Chemical Engineering Journal 270 (2015) 612–620

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Solar-photo-Fenton treatment of wastewater from the beverage

industry: Intensification with ferrioxalate
A. Durán ⇑, J.M. Monteagudo, J. Gil, A.J. Expósito, I. San Martín
Department of Chemical Engineering, Grupo IMAES, Escuela Técnica Superior de Ingenieros Industriales, Instituto de Investigaciones Energéticas y Aplicaciones Industriales
(INEI), Universidad de Castilla-La Mancha, Avda. Camilo José Cela 3, 13071 Ciudad Real, Spain

h i g h l i g h t s

 Beverage industry wastewater mineralized by solar photo-Fenton/ferrioxalate process.

 Under optimal conditions, 70.6% of TOC was removed in 55 min and 96.6% in 125 min.
 The process also removes completely the toxicity, COD and 99.8% of BOD-5.
 This process considerably reduces the time of a biological treatment.
 Synergism of photo-Fenton process and ferrioxalate photochemistry is 22.9%.

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

Article history: The mineralization of industrial wastewater from beverage industries during a solar photo-Fenton
Received 9 December 2014 enhanced process mediated by ferrioxalate complexes was evaluated as an alternative to reduce the total
Received in revised form 16 February 2015 treatment time required for conventional anaerobic digestion procedures in a compound parabolic
Accepted 17 February 2015
collector (CPC) pilot plant. Under selected conditions (H2O2 flowrate = 460 mL/h, H2C2O4
Available online 24 February 2015
flowrate = 2100 mL/h, [Fe]0 = 150 mg/L, pH = 2.79, medium solar power = 35.8 Wh) and continuous
operation, 70.6% and 96.6% of the total organic carbon (TOC) was removed from industrial effluent with
Keywords:
an initial TOC concentration of 1386.8 mg/L after 55 and 125 min, respectively. In addition, this process
Beverage
Wastewater
completely removed the toxicity and COD and removed 99.8% of the BOD-5.
Ferrioxalate First, the physico-chemical pre-treatment of raw wastewater was performed based on sedimentation
Photo-Fenton to remove suspended solids and reduce the turbidity by 91%.
CPC The effects of the variables were studied during two different irradiation periods. Solar power is the
main factor that influences mineralization during the first 60 Wh of accumulated energy due to the
generation of hydroxyl radicals. However, solar power is unimportant at the end of the process
(150 Wh of accumulated energy), when the molecular reaction mechanism between H2O2 and the
intermediates is predominant.
The overall mineralization process (k = 0.0096 min1) occurs due to the contributions of the
photo-Fenton process (k = 0.0044 min1) and the ferrioxalate photochemistry (k = 0.003 min1). The
synergism between both processes was 22.9% based on the first order rate constants for TOC removal.
Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction
The beverage industry generates large amounts of wastewater
effluent. In the beverage industry, water is used as an ingredient,
an initial and intermediate cleaning source, an efficient conveyor
of raw materials, and a principal agent for sanitizing plant machin-
ery and work areas. These effluents contain high concentrations of
⇑ Corresponding author. Tel.: +34 926295300x3814; fax: +34 926295361.
E-mail address: Antonio.Duran@uclm.es (A. Durán).
organics due to the use of raw materials, such as oranges, grapes or
sugar [1].
Conventional biological processes are usually the most cost-
effective alternatives for treating these effluents, but it is widely
known that high concentrations of toxic or non-biodegradable
compounds prevent mineralization [2,3]. Moreover, 12 h are
required to achieve chemical oxygen demand (COD) removal
efficiencies higher than 90% [4].
Advanced oxidation processes (AOPs) are an efficient alterna-
tive that could be used to solve this problem. Among AOPs, the

