fate, and removal of pharmaceu-tical residues in the aquatic environment: a review of recentresearch data. Toxicol. Lett. 131 (1–2), 5–17.

Irmak, S., Yavuz, H.I., Erbatur, O., 2006. Degradation of 4-chloro-2-methylphenol in aqueous solution by electro-Fenton andphotoelectro-Fenton processes. Appl. Catal. B Environ. 63(3–4), 243–248.

Kraft, A., Stadelmann, M., Blaschke, M., 2003. Anodic oxidationwith doped diamond electrodes: a new advanced oxidationprocess. J. Hazard. Mater. 103 (3), 247–261.

Kummerer, K. (Ed.), 2001. Pharmaceuticals in the Environment.Sources, Fate and Risks. Springer, Berlin.

Marselli, B., Garcıa-Gomez, J., Michaud, P.A., Rodrigo, M.A.,Comninellis, Ch., 2003. Electrogeneration of hydroxyl radicalson boron-doped diamond electrodes. J. Electrochem. Soc. 150(3), D79–D83.

Montilla, F., Michaud, P.A., Morallon, E., Vazquez, J.L., Comninellis,Ch., 2002. Electrochemical oxidation of benzoic acid at boron-doped diamond electrodes. Electrochim. Acta 47 (21),3509–3513.

Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H., 2006.Pharmaceutical chemicals and endocrine disrupters in muni-cipal wastewater in Tokyo and their removal during activatedsludge treatment. Water Res. 40 (17), 3297–3303.

Ogawa, K., Tobe, N., 1966. A spectrophotometric study of thecomplex formation between iron(III) and salicylic acid. Bull.Soc. Chem. Jpn. 39 (2), 227–232.

Oturan, M.A., Pinson, J., Deprez, D., Terlain, B., 1992. Polyhydrox-ylation of salicyclic acid by electrochemically generated OHd

radicals. New J. Chem. 16 (6), 705–710.Oturan, M.A., Oturan, N., Lahitte, C., Trevin, S., 2001. Production of

hydroxyl radicals by electrochemically assisted Fenton’sreagent. Application to the mineralization of an organicmicropollutant, the pentachlorophenol. J. Electroanal. Chem.507 (1–2), 96–102.

Panizza, M., Cerisola, G., 2005. Application of diamond electrodes toelectrochemical processes. Electrochim. Acta 51 (2), 191–199.

Sires, I., Garrido, J.A., Rodrıguez, R.M., Cabot, P.L., Centellas, F.,Arias, C., Brillas, E., 2006. Electrochemical degradation ofparacetamol from water by catalytic action of Fe2+, Cu2+, andUVA light on electrogenerated hydrogen peroxide. J. Electro-chem. Soc. 153 (1), D1–D9.

Sires, I., Centellas, F., Garrido, J.A., Rodrıguez, R.M., Arias, C.,Cabot, P.L., Brillas, E., 2007. Mineralization of clofibric acid byelectrochemical advanced oxidation processes using a boron-doped diamond anode and Fe2+ and UVA light as catalysts.Appl. Catal. B Environ. 72 (3–4), 373–381.

Sun, Y., Pignatello, J.J., 1993. Photochemical reactions involved inthe total mineralization of 2,4-D by iron(3+)/hydrogen per-oxide/UV. Environ. Sci. Technol. 27 (2), 304–310.

Tarr, M. (Ed.), 2003. Chemical Degradation Methods for Wastesand Pollutants. Environmental and Industrial Applications.Marcel-Dekker, New York.

Weichgrebe, D., Danilova, E., Rosenwinkel, K.H., Vedenjapin, A.A.,Baturova, M., 2004. Electrochemical oxidation of drug residuesin water by the example of tetracycline, gentamicin andaspirin. Water Sci. Technol. 49 (4), 201–206.

Zuo, Y., Hoigne, J., 1992. Formation of hydrogen peroxide anddepletion of oxalic acid in atmospheric water by photolysis ofiron(III)–oxalate complexes. Environ. Sci. Technol. 26 (5),1014–1022.