http://dx.doi.org/10.1016/j.cej.2015.02.069
1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620
ferrioxalate-induced solar photo-Fenton process is a clean
technology that is considered environmentally friendly and
competitive at the industrial scale [5]. The ferrioxalate-induced
solar photo-Fenton process is based on the generation of highly
reactive oxidative intermediates species (hydroxyl radical, OH,
singlet oxygen, 1O2, or superoxide radical anion, O 2 ) [6]. In
contrast, compound parabolic collector (CPC) pilot plants can
efficiently use direct and diffuse solar radiation and permit
continuous wastewater treatment at the industrial scale.
The use of ferrioxalates for organic pollutant degradation under
sunlight radiation has been previously reported as effective [7–9].
The process of using ferrioxalates employs a higher portion of the
solar spectrum. The photolysis of ferrioxalate generates Fe(II)
under acidic pH conditions through a well-known mechanism
[10], improving the mineralization rate. Oxalic acid was used in
this study for this purpose because it is very active in the photo-
Fenton reaction [11,12].
In this study, the mineralization of actual beverage wastewater
effluent was studied using a continuous ferrioxalate-induced solar
photo-Fenton process as an alternative to conventional anaerobic
digestion or as a rapid pre-treatment step (to reduce the total
treatment time). Physico-chemical pre-treatment of the raw
wastewater was performed using precipitation and sedimentation
to improve the photocatalytic treatment efficiency.
Six variables were studied in this work using a Central
Composite Experimental Design and the hydrogen peroxide and
oxalic acid flowrate, initial TOC concentration, pH, temperature
and solar power. The response function was the degree of mineral-
ization (after 60 and 150 Wh of accumulated solar power).
The experimental results were fit with neural networks (NNs)
[13,14] to simulate the effects of the different variables on the
degree of mineralization degree as a function of the accumulated
solar power. The remaining H2O2 concentrations in the water and
the concentrations of the different iron species and dissolved
oxygen (DO) were also measured.
The novelty of this research includes several items: it is the first
time that reagents are added continuously to the system, the effect
of the variables was studied under two different irradiation periods
and finally since the overall mineralization process occurs by the
contribution of the photo-Fenton process and the ferrioxalate
photochemistry, the synergism between both processes was
estimated from the first order rate constants for TOC removal.
2. Experimental
2.1. Materials and methods
A factory that produces juice drinks from orange, apple and
grape supplied the wastewater used in this study. The composition
of the beverage wastewater is given in Table 1.
Hydrogen peroxide (H2O2, 30% w/v, obtained from Merck),
analytical-grade ferrous sulfate (FeSO47H2O) and oxalic acid
Table 1
Main characteristics of the beverage industry wastewater.
Parameter Average value
pH 5.35
Conductivity (mS cm1) 2.43
Total suspended solids (mg L1) 1300
Turbidity (NTU) 1850
TOC (mg L1) 1987
COD (mg L1) 6500
BOD5 (mg L1) 4400
Acetate (mg L1) 660
Formate (mg L1) 110
613
(H2C2O42H2O, 99.5%) were used as received and immediately
added to the wastewater in situ to form ferrioxalate complexes
because of their light sensitivity.
2.2. Solar-CPC pilot plant
The wastewater was mineralized during the photo-Fenton
process using ferrioxalate at a pilot plant consisting of a solar
compound parabolic collector (CPC) manufactured by Ecosystem,
S.A. (Fig. 1). This pilot plant included a solar reactor with a
continuously stirred tank (50 L), a centrifugal recirculation pump
(30 L min1), and a solar collector unit with an area of 2 m2 (more
details can be found elsewhere [6]). The incident solar power
(W m2) and accumulated solar energy (Wh) were measured using
a programmable logic controller (PLC) coupled with a radiometer
(Ecosystem, model ACADUS-85).
The experimental set-up is also completed with an UV pilot
plant composed of a 28-L reactor with two UV-C lamps and two
UV-A lamps which can be operated in series with the solar CPC
plant to improve efficiency or in not sunny days, but it was not
used in this research.
2.3. Analysis
The degree of mineralization, calculated as (TOC0  TOC(t)/
TOC0), was monitored using a TOC-5050 Shimadzu analyzer (stan-
dard deviation <0.2 mg L1). The TOC contributions of the added
oxalic acid had relatively low significance because ferrioxalate is
easily photolyzed, resulting in oxalate mineralization. In addition,
the H2O2 concentration in solution was determined using titration
with an aqueous solution of potassium permanganate (0.02 M)
and an automatic Titrino SET/MET 702 (Metrohm). The dissolved
O2 concentration (DO) was measured using a Lange LXV416
Luminescent Dissolved Oxygen (LDO) sensor, and the ferrous Fe
concentrations were obtained using photometric measurements
with 1,10-phenanthroline (according to ISO 6332) and a UV–Vis
spectrophotometer (Zuzi 4418PC). Before analysis, all samples were
withdrawn from the reactor and immediately treated with excess
Na2SO3 solution to prevent further oxidation (this procedure was
performed to prevent the overestimation of degradation).
Chemical oxygen demand (COD) and biochemical oxygen
demand (BOD-5) analyses were performed according to methods
already described in literature [5]. Suspension solids were deter-
mined according to APHA Standard Methods [15].
The formate and acetate concentrations were determined by
high performance ion chromatography using a Metrohm chro-
matograph fit with an ASUPP5 250 column and an ionic conduc-
tivity detector at a pressure of 10–12 MPa. In addition, 0.7 mL/
min of a 50/50 3.2 mM NaHCO3/mM Na2CO3 solution was used
as the moving phase, and 0.33 mL/min of deionized water and
0.5 mL/min of sulfuric acid (50 mM) were used for the suppression
module.
The toxicity was evaluated by determining the inhibitory effects
of water samples on the light emission of Vibrio fischeri
(Luminescent bacteria test) using a luminometer (Optocomp
BG-1, Gomensoro) according to ISO 11348-3:1998.
2.4. Experimental design
A central-composite experimental design (CCED) consisting of
11 experiments was applied to investigate the effects of two
variables [flowrate of H2O2 (mL h1) and flowrate of oxalic acid
(mL h1)] in the ferrioxalate-induced solar photo-Fenton process
(see Table 2; complete details of the CCED were described
previously [6]).
614 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620

(a)
radiometers
Solar radiation

UV-A/C Reactor CPC Reactor

sampling

Heater
• RB4
• Fe (II)
O 2 sensor • (COOH)2
• H2 O2

Water tank

Temperature Sy ringe Air

pump bubbling

Recirculation pump pH-meter

pouring

(b)
Fig. 1. Experimental setup based on a CPC pilot plant. Reactor volume: 16 L; (a) photo; (b) schematic.

The design consisted of a series of the three following mineralization in the treated water at accumulated solar energies
experiments: (i) a 2k factorial design (all possible combinations of 60 and 150 Wh. The complete experimental design is shown in
of the coded values +1 and 1), which consisted of 4 experiments Table 2.
when k = 2 (experiments 1–4); (ii) axial or star points (coded In all the experiments, the disappearance of TOC followed pseu-
values a = 2k/4 = ± 1.6817) consisting of 2k = 4 experiments do-first-order kinetics with respect to the pollutant concentration,
(experiments 5–8); and (iii) replicates of the central point as follows:
(3 experiments, 9–11). One additional experiment (12) was carried
out under optimal conditions. dC
r¼ ¼ kC ð1Þ
The incident solar power, temperature and pH were not dt
controlled during the experiment. Instead, these parameters were
where r is the reaction rate, C is the concentration of TOC (mg L1)
measured during the reaction, and their average values were used
at normalized illumination time t and k is the pseudo-first-order
during the fitting process. The initial TOC concentration was also
mineralization-rate constant. This equation can be integrated
used for fitting because it fluctuated between 1897 and 1386 mg/L
between t = 0 and t = t, yielding:
due to biological degradation during storage.
The chosen response function for optimizing the photocatalytic C0
ln ¼ kt ð2Þ
process through experimental design was the degree of C
 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620 615

Table 2
The 2-factor central composite design matrix.
a a a
Exp. Q H2 O2 (mL h1) Q H2 C2 O4 (mL h1) [TOC]o (mg L1) T (°C) pH Solar power (W m2) %TOC removal (60 Wh) %TOC removal (150 Wh)

1 395.6 1792.5 1416 37.7 2.72 36.6 44.1 86.3

2 84.4 1792.5 1422.6 40.5 2.93 34.8 31.2 36.3
3 395.6 307.5 1595.2 35 2.74 37.1 26.6 43.4
4 84.4 307.5 1871.2 38.9 2.72 32.6 26.2 33.0
5 460 1050 1631.4 43.5 2.75 41.2 31.6 74.1
6 20 1050 1386.8 39.7 2.72 26.9 27.2 32.5
7 240 2100 1897.8 40.8 2.74 34.3 31.6 55.6
8 240 0 1691.2 38.8 2.75 33.9 23.9 51.2
9 240 1050 1668 38.9 2.75 37.9 30.6 51.8
10 240 1050 1600 38.4 2.80 37.1 30.1 50.9
11 240 1050 1633 38.1 2.78 38.1 30.5 51.9
12b 460 2100 1386 40.2 2.79 35.8 70.6 96.6
Coded levels
(+a) 460 2100
(a) 20 0
(+1) 395.6 1792.5
(1) 84.4 307.5
(0) 240 1050
a
Average values in experiments.
b
Optimal conditions.

where C0 is the initial concentration of TOC. According to this
expression, a plot of the first term versus t must yield a straight line
validating Eq. (2), where the slope is k.
In order to compare experiments over several days, it is neces-
sary to use a simple equation to normalize the data so that the time
used in Eqs. (1) and (2) comes from the following correlation [16]:
UV V i
t 30W;n ¼ t 30W;n1 þ Dtn ; Dt n ¼ tn  t n1 ; t0 ¼ 0 ðn ¼ 1Þ
30 V T
where tn is the experimental time for each sample, UV is the aver-
age solar ultraviolet radiation (k < 400 nm) measured between tn1
and tn, and t30W is a normalized illumination time that refers to a
constant solar power of 30 W m2 (typical solar UV radiation on a
perfectly sunny day around noon).
2.5. Neural network strategy
The neural network used in this work to determine the miner-
alization constants and obtain optimal operation conditions was
ð3Þ solved for two neurons and by using an exponential function and
solution strategy based on a back-propagation algorithm [13,14].
Additional details regarding this method can be found in our pre-
vious report [17]. Finally, the saliencies of the input variables were
measured based on the connection weights of the NNs (Table 3).

Table 3
Equation and neural network fitting parameters for the degradation of the beverage industry wastewater. Ferrioxalate-induced solar photo-Fenton process.

Neural network fitting

Equation* for beverage industry wastewater mineralization
%TOC removal ¼ N 1  ð1=ð1 þ 1=EXPðQ H2 O2  W 11 þ Q H2 C2 O4  W 12 þ ½TOCo  W 13 þ Temperature  W 14 þ pH  W 15 þ Solar Power  W 16 ÞÞÞ
þN2  ð1=ð1 þ 1=EXPðQ H2 O2  W 21 þ Q H2 C2 O4  W 22 þ ½TOCo  W 23 þ Temperature  W 24 þ pH  W 25 þ Solar Power  W 26 ÞÞÞ
Weight factors Parameters Values of neurons and factors
60 Wh 150 Wh
N1 Neuron 0.5160 0.7544
W11 Q H2 O2 0.0889 1.3494
W12 Q H2 C2 O4 0.0953 0.6217
W13 [TOC]o 1.0887 1.4189
W14 Temperature 1.5264 1.9767
W15 pH 1.0417 1.9150
W16 Solar power 1.2037 0.3434
N2 Neuron 0.2525 0.0415
W21 Q H2 O2 2.5245 0.9272
W22 Q H2 C2 O4 5.6696 0.6146
W23 [TOC]o 6.2337 0.9442
W24 Temperature 4.6675 0.7728
W25 pH 1.5116 1.4092
W26 Solar power 4.7397 0.1574
Saliency analysis of the input variables for the neural network (%)
Neural network output Parameters
Q H2 O2 Q H2 C2 O4 [TOC]o Temperature pH Solar power

%TOC removal (60 Wh) 6.09 12.38 25.90 28.27 2.98 24.38
%TOC removal (150 Wh) 21.42 11.81 22.21 25.32 14.60 4.64
*
Parameter values in equations were previously normalized to the (0.1) interval.
616 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620

This process allowed us to analyze the relevance of each variable Table 4

(in percentage) relative to the other variables. Selected optimal conditions and experimental results.

Variable Optimal conditions

3. Results and discussion

1
H2O2 flowrate (mL h ) 460a
[Fe(II)]o (mg L1) 150
H2C2O4 flowrate (mL h1) 2100b
3.1. Physico-chemical pre-treatment Average pH 2.79
Average temperature 40.2
The sedimentation test revealed (data not shown) that the Average solar power (W m2) 35.8
removal of turbidity in 10 min was between 89% and 91% in all Response function after 60 Wh (55 min) Optimal results
TOC removal 70.6
cases (residual turbidity = 167 NTU), regardless of the initial pH
Response functions after 150 Wh (125 min) Optimal results
value, which varied from 2.5 to 8.6. Thus, pre-treatment was TOC removal 98.6
carried out without modifying the pH of the raw wastewater. COD removal (%) 100
Pre-treatment was essential for achieving an efficient BOD-5 removal (%) 99.8
Toxicity removal (%) 100
photo-catalytic process. Previous tests indicated that the degree
of mineralization reached in wastewater without pre-treatment a
[H2O2]0 = 1000 mg L1.
b
was always less than 20% due to a screen effect linked to radiation [H2C2O4]0 = 500 mg L1.
scattering that decreased the depth of radiation penetration.

3.2. Mineralization in the solar CPC pilot plant

All experiments presented in this section were based on the

mineralization of wastewater from the beverage industry (initial
TOC concentration varied between 1386 and 1897 mg L1). The
pH value was adjusted to 2.7–2.9 using sulfuric acid. Next, the
wastewater was degraded using a ferrioxalate-induced solar
photo-Fenton process. Hydrogen peroxide and oxalic acid were
continuously dosed to the system at concentrations of 3  105
and 104 mg/L, respectively. Initially, the following reagent concen-
trations were present in the wastewater: [H2O2]0 = 1000 mg/L,
[H2C2O4]0 = 500 mg/L and [Fe(II)]0 = 150 mg/L.
The experimental tests and the values of the achieved %TOC
removal under different accumulated energy values are shown in
Table 2. In addition, the experimental results and NN predictions
of %TOC removal agreed, with an average error of less than 5%. Fig. 2. Correlation between consumed H2O2 and removed %TOC under optimal
The equation and fitting parameters for %TOC removal are shown operating conditions. (H2O2 flowrate = 460 mL h1, H2C2O4 flowrate = 2100 mL h1,
in Table 3. Here, N1 and N2 are general factors related to the first [Fe]0 = 150 mg L1, pH = 2.79, medium solar power = 35.8 Wh).

and the second neuron, respectively, and W11 through W16 are
the contribution parameters for the first neuron and represent
the influences of the following different variables: H2O2 flowrate
(Q H2 O2 ), H2C2O4 flowrate (Q H2 C2 O4 ), [TOC]0, average temperature,
pH and average incident solar power, respectively. In addition,
W21 through W26 represent the contributions for the second
neuron for the same variables.
The saliency analysis results (Table 3) indicate that temperature
and solar power are the main variables that affect the degree of
mineralization after 60 Wh of accumulated energy in the studied
range (28.27% and 24.38%, respectively), regardless of the
presumed effects of the initial TOC concentrations. However, when
the reaction is nearly completed (after 120 Wh of accumulated
energy), the effects of solar power dramatically decrease (4.64%),
and the influences of the H2O2 flowrate increase (from 6.09% to
21.42%), as explained below.
3.3. Optimal conditions
Table 4 shows the selected optimal operating conditions and
the degree of mineralization obtained when treating wastewater
from a beverage industry in the CPC pilot plant. Under such condi-
tions, a 70.6% degree of mineralization was achieved after 60 Wh
(55 min of reaction), reaching up of 98.6% after 150 Wh
(125 min). This process considerably reduces conventional biolo-
gical treatment time. Furthermore, this process could be used as a
pre-treatment step by stopping the reaction earlier (when a resi-
dual organic content is present) before performing a biological step
to reduce costs. With this type of pre-treatment, a loss of toxicity
occurs that is usually accompanied by enhanced biodegradability
of the treated water [18].
It is difficult to determine which products are present in the ini-
tial industrial wastewater and form intermediate compounds.
Thus, we analyzed the acetate and formate concentrations during
the optimal degradation test. It this case (Fig. 2), acetate is present
at the beginning of the process and is completely degraded
later, confirming that this compound is one intermediate in the
mineralization reactions.
Carboxylic acids (such as formic acid and acetic acid) are stable
and invariably develop a resistance to further oxidation [19]. In the
food industry, acetic acid is used as an acidity regulator and condi-
ment under the food additive code E260. Additionally, acetic acid is
formed during the fermentation processes of some foods. Acetic
acid is converted to acetate at pH values above 5.5. In this work,
as shown in Fig. 2, the acetate concentration decreases as the
reaction progresses. This result indicates an efficient oxidation
process because the methyl group in the a-position in the
carboxylic group is difficult to oxidize.
Finally, a linear relationship between the consumed peroxide
and the percentage of removed TOC was observed under optimal
conditions. Overall, 2.5 mol of H2O2 were consumed per each mole
of total organic carbon removed from solution.
3.4. Effects of H2O2 and H2C2O4 dosing
3.4.1. High flowrates of H2O2 and H2C2O4
When both flowrates (oxalic acid and hydrogen peroxide) are
high, as shown in Fig. 3a, the mineralization process is initially very
 Gráfico mal expostos
A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620 617

1 1200 20
H2O2 Oxygen 18
1000
16
Tempo

Dissolved oxygen (mg/L)

14
800
12 para

H2O2 (mg/L)
600 10 degradar
8 todo
400
6 efluente?
4
200
2
0 0
0 50 100 150
t30W (min)

(a)
1200 4
H2O2
H2O2 Oxygen
3,5
1000
3

Dissolved oxygen (mg/L)

800
2,5

H2O2 (mg/L)
600 2

1,5
400
1
200
0,5

0 0
0 50 100 150
t30W (min)

(b)

3500 16
H2O2
H2O2 Oxygen
3000 14

12
2500

Dissolved oxygen (mg/L)

10
H2O2 (mg/L)

2000
8
1500
6
1000
4
500 2

0 0
0 50 100 150
t30W (min)

(c)
Fig. 3. Mineralization of actual beverage industry wastewaters during the ferrioxalate-induced solar photo-Fenton process. Evolution of %TOC removal, iron species, H2O2
and dissolved O2 in solution under different operation conditions. (a) Q H2 O2 = 395 mL h1; Q H2 C2 O4 = 1792 mL h1. (b) Q H2 O2 = 84 mL h1; Q H2 C2 O4 = 1792 mL h1.
(c) Q H2 O2 = 395 mL h1; Q H2 C2 O4 = 307 mL h1.

fast, suggesting a radical mechanism. In the presence of sufficient
hydrogen peroxide and Fe concentrations, the addition of oxalic
acid to the system increases the amount of mineralization in
wastewater due to the continuous regeneration of Fe(II) from the
photo-reduction of Fe(III) by solar light [according to Reactions
(4) and (5)] and the generation of extra free radicals (mainly OH)
due to ferrioxalate photochemistry as commented in the following
section [18]:
FeðIIIÞ þ H2 O ! FeðIIIÞðOHÞ2þ þ Hþ
FeðIIIÞðOHÞ2þ þ hm ! FeðIIÞ þ  OH
Ferric ions are formed from the oxidation of ferrous ions by
dissolved oxygen (Reaction (6)), mainly during the Fenton
reaction (Reaction (7)), and generate extra hydroxyl radicals as
follows:
FeðIIÞ þ O2 ! FeðIIIÞ þ O
2 ð6Þ
FeðIIÞ þ H2 O2 ! FeðIIIÞ þ  OH þ OH ð7Þ
In this test, Fe(II) is very quickly converted into Fe(III) in the
presence of excess hydrogen peroxide (Fig. 3a). Thus, the mineral-
ization process is catalyzed by Fe(III)–H2O2. The hydrogen peroxide
ð4Þ
concentrations in the solutions decrease during this first stage
because it is being consumed in an efficient degradation process.
ð5Þ In contrast, because the reaction of hydrogen peroxide with
hydroxyl radicals produces oxygen (Reactions (8) and (9)) and
the reagent is decomposed into water and oxygen (Reaction
618 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620

70

35
60
%

TOC removal, %
Mineralization degree,

50
30

40

25
30

20 40 20
38 44
36 42

-2
200

C
m
400 34

,º
10

,W
600 50 40

re
800 32 100

er

tu
1000 150
ow
30

ra
Oxa 1200 200 38
lic ac

pe
rP
id flo 1400 28 250

m
300
la

wrat 1600

Te
e, m HO
So

L /h 1800 2 2 flo
350 36
wr ate, 400
mL h -1
(a)
Fig. 5. Effects of temperature and hydrogen peroxide flowrate on the degree of
mineralization after 150 Wh. (Simulation conditions: H2C2O4 flowrate = 1050
mL h1; [Fe(II)]o = 150 mg L1; [TOC]o = 1600 mg L1; pH = 2.75; average solar
50 power = 37.9 W m2.)

During the first stage of this test, the DO content remains very
45 low because it is efficiently used in mineralization reactions.
,%

However, after 110 Wh of accumulated solar power, this process

Mineralization degree

becomes slower, which indicates that the intermediates products

40 are not easily degraded. In this case, the significance of hydrogen
peroxide in the saliency analysis was higher (21.4% in Table 3),
indicating that a slow direct reaction of molecular H2O2 with some
intermediates is feasibly. This mineralization stage consumes
35
lower amounts of H2O2 so that increases in the hydrogen peroxide
40
in solution can be observed. In this stage, the DO content increases
38 dramatically after 110 Wh of accumulated solar power, confirming
-2

30 36 the inefficiency of the later process stage and indicating that a slow
m

200 34
,W

400 molecular reaction may occur.

600 32
er

800
ow

Oxa 1000 30 The mineralization constants for each step (fast and slow) were
rP

lic A 1200
cid f 1400 28 0.0096 and 0.0016 min1, respectively.
la

lowr
So

ate, 1600
mL/h
(b)
Fig. 4. Effects of solar power and the oxalic acid flowrate on the degree of
mineralization. (Conditions: H2O2 flowrate = 240 mL h1; [Fe(II)]o = 150 mg L1;
[TOC]o = 1600 mg L1; pH = 2.75; average temperature = 39 °C.) (a) After 60 Wh.
(b) After 150 Wh.
(10)), monitoring the dissolved oxygen (DO) content has been
suggested as an easy procedure for measuring the progress of the
photo-Fenton process and the efficiency of H2O2 consumption
[20–22]. The generation of excess oxygen indicates inefficient
reaction conditions, and a decrease in the dissolved oxygen con-
centration (to near zero) indicates the participation of competitive
superoxide radicals in the degradation of organic compounds,
resulting in low available hydrogen peroxide concentrations [23].
Therefore, based on DO measurements, different strategies for
optimizing hydrogen peroxide consumption are proposed:
1800
3.4.2. Low H2O2 and high H2C2O4 flowrates
If the hydrogen peroxide flowrate decreases, important changes
in the mineralization process occur. During the first 20 Wh of
accumulated solar power, the process is similar to the first stage
described in the previous section. Despite the low amounts of
added H2O2, a high H2O2 concentration is present in the solution
due to the initial dose of H2O2 present in the wastewater
(1000 mg L1). Consequently, degradation is faster during this
stage. However, this peroxide is quickly consumed and the process
becomes slower (k = 0.003 min1). Indeed, an H2O2 concentration
of nearly zero is not sufficient for oxidizing Fe(II) to Fe(III); thus,
Fe(II) remains the main species in solution.
Mineralization during this stage (after 20 Wh) results from the
continuous regeneration of Fe(II) from the photo-reduction of
Fe(III) by solar UV-A light, according to Reaction (5), and the
extra generation of free radicals (mainly OH) due to ferrioxalate
photochemistry [18]:

H2 O2 þ  OH ! HO2 þ H2 O ð8Þ ½FeðC2 O4 Þ3 3 þ hm ! FeðIIÞ þ 2C2 O2 

4 þ C2 O4 ð11Þ

3
HO2 þ HO2 ! H2 O2 þ O2 ð9Þ C2 O
4 þ ½FeðC2 O4 Þ3  ! FeðIIÞ þ 3C2 O2
4 þ 2CO2 ð12Þ

2H2 O2 ! O2 þ 2H2 O ð10Þ C2 O 

4 þ O2 ! O2 þ 2CO2 ð13Þ
 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620
þ
O 
2 þ H ! HO2
HO2 þ HO2 ! H2 O2 þ O2
TOC reduction results from the decarboxylation of organic-acid
intermediates as follows [24]:
þ2
½FeðIIIÞ ðRCO2 Þ  ! FeðIIÞ þ CO2 þ R
3.4.3. High H2O2 and low H2C2O4 flowrates
In this case, the mineralization reaction is mainly due to OH
radicals coming from the photo-Fenton process. Because hydrogen
peroxide is added in excess, its concentration in solution increases
from the beginning of the process. Thus, excess hydrogen peroxide
reduces catalytic activity because of the scavenger effect, in which
OH radicals react with each other to produce hydrogen peroxide
according to Reaction (17), reducing the number of radicals avail-
able for destroying pollutants and decreasing the overall constant
(k = 0.0044 min1). In addition, the decomposition of hydrogen
peroxide to water and oxygen (Reactions (8) + (9) + (17)) is
favored:

OH þ  OH ! H2 O2
In summary, the overall mineralization process (k = 0.0096
min1) results from the photo-Fenton process (k = 0.0044 min1)
and the ferrioxalate photochemistry (k = 0.003 min1).
The synergism between the photo-Fenton process and the
ferrioxalate photochemistry was 22.9% when using the first order
rate constants for TOC removal according to Eq. (18) [25]:


koverall  kphotonFenton þ kferrioxalate
Synergy ð%Þ ¼  100
koverall
0:0096  ð0:0044 þ 0:003Þ
¼  100 ¼ 22:9
0:0096
3.5. Effect of solar power
Fig. 4a shows the effects of solar power and the oxalic acid
flowrate on the %TOC removal from beverage wastewater during
the ferrioxalate-induced solar photo-Fenton process after 60 Wh
of accumulated solar energy. The following operating conditions
were used: H2O2 flowrate = 240 mL h1; [Fe(II)]o = 150 mg L1;
[TOC]o = 1600 mg L1; pH = 2.75; average temperature = 39 °C. As
shown in Fig. 4, the influence of solar power on the mineralization
of wastewater is always positive, regardless of the oxalic acid
flowrate.
However, when 150 Wh of solar power has accumulated
(Fig. 4b) under the same operation conditions, the effects of solar
power are negligible (4.6% in the saliency analysis). Thus, under
these circumstances (when most of TOC has already been degraded
and an excess of solar power is received by the wastewater), it can
be concluded that an unnecessary amount of radicals is generated
and that some hydrogen peroxide is consumed by unproductive
reactions (as mentioned above).
3.6. Effects of temperature
One example of the effects of temperature was simulated and is
shown in Fig. 5 (simulation conditions: H2C2O4 flowrate = 1050
mL h1; [Fe(II)]o = 150 mg L1; [TOC]o = 1600 mg L1; pH = 2.75;
average solar power = 37.9 W m2). This parameter always result-
ed in increasing reaction rates based on Arrhenius law and gener-
ated more OH radicals during the ferrioxalate photochemistry
reactions. Although temperature also influences the inefficient
decomposition of hydrogen peroxide into H2O and O2 (inactive
species) and the formation of radicals with smaller oxidation
619
ð14Þ potentials, the enhancement of OH generation prevails over the
inefficient species formation.
ð15Þ
4. Conclusions
Wastewater from a beverage industry can be photo-degraded
ð16Þ using a ferrioxalate-mediated solar photo-Fenton process. Under
optimal conditions, 70.6% and 96.6% of total organic carbon
(TOC) was removed from industrial effluent with 1386.8 mg/L of
initial TOC after 55 and 125 min, respectively. In addition, the
treatment process completely removed the toxicity and COD and
removed 99.8% of the BOD-5. This process required considerably
less time than the conventional biological treatment process.
Furthermore, this process could be used as a pre-treatment step
by stopping the reaction earlier (when a residual organic content
is present) to perform a biological step later, which would reduce
treatment costs.
Finally, the synergism between the photo-Fenton process and
the ferrioxalate photochemistry was 22.9% according to the first
order rate constants for TOC removal.
ð17Þ Acknowledgements
We gratefully acknowledge financial support from JCCM
(POII10-0114-3563), MINECO (CTM2013-44317-R) and FEDER
(UNCM08-1E-008).
References
[1] N.Z. Al-Mutairi, M.F. Hamoda, I. Al-Ghusain, Coagulant selection and sludge
conditioning in a slaughterhouse wastewater treatment plant, Bioresour.
Technol. 96 (2) (2004) 115–119.
[2] M. Rodríguez, S. Malato, C. Pulgarín, S. Contreras, D. Curcó, J. Giménez, S.
ð18Þ Esplugas, Optimizing the solar photo-Fenton process in the treatment of
contaminated water. Determination of intrinsic kinetic constants for scale-up,
Sol. Energy 79 (2005) 360–368.
[3] R.F.P. Nogueira, M.R.A. Silva, A.G. Trovó, Influence of the iron source on the
solar photo-Fenton degradation of different classes of organic compounds, Sol.
Energy 79 (2005) 384–392.
[4] E.E. Ozbas, N. Tufeckci, G. Yilmaz, S. Ovez, Aerobic and anaerobic treatment of
fruit juice industry effluents, J. Sci. Ind. Res. 65 (2006) 830–837.
[5] A. Durán, J.M. Monteagudo, A. Carnicer, I. San Martín, P. Serna, Solar
photodegradation of synthetic apple juice wastewater: process optimization
and operational cost study, Sol. Energy Mater. Sol. Cells 107 (2012) 307–315.
[6] J.M. Monteagudo, A. Durán, M. Aguirre, I. San Martín, Optimization of the
mineralization of a mixture of phenolic pollutants under a ferrioxalate-
induced solar photo-Fenton process, J. Hazard. Mater. 185 (2010) 131–139.
[7] M.S. Lucas, J.A. Peres, Degradation of Reactive Black 5 by Fenton/UV-C and
ferrioxalate/H2O2/solar light processes, Dyes Pigments 74 (2007) 622–629.
[8] D. Prato-García, R. Vasquez-Medrano, M. Hernández-Esparza, Solar
photoassisted advanced oxidation of synthetic phenolic wastewaters using
ferrioxalate complexes, Sol. Energy 83 (2009) 306–315.
[9] N. Klamerth, S. Malato, M.I. Maldonado, A. Aguera, A. Fernández-Alba, Modified
photo-Fenton for degradation of emerging contaminants in municipal
wastewater effluents, Catal. Today 161 (2011) 241–246.
[10] K. Selvam, M. Muruganandham, M. Swaminathan, Enhanced heterogeneous
ferrioxalate photo-fenton degradation of reactive orange 4 by solar light, Sol.
Energy Mater. Sol. Cells 89 (2005) 61–74.
[11] F.B. Li, X.Z. Li, C.S. Liu, X.M. Li, T.X. Liu, Effect of oxalate on photodegradation of
bisphenol a at the interface of different iron oxides, Ind. Eng. Chem. Res. 46
(2007) 781–787.
[12] E.M. Rodriguez, B. Nuñez, G. Fernández, F.J. Beltrán, Effects of some carboxylic
acids on the Fe(III)/UVA photocatalytic oxidation of muconic acid in water,
Appl. Catal., B 89 (2009) 214–222.
[13] D.P. Morgan, C.L. Scofield, Neural Networks and Speech Processing, Kluwer
Academic Publishers, London, 1991.
[14] R. Nath, B. Rajagopalan, R. Ryker, Determining the saliency of input variables in
neural network classifiers, Comput. Oper. Res. 24 (1997) 767–773.
[15] APHA, AWWA, WEF, Standard Methods: For the Examination of Water and
Wastewater, APHA Publication, 1995.
[16] N. Klamerth, L. Rizzo, S. Malato, M.I. Maldonado, A. Agüera, A.R. Fernández-
Alba, Degradation of fifteen emerging contaminants at lg L1 initial
concentrations by mild solar photo-Fenton in MWTP effluents, Water Res.
44 (2010) 545–554.
[17] J.M. Monteagudo, A. Durán, I. San Martin, M. Aguirre, Effect of continuous
addition of H2O2 and air injection on ferrioxalate-assisted solar photo-Fenton
degradation of Orange II, Appl. Catal., B 89 (2009) 510–518.
620 A. Durán et al. / Chemical Engineering Journal 270 (2015) 612–620
[18] V. Sarria, M. Deront, P. Péringer, C. Pulgarín, Degradation of a biorecalcitrant
dye precursor present in industrial wastewaters by a new integrated
iron(III) photoassisted–biological treatment, Appl. Catal., B 40 (2003) 231–
246.
[19] S. Fındık, G. Gündüz, Sonolytic degradation of acetic acid in aqueous solutions,
Ultrason. Sonochem. 14 (2007) 157–162.
[20] W. Gernjak, M. Fuerhacker, P. Fernández-Ibañez, J. Blanco, S. Malato, Solar
photo-Fenton treatment – process parameters and process control, Appl.
Catal., B 64 (2006) 121–130.
[21] L. Prieto-Rodríguez, I. Oller, A. Zapata, A. Agüera, S. Malato, Hydrogen peroxide
automatic dosing based on dissolved oxygen concentration during solar
photo-Fenton, Catal. Today 161 (2011) 247–254.
[22] L. Santos-Juanes, J.L.G. Sánchez, J.L.C. López, I. Oller, S. Malato, J.A. Sánchez
Pérez, Dissolved oxygen concentration: a key parameter in monitoring the
photo-Fenton process, Appl. Catal., B 104 (2011) 316–323.
[23] A. Kunai, S. Hata, S. Ito, K. Sasaki, The role of oxygen in the hydroxylation
reaction of benzene with Fenton’s reagent. Oxygen 18 tracer study, J. Am.
Chem. Soc. 108 (1986) 6012–6016.
[24] G. Sagawe, A. Lehnard, M. Lübber, D. Bahnemann, The insulated solar fenton hybrid
process: fundamental investigations, Helv. Chim. Acta 84 (2001) 3742–3759.
[25] C.G. Joseph, G. Li Puma, A. Bono, Y.H. Taufiq-Yap, D. Krishnaiah, Operating
parameters and synergistic effects of combining ultrasound and ultraviolet
irradiation in the degradation of 2,4,6-trichlorophenol, Desalination 276
(2011) 303–309.